Master Art of Translation | The Discourse Marker Function Fidelity System — How Well, Anyway, Right and You Know Carry Meaning

How to translate discourse markers without losing what a speaker is doing

When you translate discourse markers such as well, anyway, right, so, actually, you know or I mean, a word-for-word translation can preserve dictionary meanings while changing the conversation. The most reliable approach to discourse marker translation is to identify the expression’s function in its turn—acknowledgement, correction, hesitation, topic change, inference, concession or interpersonal cushioning—before deciding whether the target language needs a particle, an adverb, a phrase, a change of word order, or no overt marker at all.

Readers looking for how to translate ‘well’ or ‘you know’ naturally, how to keep conversational tone in subtitles, or how to review pragmatic markers in machine translation need more than a bilingual list. The hidden problem is that the same short expression can organise the exchange, signal a response to another person’s stance, or leave a proposition unchanged while altering how it is received. A fluent target sentence can therefore be pragmatically wrong even when every concrete fact is accurate.

This world-facing guide gives a repeatable system for translating discourse particles, conversational connectors and pragmatic markers across languages, registers and media. It separates what a speaker says from what the speaker is doing, supplies original worked diagnostics, identifies common omission and addition errors, and shows how to evaluate translation and AI post-editing without inventing agreement, conflict or certainty. For the wider architecture, begin with Master Art of Translation | The Complete System for Moving Meaning Between Languages.

Core answer: translate the function, preserve the relationship between turns, check the implied stance, and verify how a target-language reader would act or respond. A marker does not have to survive as the same number of words, but its consequential communicative effect must be accounted for.

Recognise the marker: Is it propositional content, a discourse cue, a hesitation, or a literal verb/adverb?
Read the whole exchange: The response before and after the marker often determines its job.
Name the function: Identify correction, confirmation, concession, restart, elaboration, topic change or affiliation.
Translate natural interaction: Choose the target-language construction that performs the same job, not a fixed lexical match.
Check the risk: Ask whether the reader now assumes agreement, refusal, certainty, blame or urgency that was absent.
Revise pragmatically: Compare the translated conversation against the source and its intended audience.

The hidden problem: correct facts, wrong interaction
What discourse markers actually do
The repeatable sixteen-gate method
Ninety original worked diagnostics
Failure modes and repair tests
Transfer into subtitles, literature, meetings, interfaces and AI
Four-week learning and review plan
Glossary, FAQs, sources and next routes

Scope: an advanced chapter, not a replacement for the translation root

This chapter examines the micro-signals that manage a discourse, especially the particles, formulae and short connectors that are easy to omit because they do not always change the sentence’s basic truth conditions. It deliberately does not replace the wider translation framework, the prior analysis of presupposition, the independent study of full conversation repair, or specialised work on vocabulary development. The topic is precise: functional fidelity of discourse markers, including their optional expression and recoverable social effect.

A word can be a marker in one sentence and ordinary content in another. Compare Now, about the report with We need it now. The first opens a topic; the second specifies time. Compare Look, I disagree with Look at the chart. The first takes the conversational floor; the second commands visual inspection. Those distinctions are not stylistic niceties. In instructions, meetings, service messages and witness records they change observable behaviour.

The hidden problem: correct facts, wrong interaction

The propositional trap

A translator may preserve who acted, what happened and when, then remove a marker as nonessential. Yet a response like ‘Well, perhaps’ may register limited willingness, not a reliable yes. A deletion can therefore change the degree of commitment. A similar failure occurs when a response token is translated as agreement although it merely tells a speaker to continue.

The practical safeguard is to separate content, relation and effect. Content is the claim that may be checked against the world. Relation locates the claim within prior speech and next action. Effect is the signal a reasonable interlocutor would perceive: invitation, challenge, qualification, reassurance, transition or continued attention. If any of those three changes materially, the translation requires a second pass.

The unit-of-translation trap

Single-word dictionary entries make a fluent false friend attractive. A bilingual glossary can suggest equivalents for ‘right’, ‘well’ or ‘actually’, but those words perform several incompatible functions. The relevant unit is usually a turn, an adjacency pair, a paragraph transition or an entire interactional sequence. Without the neighbouring turns, a translator cannot know which use is active.

The cultural-form trap

Languages do not divide interpersonal work into the same particles. One language may encode recognition with a clause-final particle, another with a short independent response, another with intonation or politeness morphology. Exact placement and form depend on target grammar and social relationship. This article provides language-neutral analytical tests rather than unverified universal one-to-one equivalents.

The register trap

A spoken ‘anyway’ in a friendly exchange cannot always be replaced with a formal ‘nevertheless’. Likewise ‘moreover’ in a structured analysis rarely means the breezy ‘oh, and’. The relation may be similar, but the target sentence must preserve the intended register and the speaker’s relationship to listeners.

The transcript trap

Interviews, dialogue, subtitles and written minutes have different documentary goals. A verbatim transcript records interactional behaviour, including meaningful hesitations. A clean set of minutes may omit many surface cues while retaining decisions and cautions. An accessible subtitle may need an economical substitute. A responsible translator records the assignment’s specification before choosing a level of representation.

The machine-fluency trap

Automated output can be grammatical and still erase a soft challenge or manufacture a confident answer. A post-editor who checks only lexical fidelity may miss those changes. Compare dialogue function turn by turn; test whether the same information could be followed by the same social response. Fluency is necessary but not sufficient for communicative accuracy.

Mechanism: a three-layer map of discourse meaning

Truth-conditional content

An ordinary proposition may state that a device is charging. Removing the word ‘still’ when it means continuing-in-time changes that proposition; removing a merely transitional ‘now’ may not. Test the temporal and logical role before treating a word as optional.

A reviewer can document this layer with a brief note: the relevant source span; the function in context; the target construction; and the test that would detect a change. This is not a demand that every informal utterance carry a long annotation. It is a practical record for high-risk or disputed decisions, especially when multiple possible target versions are equally fluent but imply different social outcomes.

Interactional action

Every turn does something: requests, refuses, acknowledges, qualifies, offers or closes. Discourse markers can reveal that action when the proposition alone is ambiguous. A reply ‘Okay, but…’ accepts only part of a request. The untranslated boundary can create false approval.

Discourse structure

Some markers organise information: returning to an earlier point, moving to a new agenda item, adding a reason, or presenting an example. Translation must preserve whether the information is evidence, illustration, exception, concession, summary or conclusion.

Epistemic stance

A response cue can disclose uncertainty, expectation or recognition. Do not read ‘hmm’ as disbelief without the surrounding turn; do not read ‘actually’ as aggression without an observable challenge. The target should carry the supported degree of certainty.

Social positioning

Markers may maintain solidarity, create respectful distance, soften a correction or claim authority. The same device can sound friendly or hostile depending on hierarchy, register and sequence. The translator should neither flatter nor harden the speaker beyond the source.

Processing guidance

Markers help the reader or hearer process the message. They indicate which earlier statement matters, where to pause, when a topic closes, or what follows as a reason. A translation can drop a token when structure supplies the same guidance; it cannot drop a necessary relationship.

Multimodal support

Prosody, gaze, timing and gesture can share the marker’s function. A transcript may need an annotation that subtitles do not, while subtitles may use punctuation or timing. Never assert that an image or intonation supplies a function unless the source actually gives that evidence.

Functional equivalence with restraint

A target-language particle may be idiomatic without being a literal match. Rephrasing is justified when it keeps the speaker’s action and the recipient’s plausible inference. Do not invent hidden subtext just to make the target sentence colourful. When evidence is insufficient, preserve ambiguity or query the source.

The repeatable sixteen-gate discourse marker method

Gate 1: Brief and audience

Record the requested target variety, intended audience, medium, formality, and whether the output is verbatim, readable, literary, audiovisual or operational. A reader-facing adaptation and an evidentiary transcript should not silently follow the same omission rules.

Apply this gate to a specific exchange before deciding on the final wording. For a high-consequence message, write down what would count as failure: a new promise, a suppressed correction, a closed category where the source gave an example, a false allegation, a lost warning, or a changed speaker’s role. If the only available evidence is a short isolated sentence, record the unresolved reading and prefer the least committal faithful option until context can be obtained. This safeguard is particularly important for the expressions analysed below because their form is short while their contextual reach can be long.

Gate 2: Source provenance

Check whether the source is a transcript, a rewritten subtitle, a prepared script or a summary. Markers may have been added or removed upstream. Do not restore a conversational feature from imagination as though it were documentary fact.

Gate 3: Turn boundary

Mark who speaks and where each turn starts and finishes. A ‘right’ spoken alone by the listener is not the same as a tag at the end of the other speaker’s question.

Gate 4: Prior expectation

Write the claim or request to which the marker responds. Without that expectation, ‘actually’ cannot be classified reliably as correction, surprise or soft intervention.

Gate 5: Next action

Inspect the next clause and turn. If ‘anyway’ is followed by resumed scheduling it is likely a topic return; if followed by leaving it may be an exit.

Gate 6: Lexical-versus-pragmatic

Ask whether the word is literal content or an interactional cue. ‘Look at the diagram’ requires visual action. ‘Look, I disagree’ need not.

Gate 7: Function label

Choose a narrow label: confirmer, continuer, repair preface, agenda opener, concession, exemplifier, reminder, hedge, departure, appeal or qualified acceptance. Several labels may coexist; record the primary one.

Gate 8: Commitment check

Assess whether the utterance promises, permits, refuses or merely acknowledges. Never let translation change a minimal continuer into approval or a tentative offer into a guaranteed delivery.

Gate 9: Stance check

Assess certainty, warmth, irritation, irony, humility and distance using evidence, not stereotypes. If the source gives no tone, avoid maximally emphatic target forms.

Gate 10: Relationship check

Determine whether the marker connects reasons, examples, alternatives, objections or paraphrases. A cause relation is not interchangeable with a contrast or an optional illustration.

Gate 11: Candidate options

Generate at least two natural target constructions for a difficult marker. Options can include an overt particle, a clause-level paraphrase, word order, punctuation or justified zero realisation.

Gate 12: Function back-translation

Paraphrase the target version back into the source language as a conversational action. Does it still sound like a gentle correction, a reluctant yes, or merely listening?

Gate 13: Participant response test

Imagine a reasonable recipient answering or acting. Would they schedule, approve, apologise, challenge, continue telling the story or wait? A changed likely response is a warning sign.

Gate 14: Register and medium

Validate length, subtitle timing, readability, audience expectations and speaker voice. Do not use formal legalistic connectors in everyday dialogue unless the character or document calls for them.

Gate 15: Traceable decision

Record uncertain cases, disputed readings, query choices and approved preferences. A reusable function map is more reliable than an unqualified find-and-replace table.

Gate 16: Independent review

Have a second competent reader inspect ambiguous or consequential exchanges. Review the source and the target in context, not merely a list of tokens marked as translated.

Worked diagnostic library: 90 original contexts, not a word-substitution table

The examples below are original English-language diagnostic scenarios. The proposed transfer is a function-level instruction, not a claimed literal translation into every language. Every target language must implement that instruction with its own grammatical, pragmatic and cultural resources. An utterance’s punctuation is supplied to assist reading; real audio, timing and gestures may justify a different interpretation. Each case states what must survive, a tempting failure, and a concrete acceptance test.

Mechanism 1: well

The central job of well in this group is to delays a response, reframes an expectation, or prepares a qualified answer. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the next contribution is reluctant, corrective, or simply a normal opening. A common risk is translating every occurrence as an adjective meaning healthy or good. A workable first approach is to retain the response relationship rather than a literal lexical counterpart. Those instructions are function-level; they do not assert an identical particle exists in every language.

To diagnose an occurrence, underline the prior utterance or surrounding claim, mark the first clause after the expression, and ask whether removing it changes the next reader’s likely response. Then compare two candidates: one with a natural explicit target-language cue and one with the relationship carried through syntax or punctuation. Prefer the candidate that preserves the speaker’s observable intention, the degree of certainty, and the next conversational move. No expression receives an automatic keep-or-delete rule; its status follows from the evidence of the local exchange.

Diagnostic 01: Well, I can try.

Source utterance: “Well, I can try.” Situation: A colleague is asked to accept an unfamiliar task; the speaker is willing but cannot promise delivery. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning the reply into an enthusiastic unqualified yes. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The commitment is tentative, not a confirmed outcome. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a hesitation or qualified acceptance natural in the receiving language. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would a reader treat this as a firm promise? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning the reply into an enthusiastic unqualified yes. The acceptable middle ground is the one that keeps the observed function—The commitment is tentative, not a confirmed outcome.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 02: Well, that is not what the report says.

Source utterance: “Well, that is not what the report says.” Situation: A reviewer answers a confident but mistaken summary during a meeting. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Removing the marker and making the correction sound abrupt or openly hostile. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker is preparing a measured challenge. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Choose a gentle prefatory correction or an equivalent shift in rhythm. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the response disagree without adding contempt? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Removing the marker and making the correction sound abrupt or openly hostile. The acceptable middle ground is the one that keeps the observed function—The speaker is preparing a measured challenge.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 03: Well then, where do we begin?

Source utterance: “Well then, where do we begin?” Situation: A facilitator moves a group from discussion into work after the decision has been made. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Reading well as doubt and then as mere time filler. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The phrase frames the start of a new stage. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Translate the launch of action with an appropriate opening transition. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can readers hear the meeting move forward? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Reading well as doubt and then as mere time filler. The acceptable middle ground is the one that keeps the observed function—The phrase frames the start of a new stage.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 2: anyway

The central job of anyway in this group is to returns to an interrupted thread, changes topic, or sets aside an objection. Its meaning cannot be chosen solely from a bilingual headword. First examine whether a prior topic is abandoned, merely paused, or explicitly dismissed. A common risk is rendering a dismissal when the speaker intends only a neutral return. A workable first approach is to mark topic trajectory and decide whether an overt marker is needed. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 04: Anyway, back to the schedule.

Source utterance: “Anyway, back to the schedule.” Situation: Two people have digressed into stories before returning to a planning agenda. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Treating the word as indifference to the schedule. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker restores the earlier agenda without rejecting the digression. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a resumption cue or restructure the first sentence after the digression. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the transition point backward to the suspended topic? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Treating the word as indifference to the schedule. The acceptable middle ground is the one that keeps the observed function—The speaker restores the earlier agenda without rejecting the digression.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 05: It might rain. Anyway, we have the covered hall.

Source utterance: “It might rain. Anyway, we have the covered hall.” Situation: An organiser responds to a weather concern with a practical fallback. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Deleting the marker and making the covered hall seem like unrelated information. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker downgrades the obstacle and pivots to a solution. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use an obstacle-discounting transition suited to reassurance. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the audience see why the plan can continue? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Deleting the marker and making the covered hall seem like unrelated information. The acceptable middle ground is the one that keeps the observed function—The speaker downgrades the obstacle and pivots to a solution.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 06: Anyway, I should leave.

Source utterance: “Anyway, I should leave.” Situation: A guest closes a long visit while trying not to sound dismissive. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the departure sound like an accusation that the conversation has wasted time. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The word initiates a socially delicate exit. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Select a natural closing cue rather than an aggressively contrastive connector. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would the host hear a courteous leave-taking? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the departure sound like an accusation that the conversation has wasted time. The acceptable middle ground is the one that keeps the observed function—The word initiates a socially delicate exit.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 3: actually

The central job of actually in this group is to corrects an assumption, highlights a surprising fact, or quietly revises an earlier claim. Its meaning cannot be chosen solely from a bilingual headword. First examine the strength and target of the correction. A common risk is inflating a small update into a confrontation or suppressing a necessary correction. A workable first approach is to map the challenged assumption before selecting a target expression. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 07: Actually, the delivery is on Thursday.

Source utterance: “Actually, the delivery is on Thursday.” Situation: A client has copied an old delivery day into a message. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Losing the implicit correction and presenting the date as fresh unrelated information. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker gently replaces an incorrect previous date. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Retain a corrective signal or place the corrected date in the contrastive position. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the recipient identify which plan changed? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Losing the implicit correction and presenting the date as fresh unrelated information. The acceptable middle ground is the one that keeps the observed function—The speaker gently replaces an incorrect previous date.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 08: I actually enjoyed the rehearsal.

Source utterance: “I actually enjoyed the rehearsal.” Situation: A performer speaks after expressing doubts about a new format. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Reading actually as evidence that all previous judgments were false. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The enjoyment is unexpected relative to a mild negative expectation. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a surprise or counterexpectation cue at the appropriate intensity. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the line convey unexpected pleasure rather than scorn? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Reading actually as evidence that all previous judgments were false. The acceptable middle ground is the one that keeps the observed function—The enjoyment is unexpected relative to a mild negative expectation.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 09: Actually, could we pause for a minute?

Source utterance: “Actually, could we pause for a minute?” Situation: A participant interrupts an accelerating discussion to ask for a short break. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Producing an authoritative order that overrides the speaker’s tact. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker cushions a request that redirects the group. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Pair the request with an interactionally natural softening device. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the interruption still feel polite? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Producing an authoritative order that overrides the speaker’s tact. The acceptable middle ground is the one that keeps the observed function—The marker cushions a request that redirects the group.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 4: you know

The central job of you know in this group is to invokes shared knowledge, seeks recognition, or softens an explanation. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the recipient truly shares the fact or is merely being invited to recognise it. A common risk is claiming that the listener already knows a fact they may never have heard. A workable first approach is to preserve the appeal to shared ground without imposing false knowledge. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 10: It was that small room, you know, near the stairs.

Source utterance: “It was that small room, you know, near the stairs.” Situation: A speaker helps a friend retrieve a half-remembered place within a familiar building. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Interpreting the phrase as a demand that the friend should already know. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker is facilitating memory and identification. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a recognition prompt or an economical descriptive pause. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would the addressee feel invited to remember rather than blamed? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Interpreting the phrase as a demand that the friend should already know. The acceptable middle ground is the one that keeps the observed function—The speaker is facilitating memory and identification.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 11: You know, we could ask for another opinion.

Source utterance: “You know, we could ask for another opinion.” Situation: A hesitant group member proposes an alternative to a dominant plan. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the suggestion sound like a lecture. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker opens a cooperative proposal rather than declaring settled knowledge. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Translate with a gentle proposal opener and appropriate interpersonal distance. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the new idea enter without implying superiority? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the suggestion sound like a lecture. The acceptable middle ground is the one that keeps the observed function—The marker opens a cooperative proposal rather than declaring settled knowledge.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 12: That was difficult, you know.

Source utterance: “That was difficult, you know.” Situation: A person describes an upsetting task to someone who did not witness it. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Reducing the appeal to a bare factual assertion. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker seeks emotional acknowledgement, not information transfer alone. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a recognition-seeking cue or sentence-final interpersonal shading. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the translation leave room for empathy? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Reducing the appeal to a bare factual assertion. The acceptable middle ground is the one that keeps the observed function—The speaker seeks emotional acknowledgement, not information transfer alone.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 5: I mean

The central job of I mean in this group is to repairs wording, narrows a category, or clarifies intention. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the speaker retracts, restates, or strengthens the previous phrase. A common risk is treating every I mean as an instruction about the speaker’s literal intention. A workable first approach is to identify the repair relation and its correction span. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 13: We need it soon—I mean, before noon.

Source utterance: “We need it soon—I mean, before noon.” Situation: A teammate changes a vague deadline to a precise one. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Omitting the repair so the deadline looks known from the outset. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker visibly tightens the requirement. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Keep the repair or represent the revision without inventing prior certainty. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can a reader detect that precision was added? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Omitting the repair so the deadline looks known from the outset. The acceptable middle ground is the one that keeps the observed function—The speaker visibly tightens the requirement.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 14: It’s not broken. I mean, it still runs, but slowly.

Source utterance: “It’s not broken. I mean, it still runs, but slowly.” Situation: A technician corrects an overly broad assurance. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Merging the clauses into an absolute guarantee of function. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker introduces a narrower and more honest operational claim. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve partial self-correction and the residual limitation. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the reader understand why an inspection may still be needed? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Merging the clauses into an absolute guarantee of function. The acceptable middle ground is the one that keeps the observed function—The marker introduces a narrower and more honest operational claim.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 15: I mean, thank you for staying late.

Source utterance: “I mean, thank you for staying late.” Situation: A team lead restarts after an awkward pause to express gratitude. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the line sound like a formal definition of gratitude. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The phrase is a conversational restart with social repair. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Consider natural hesitation or omit only if the warmth survives. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would the thank-you feel sincere and unrehearsed? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the line sound like a formal definition of gratitude. The acceptable middle ground is the one that keeps the observed function—The phrase is a conversational restart with social repair.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 6: so

The central job of so in this group is to announces a consequence, opens a new turn, or sums up discussion. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the relationship is logical inference or conversational floor management. A common risk is adding a causal claim whenever so appears at a turn opening. A workable first approach is to distinguish reason-result logic from discourse-stage movement. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 16: So, we meet at eight?

Source utterance: “So, we meet at eight?” Situation: A participant checks a conclusion after several options have been discussed. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning the question into a command to attend. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The word signals a tentative synthesis requiring confirmation. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a recap and confirmation frame, not a cause-only conjunction. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the audience understand that agreement is being checked? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning the question into a command to attend. The acceptable middle ground is the one that keeps the observed function—The word signals a tentative synthesis requiring confirmation.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 17: The file is missing, so we cannot verify the figure.

Source utterance: “The file is missing, so we cannot verify the figure.” Situation: An analyst explains a concrete consequence of absent evidence. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Deleting the consequence relation and leaving two disconnected facts. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The second clause follows from the first under the task’s verification rule. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Retain an explicit causal connector or equivalent construction. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the reader identify the reason verification is blocked? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Deleting the consequence relation and leaving two disconnected facts. The acceptable middle ground is the one that keeps the observed function—The second clause follows from the first under the task’s verification rule.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 18: So, tell me about your idea.

Source utterance: “So, tell me about your idea.” Situation: An interviewer begins the main question after greetings. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Inventing a previous argument from which the question supposedly follows. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The word manages the transition into an interview phase. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a natural topic-opening cue or a clean question when that is idiomatic. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the conversation begin rather than conclude? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Inventing a previous argument from which the question supposedly follows. The acceptable middle ground is the one that keeps the observed function—The word manages the transition into an interview phase.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 7: right

The central job of right in this group is to acknowledges, checks comprehension, or claims a shared premise. Its meaning cannot be chosen solely from a bilingual headword. First examine prosody, sequence position, and whether it is a question or a response. A common risk is converting a tentative check into a declaration of agreement. A workable first approach is to recover the interactional role before choosing punctuation. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 19: Right, I will send the draft tonight.

Source utterance: “Right, I will send the draft tonight.” Situation: A writer confirms what has been agreed after a task allocation. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Reading right as a spatial direction. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The first word signals acceptance and closure of negotiation. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use an acknowledgement equivalent and keep the future commitment intact. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the response accept the task without inventing approval of the draft? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Reading right as a spatial direction. The acceptable middle ground is the one that keeps the observed function—The first word signals acceptance and closure of negotiation.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 20: You received the updated file, right?

Source utterance: “You received the updated file, right?” Situation: A coordinator seeks confirmation before continuing. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Translating the tag as a rhetorical accusation. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker is checking receipt, not presuming negligence. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use the target language’s confirmation device with appropriate politeness. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the recipient answer no without feeling accused? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Translating the tag as a rhetorical accusation. The acceptable middle ground is the one that keeps the observed function—The speaker is checking receipt, not presuming negligence.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 21: Right. Let’s examine the second option.

Source utterance: “Right. Let’s examine the second option.” Situation: A moderator leaves a completed discussion and moves the agenda onward. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the word sound like a judgment that one option was morally correct. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker closes a unit and authorises a transition. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a phase-change cue rather than a truth-value label. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the audience follow the change of topic? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the word sound like a judgment that one option was morally correct. The acceptable middle ground is the one that keeps the observed function—The marker closes a unit and authorises a transition.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 8: okay

The central job of okay in this group is to signals acceptance, reluctant consent, or process management. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the response is genuine agreement or mere readiness to continue. A common risk is treating every okay as legal or procedural approval. A workable first approach is to read the next action and any visible resistance. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 22: Okay, I see the problem now.

Source utterance: “Okay, I see the problem now.” Situation: A colleague finally understands an explanation. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning comprehension into consent to a proposed solution. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker registers understanding only. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve the distinction between grasping the issue and approving a remedy. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would the person be expected to take action without further permission? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning comprehension into consent to a proposed solution. The acceptable middle ground is the one that keeps the observed function—The marker registers understanding only.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 23: Okay, but I need the figures first.

Source utterance: “Okay, but I need the figures first.” Situation: A manager gives conditional cooperation. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Dropping but and presenting unqualified consent. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: Acceptance is limited by an unmet dependency. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Maintain the boundary and the prerequisite plainly. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the request be acted on without the figures? It should not be. For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Dropping but and presenting unqualified consent. The acceptable middle ground is the one that keeps the observed function—Acceptance is limited by an unmet dependency.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 24: Okay, next question.

Source utterance: “Okay, next question.” Situation: A host ends one discussion and opens another. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Reading the word as satisfaction with the previous answer. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: This is a procedural transition, not an evaluation. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Choose a chairing cue or an economical transition. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the text preserve the host’s control of sequence? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Reading the word as satisfaction with the previous answer. The acceptable middle ground is the one that keeps the observed function—This is a procedural transition, not an evaluation.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 9: now

The central job of now in this group is to shifts to a new phase, warns, or places a time reference. Its meaning cannot be chosen solely from a bilingual headword. First examine whether now is temporal or discourse-organisational. A common risk is adding an immediate deadline where there is only a change of topic. A workable first approach is to test the sentence for real time reference versus rhetorical transition. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 25: Now, about the budget.

Source utterance: “Now, about the budget.” Situation: A chair moves to a new topic on a written agenda. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Demanding an immediate budget payment. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The word is an agenda switch rather than a clock instruction. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Select a topic transition, possibly with punctuation alone. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the text avoid creating a new payment deadline? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Demanding an immediate budget payment. The acceptable middle ground is the one that keeps the observed function—The word is an agenda switch rather than a clock instruction.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 26: Now, be careful with that cable.

Source utterance: “Now, be careful with that cable.” Situation: A supervisor notices a live physical risk. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Flattening the opening into neutral background information. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker helps make the warning urgent. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Retain the force of an immediate caution in idiomatic speech. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Will someone understand they must take care at once? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Flattening the opening into neutral background information. The acceptable middle ground is the one that keeps the observed function—The marker helps make the warning urgent.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 27: Now we have all the measurements.

Source utterance: “Now we have all the measurements.” Situation: A researcher has just completed the final test. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Treating now as merely a rhetorical paragraph opener. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: Here the word genuinely reports a changed state of knowledge. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Keep the temporal result and its relationship to completed work. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the reader infer that the measurements were not previously complete? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Treating now as merely a rhetorical paragraph opener. The acceptable middle ground is the one that keeps the observed function—Here the word genuinely reports a changed state of knowledge.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 10: look

The central job of look in this group is to calls attention, prefaces disagreement, or makes a direct appeal. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the imperative is literal seeing or a conversational summons. A common risk is turning an appeal to listen into an instruction to inspect an object. A workable first approach is to locate the attention target and any conflict trajectory. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 28: Look, the numbers do not match.

Source utterance: “Look, the numbers do not match.” Situation: A reviewer faces resistance from someone using an outdated spreadsheet. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the first word a visual instruction unrelated to the disagreement. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: It claims the floor to present corrective evidence. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use an attention-getter with no unintended insult. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the reader hear a firm evidence-based correction? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the first word a visual instruction unrelated to the disagreement. The acceptable middle ground is the one that keeps the observed function—It claims the floor to present corrective evidence.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 29: Look at the label before opening it.

Source utterance: “Look at the label before opening it.” Situation: A worker is given a physical inspection instruction. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Suppressing the literal requirement to check the object. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The verb actually governs the target and must remain actionable. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve the explicit direction to inspect the label. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would a worker visually check first? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Suppressing the literal requirement to check the object. The acceptable middle ground is the one that keeps the observed function—The verb actually governs the target and must remain actionable.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 30: Look, I am trying to help.

Source utterance: “Look, I am trying to help.” Situation: A volunteer responds to an accusation of indifference. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Rendering the marker as a threat or a command to stop speaking. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The word frames a defensive interpersonal appeal. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Choose a form that conveys urgency and frustration without excess aggression. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the speaker remain human rather than authoritarian? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Rendering the marker as a threat or a command to stop speaking. The acceptable middle ground is the one that keeps the observed function—The word frames a defensive interpersonal appeal.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 11: listen

The central job of listen in this group is to requests attention, introduces reassurance, or marks a conversational boundary. Its meaning cannot be chosen solely from a bilingual headword. First examine whether literal auditory attention or figurative floor-taking is intended. A common risk is overstating dominance or underplaying necessary urgency. A workable first approach is to select the level of interpersonal force supported by the scene. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 31: Listen, you have done enough for today.

Source utterance: “Listen, you have done enough for today.” Situation: A mentor reassures a tired colleague. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Changing reassurance into a reprimand. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker takes the floor to offer comfort and permission to rest. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a warm attention cue when appropriate. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would the hearer feel supported? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Changing reassurance into a reprimand. The acceptable middle ground is the one that keeps the observed function—The marker takes the floor to offer comfort and permission to rest.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 32: Listen carefully to the warning tone.

Source utterance: “Listen carefully to the warning tone.” Situation: An operator must identify a real alarm pattern. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Treating listen as a discourse filler. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The command requires actual auditory action. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Retain the literal listening instruction and specific object. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Will an operator know what sense to use? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Treating listen as a discourse filler. The acceptable middle ground is the one that keeps the observed function—The command requires actual auditory action.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 33: Listen, I cannot approve this without a review.

Source utterance: “Listen, I cannot approve this without a review.” Situation: A supervisor responds to pressure to skip a check. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the refusal sound like personal dislike. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker introduces a firm procedural boundary. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a controlled refusal preface and keep the review condition exact. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the hearer distinguish process limits from hostility? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the refusal sound like personal dislike. The acceptable middle ground is the one that keeps the observed function—The marker introduces a firm procedural boundary.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 12: by the way

The central job of by the way in this group is to introduces an incidental topic, a relevant reminder, or a delayed disclosure. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the new item is minor or merely delayed in the conversation. A common risk is downgrading important information into an optional footnote. A workable first approach is to evaluate practical importance separately from conversational packaging. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 34: By the way, the venue has changed.

Source utterance: “By the way, the venue has changed.” Situation: A planner remembers to mention a critical logistical update. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Treating the venue change as optional because it is introduced casually. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker frames a late-added topic, not its significance. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve the change visibly even if the transition is conversational. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Will every attendee notice the change? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Treating the venue change as optional because it is introduced casually. The acceptable middle ground is the one that keeps the observed function—The marker frames a late-added topic, not its significance.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 35: By the way, how was your interview?

Source utterance: “By the way, how was your interview?” Situation: A friend pivots from a technical discussion to personal interest. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning the aside into an abrupt demand. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The phrase manages a friendly topic change. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Select a light transition of appropriate intimacy. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the question sound considerate rather than intrusive? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning the aside into an abrupt demand. The acceptable middle ground is the one that keeps the observed function—The phrase manages a friendly topic change.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 36: Oh, by the way, the file password is in a separate message.

Source utterance: “Oh, by the way, the file password is in a separate message.” Situation: A sender adds a procedural reminder after sharing a document. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Misreading the note as proof that the document is already accessible. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker points to a second required communication. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Keep the reference and the difference between file and password. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the recipient locate the missing access information? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Misreading the note as proof that the document is already accessible. The acceptable middle ground is the one that keeps the observed function—The speaker points to a second required communication.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 13: mind you

The central job of mind you in this group is to qualifies a statement, points to an exception, or restores balance. Its meaning cannot be chosen solely from a bilingual headword. First examine which earlier claim is being limited. A common risk is rendering qualification as total reversal. A workable first approach is to preserve coexistence of two partly conflicting observations. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 37: The room is small. Mind you, it is very quiet.

Source utterance: “The room is small. Mind you, it is very quiet.” Situation: A visitor evaluates a work area with mixed features. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Canceling the size criticism altogether. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker introduces a compensating benefit. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a concessive qualifier that preserves both assessments. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can a reader see both the drawback and the advantage? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Canceling the size criticism altogether. The acceptable middle ground is the one that keeps the observed function—The speaker introduces a compensating benefit.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 38: It was cheap. Mind you, we had to assemble it ourselves.

Source utterance: “It was cheap. Mind you, we had to assemble it ourselves.” Situation: A buyer recalls a purchase. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Treating the assembly cost as a condition of the original sale. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The second fact qualifies the meaning of cheap. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a corrective afterthought that restores total effort to the evaluation. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the low price still remain true? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Treating the assembly cost as a condition of the original sale. The acceptable middle ground is the one that keeps the observed function—The second fact qualifies the meaning of cheap.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 39: You could ask Mina. Mind you, she is away this week.

Source utterance: “You could ask Mina. Mind you, she is away this week.” Situation: A colleague suggests a possible source of help. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Recasting the suggestion as a firm recommendation for immediate contact. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The second clause restricts short-term usefulness. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Keep the caveat linked to the suggested person. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would readers expect an immediate reply? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Recasting the suggestion as a firm recommendation for immediate contact. The acceptable middle ground is the one that keeps the observed function—The second clause restricts short-term usefulness.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 14: still

The central job of still in this group is to marks persistence despite resistance, adds a counterargument, or indicates continuation. Its meaning cannot be chosen solely from a bilingual headword. First examine whether it is temporal persistence or adversative concession. A common risk is mistaking despite-this for until-now or vice versa. A workable first approach is to test whether a duration reading is grammatically and contextually available. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 40: It is late. Still, we should finish the safety check.

Source utterance: “It is late. Still, we should finish the safety check.” Situation: A team weighs fatigue against an incomplete inspection. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Interpreting still as a time adverb before finish. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The word means nevertheless; obligation survives the obstacle. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a concessive link and retain the safety duty. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Will the check remain required despite lateness? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Interpreting still as a time adverb before finish. The acceptable middle ground is the one that keeps the observed function—The word means nevertheless; obligation survives the obstacle.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 41: The device is still charging.

Source utterance: “The device is still charging.” Situation: A user reports status after checking the indicator. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Replacing the duration meaning with a contrastive conjunction. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The process has not yet ended. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve ongoing state and avoid implying completion. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would the user know the device is not ready? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Replacing the duration meaning with a contrastive conjunction. The acceptable middle ground is the one that keeps the observed function—The process has not yet ended.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 42: The review was harsh. Still, the proposal improved.

Source utterance: “The review was harsh. Still, the proposal improved.” Situation: An author reflects on uncomfortable feedback. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the improvement a logical consequence guaranteed by harshness. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The second statement contrasts with the negative first impression. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a balanced contrast rather than a causal guarantee. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can both harsh review and improvement coexist? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the improvement a logical consequence guaranteed by harshness. The acceptable middle ground is the one that keeps the observed function—The second statement contrasts with the negative first impression.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 15: after all

The central job of after all in this group is to offers an overlooked reason, appeals to common ground, or justifies a decision. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the speaker cites a shared fact or introduces a new premise. A common risk is making a justification sound like an unquestionable axiom. A workable first approach is to keep the reason attributable and appropriately contestable. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 43: We can take our time; the deadline is next month, after all.

Source utterance: “We can take our time; the deadline is next month, after all.” Situation: A group is debating whether to rush a draft. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning the deadline fact into permission to ignore quality control. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The shared date is used as a reason against unnecessary speed. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve the conversational reason while leaving other obligations unchanged. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would readers infer only that there is more time? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning the deadline fact into permission to ignore quality control. The acceptable middle ground is the one that keeps the observed function—The shared date is used as a reason against unnecessary speed.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 44: After all, you wrote the first version.

Source utterance: “After all, you wrote the first version.” Situation: A reviewer asks the original author to clarify a design choice. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning the appeal to prior involvement into blame. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker cites relevant experience as a reason to ask. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a reminder rather than a reproach when the tone supports it. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the request respect the author’s knowledge? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning the appeal to prior involvement into blame. The acceptable middle ground is the one that keeps the observed function—The speaker cites relevant experience as a reason to ask.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 45: Maybe we should check again. After all, one value changed.

Source utterance: “Maybe we should check again. After all, one value changed.” Situation: A data analyst urges a second verification. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Converting the request into certainty that the whole analysis is wrong. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: A specific change justifies caution. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Retain proportionality between evidence and proposed action. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would a careful reader see a reason for checking rather than a verdict? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Converting the request into certainty that the whole analysis is wrong. The acceptable middle ground is the one that keeps the observed function—A specific change justifies caution.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 16: as it happens

The central job of as it happens in this group is to presents a timely coincidence, relevant new information, or modest disclosure. Its meaning cannot be chosen solely from a bilingual headword. First examine why the new fact is newly relevant. A common risk is injecting dramatic surprise where the speaker is merely connecting context. A workable first approach is to retain relevance without exaggerating improbability. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 46: As it happens, I have the updated diagram.

Source utterance: “As it happens, I have the updated diagram.” Situation: A designer offers a resource exactly when a colleague needs it. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Suggesting that possessing a diagram is astonishing. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The phrase introduces a fortuitously relevant fact. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use an understated relevance cue or natural offer. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the offer sound helpful rather than theatrical? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Suggesting that possessing a diagram is astonishing. The acceptable middle ground is the one that keeps the observed function—The phrase introduces a fortuitously relevant fact.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 47: As it happens, she asked the same question yesterday.

Source utterance: “As it happens, she asked the same question yesterday.” Situation: A colleague connects a present query with prior discussion. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning the coincidence into evidence of collusion. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The point is parallel timing and repeated concern. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve the coincidence without implying suspicion. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Will the reader infer only that the question has arisen before? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning the coincidence into evidence of collusion. The acceptable middle ground is the one that keeps the observed function—The point is parallel timing and repeated concern.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 48: As it happens, I am free after lunch.

Source utterance: “As it happens, I am free after lunch.” Situation: A friend responds to a proposed short meeting. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making availability sound unconditional for the whole day. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker reveals timely matching availability. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Keep the specified time window and the light coincidental tone. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the recipient schedule only the offered period? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making availability sound unconditional for the whole day. The acceptable middle ground is the one that keeps the observed function—The marker reveals timely matching availability.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 17: to be fair

The central job of to be fair in this group is to concedes a mitigating fact, seeks balance, or qualifies a prior criticism. Its meaning cannot be chosen solely from a bilingual headword. First examine what is being balanced and whose interests are involved. A common risk is inventing impartial adjudication when this is an informal self-correction. A workable first approach is to reproduce a measured concession without changing the underlying criticism. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 49: To be fair, the instructions were unclear.

Source utterance: “To be fair, the instructions were unclear.” Situation: A teammate discusses a failed attempt. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Erasing the failed attempt by treating the unclear instructions as a full excuse. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker adds a mitigating explanation. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Maintain shared responsibility as an open question. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the translation balance judgment without declaring innocence? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Erasing the failed attempt by treating the unclear instructions as a full excuse. The acceptable middle ground is the one that keeps the observed function—The speaker adds a mitigating explanation.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 50: The result was slow. To be fair, we changed the requirements twice.

Source utterance: “The result was slow. To be fair, we changed the requirements twice.” Situation: A manager evaluates a delayed project. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning changed requirements into proof that no one was responsible. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The concession supplies context for the delay. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve both the observed delay and the mitigating complexity. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the reader hold both facts at once? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning changed requirements into proof that no one was responsible. The acceptable middle ground is the one that keeps the observed function—The concession supplies context for the delay.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 51: To be fair, I could have explained that better.

Source utterance: “To be fair, I could have explained that better.” Situation: A speaker revises an earlier complaint. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the admission into acceptance of all fault. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The phrase signals self-accountability for one aspect. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Retain bounded self-criticism without expanding liability. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the speaker acknowledge a contribution without confessing total blame? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the admission into acceptance of all fault. The acceptable middle ground is the one that keeps the observed function—The phrase signals self-accountability for one aspect.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 18: frankly

The central job of frankly in this group is to announces directness, candid disclosure, or a deliberate breach of tact. Its meaning cannot be chosen solely from a bilingual headword. First examine whether directness is compassionate, irritated, or simply plain. A common risk is adding insults to a statement that is merely forthright. A workable first approach is to preserve communicative candour at the observed intensity. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 52: Frankly, I do not know the answer.

Source utterance: “Frankly, I do not know the answer.” Situation: A specialist responds to a difficult question. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Changing uncertainty into indifference toward the question. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker openly acknowledges a knowledge limit. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a candid admission without adding refusal or arrogance. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the listener hear epistemic restraint? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Changing uncertainty into indifference toward the question. The acceptable middle ground is the one that keeps the observed function—The speaker openly acknowledges a knowledge limit.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 53: Frankly, that deadline is unrealistic.

Source utterance: “Frankly, that deadline is unrealistic.” Situation: A coordinator pushes back against a proposed schedule. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the statement sound like contempt for its proposer. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker flags a direct assessment of feasibility. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve firmness and separate feasibility from personal judgment. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the proposal be reconsidered without an interpersonal attack? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the statement sound like contempt for its proposer. The acceptable middle ground is the one that keeps the observed function—The marker flags a direct assessment of feasibility.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 54: Frankly, I was relieved.

Source utterance: “Frankly, I was relieved.” Situation: A witness describes feelings after an outcome. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making candid disclosure into criticism of others. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker emphasises sincerity about an emotional state. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Choose an honest-emphasis device that fits reflective narration. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Is the emotion presented as genuine rather than provocative? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making candid disclosure into criticism of others. The acceptable middle ground is the one that keeps the observed function—The speaker emphasises sincerity about an emotional state.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 19: in fact

The central job of in fact in this group is to strengthens evidence, corrects a claim, or provides a more exact account. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the new clause confirms, contrasts, or specifies. A common risk is automatically turning fact into a contradiction. A workable first approach is to trace how the follow-up relates to the previous proposition. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 55: The output improved. In fact, the error rate halved.

Source utterance: “The output improved. In fact, the error rate halved.” Situation: An engineer upgrades a general performance report with a measured result. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Presenting the error figure as a separate unrelated fact. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker introduces more precise support for improvement. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Retain the evidence relationship and exact comparative claim. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would a reader recognise the figure as support? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Presenting the error figure as a separate unrelated fact. The acceptable middle ground is the one that keeps the observed function—The marker introduces more precise support for improvement.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 56: It was not a prototype. In fact, it was the final model.

Source utterance: “It was not a prototype. In fact, it was the final model.” Situation: A product manager corrects a mistaken label. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Keeping both labels as if compatible. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The second sentence explicitly replaces the first classification. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a firm correction connector or juxtaposition that cannot mislead. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the reader identify the correct model status? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Keeping both labels as if compatible. The acceptable middle ground is the one that keeps the observed function—The second sentence explicitly replaces the first classification.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 57: The team made progress. In fact, three issues were closed.

Source utterance: “The team made progress. In fact, three issues were closed.” Situation: A project lead provides a concrete instance of advancement. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Misrepresenting the number as all issues solved. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker substantiates a limited overall claim. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve the bounded count and the evidence relation. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the reader separate three closures from complete resolution? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Misrepresenting the number as all issues solved. The acceptable middle ground is the one that keeps the observed function—The marker substantiates a limited overall claim.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 20: on the other hand

The central job of on the other hand in this group is to compares alternatives, rebalances a judgment, or introduces an opposing aspect. Its meaning cannot be chosen solely from a bilingual headword. First examine the explicit or implicit first side and the comparison dimension. A common risk is converting comparison into mutually exclusive factual contradiction. A workable first approach is to reconstruct both sides of the evaluative frame. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 58: The train is faster. On the other hand, the bus is more flexible.

Source utterance: “The train is faster. On the other hand, the bus is more flexible.” Situation: A traveller compares transport choices for a trip. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the second fact disprove the first. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker identifies a competing consideration. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve two independent criteria and their trade-off. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can a reader choose based on speed or flexibility? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the second fact disprove the first. The acceptable middle ground is the one that keeps the observed function—The marker identifies a competing consideration.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 59: The new layout uses less space. On the other hand, it is harder to maintain.

Source utterance: “The new layout uses less space. On the other hand, it is harder to maintain.” Situation: A design team evaluates a revision. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Collapsing the two dimensions into one overall judgment. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The text compares compactness with maintenance burden. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Keep the frame of trade-offs visible. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the translation avoid announcing a winner? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Collapsing the two dimensions into one overall judgment. The acceptable middle ground is the one that keeps the observed function—The text compares compactness with maintenance burden.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 60: Some readers prefer detail. On the other hand, others need a quick summary.

Source utterance: “Some readers prefer detail. On the other hand, others need a quick summary.” Situation: An editor plans two views of a report. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Translating hand literally as a physical direction. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The connector marks audience contrast. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a discourse contrast appropriate to two groups. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can readers see why two formats might be justified? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Translating hand literally as a physical direction. The acceptable middle ground is the one that keeps the observed function—The connector marks audience contrast.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 21: moreover

The central job of moreover in this group is to adds an independently relevant reason, often in structured argument. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the added point is cumulative or causally dependent. A common risk is treating additive force as evidence that earlier premises are proven. A workable first approach is to build the stack of reasons without overclaiming causation. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 61: The method is slow. Moreover, it requires specialist equipment.

Source utterance: “The method is slow. Moreover, it requires specialist equipment.” Situation: A committee evaluates the cost of a procedure. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making equipment the cause of slowness when the text offers two concerns. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker adds a second disadvantage. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve independent cumulative reasons. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can a reader count two concerns rather than one chain? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making equipment the cause of slowness when the text offers two concerns. The acceptable middle ground is the one that keeps the observed function—The marker adds a second disadvantage.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 62: The plan saves energy. Moreover, it reduces maintenance.

Source utterance: “The plan saves energy. Moreover, it reduces maintenance.” Situation: A report presents two advantages. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Inserting a false quantitative comparison between benefits. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker adds a distinct benefit. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a formal additive connector with parallel structure. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does each claim remain separately testable? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Inserting a false quantitative comparison between benefits. The acceptable middle ground is the one that keeps the observed function—The marker adds a distinct benefit.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 63: The estimate is incomplete. Moreover, the assumptions are undocumented.

Source utterance: “The estimate is incomplete. Moreover, the assumptions are undocumented.” Situation: An auditor describes reasons not to approve a figure. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning the second defect into proof that the number is false. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: Two verification gaps justify further work. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve the separate audit defects and restrained conclusion. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the reader understand why review is necessary without assuming fraud? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning the second defect into proof that the number is false. The acceptable middle ground is the one that keeps the observed function—Two verification gaps justify further work.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 22: nevertheless

The central job of nevertheless in this group is to maintains an outcome, decision, or belief despite a countervailing reason. Its meaning cannot be chosen solely from a bilingual headword. First examine the strength of the rejected objection. A common risk is making concession disappear or making it sound logically impossible. A workable first approach is to preserve both the obstacle and the continued action. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 64: The draft has gaps. Nevertheless, it is useful for discussion.

Source utterance: “The draft has gaps. Nevertheless, it is useful for discussion.” Situation: A research group decides whether an imperfect document can help. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Asserting the draft has no gaps because it is useful. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker states use despite limitations. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Retain the limited-use qualification. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Could a reader use the draft while knowing it needs revision? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Asserting the draft has no gaps because it is useful. The acceptable middle ground is the one that keeps the observed function—The marker states use despite limitations.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 65: The route is longer. Nevertheless, it avoids the blocked road.

Source utterance: “The route is longer. Nevertheless, it avoids the blocked road.” Situation: A driver selects a practical detour. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Translating the longer distance as a benefit itself. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The concession highlights a disadvantage outweighed by access. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Represent the trade-off with no invented time guarantee. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the driver see the relevant operational reason? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Translating the longer distance as a benefit itself. The acceptable middle ground is the one that keeps the observed function—The concession highlights a disadvantage outweighed by access.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 66: The evidence is preliminary. Nevertheless, the question deserves study.

Source utterance: “The evidence is preliminary. Nevertheless, the question deserves study.” Situation: A scientist frames a cautious research proposal. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning preliminary evidence into established proof. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: Interest persists despite limited certainty. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Keep the exploratory rather than conclusive status. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the proposal remain a request for study? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning preliminary evidence into established proof. The acceptable middle ground is the one that keeps the observed function—Interest persists despite limited certainty.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 23: in other words

The central job of in other words in this group is to restates, simplifies, or makes an implication explicit. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the second formulation is equivalent or narrower. A common risk is smuggling in extra content under the label of paraphrase. A workable first approach is to audit entailment in both directions before keeping an equivalence signal. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 67: The device is not calibrated. In other words, do not use its readings for final reporting.

Source utterance: “The device is not calibrated. In other words, do not use its readings for final reporting.” Situation: A supervisor translates a technical condition into a work instruction. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Presenting the instruction as a universal scientific law for all devices. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The second clause explains a local operational consequence. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Retain the setting and avoid widening the policy. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the worker understand the scope of the instruction? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Presenting the instruction as a universal scientific law for all devices. The acceptable middle ground is the one that keeps the observed function—The second clause explains a local operational consequence.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 68: We need one approver. In other words, two signatures are not required.

Source utterance: “We need one approver. In other words, two signatures are not required.” Situation: A coordinator simplifies an approval workflow. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Changing the rule into a prohibition on multiple signatures. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The phrase clarifies the minimum requirement, not necessarily a maximum. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve the difference between unnecessary and forbidden. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Could an extra signature still be possible? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Changing the rule into a prohibition on multiple signatures. The acceptable middle ground is the one that keeps the observed function—The phrase clarifies the minimum requirement, not necessarily a maximum.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 69: She is unavailable this morning. In other words, try after lunch.

Source utterance: “She is unavailable this morning. In other words, try after lunch.” Situation: A receptionist offers a practical summary of a schedule. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Inventing a guarantee that she will be free later. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The paraphrase is pragmatic advice, not a firm appointment. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Translate with a suggestion and preserve uncertainty. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would a visitor treat the afternoon as confirmed? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Inventing a guarantee that she will be free later. The acceptable middle ground is the one that keeps the observed function—The paraphrase is pragmatic advice, not a firm appointment.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 24: for example

The central job of for example in this group is to introduces an illustration, a representative member, or an optional route. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the example is exhaustive or merely indicative. A common risk is converting one sample into the full list of permitted cases. A workable first approach is to guard the open or closed status of the category. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 70: Bring a form of identification, for example a passport.

Source utterance: “Bring a form of identification, for example a passport.” Situation: An event organiser names a possible document type. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making passport the only acceptable document. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The phrase introduces a nonexclusive illustration. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use an example marker and avoid closed-list punctuation. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can participants choose another acceptable form if policy permits? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making passport the only acceptable document. The acceptable middle ground is the one that keeps the observed function—The phrase introduces a nonexclusive illustration.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 71: Some materials absorb moisture, for example untreated paper.

Source utterance: “Some materials absorb moisture, for example untreated paper.” Situation: A teacher provides one specimen of a broader property. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Treating all paper varieties as identical. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The example is specifically bounded by the adjective untreated. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Keep the restriction and exemplar relationship. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would the reader avoid overgeneralising? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Treating all paper varieties as identical. The acceptable middle ground is the one that keeps the observed function—The example is specifically bounded by the adjective untreated.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 72: Try a shorter opening, for example a single question.

Source utterance: “Try a shorter opening, for example a single question.” Situation: An editor gives a writing suggestion. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning the suggestion into an obligatory template. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The example demonstrates a style option. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Retain optionality and illustrative purpose. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the writer still choose a different short opening? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning the suggestion into an obligatory template. The acceptable middle ground is the one that keeps the observed function—The example demonstrates a style option.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 25: that said

The central job of that said in this group is to qualifies the immediately preceding view while preserving part of it. Its meaning cannot be chosen solely from a bilingual headword. First examine what portion of the prior position survives. A common risk is replacing a partial qualification with total reversal. A workable first approach is to identify the surviving claim before translating the contrast. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 73: The evidence is limited. That said, the early pattern is encouraging.

Source utterance: “The evidence is limited. That said, the early pattern is encouraging.” Situation: A scientist discusses tentative results. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Converting encouragement into a proven outcome. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker permits cautious optimism without cancelling limits. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a controlled qualification at the same confidence level. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the reader remain appropriately cautious? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Converting encouragement into a proven outcome. The acceptable middle ground is the one that keeps the observed function—The marker permits cautious optimism without cancelling limits.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 74: I like the design. That said, the labels need work.

Source utterance: “I like the design. That said, the labels need work.” Situation: A reviewer gives balanced feedback. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the design praise sarcastic or insincere. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The next concern narrows a positive assessment. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve genuine praise and a specific revision request. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Could a designer keep the overall concept and fix the labels? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the design praise sarcastic or insincere. The acceptable middle ground is the one that keeps the observed function—The next concern narrows a positive assessment.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 75: The journey is possible. That said, it will take preparation.

Source utterance: “The journey is possible. That said, it will take preparation.” Situation: An adviser weighs feasibility and effort. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Changing possibility into impossibility. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The phrase acknowledges effort while keeping feasibility open. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a concessive adjustment rather than an outright rejection. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the reader understand preparation is the new condition? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Changing possibility into impossibility. The acceptable middle ground is the one that keeps the observed function—The phrase acknowledges effort while keeping feasibility open.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 26: oh

The central job of oh in this group is to marks discovery, recognition, sympathy, or response to new information. Its meaning cannot be chosen solely from a bilingual headword. First examine the emotional state and the event that triggered it. A common risk is flattening surprise or exaggerating it into shock. A workable first approach is to read adjacent turns before matching an interjection. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 76: Oh, now I understand.

Source utterance: “Oh, now I understand.” Situation: A learner responds after a helpful explanation. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Omitting the discovery and making understanding sound previously settled. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The interjection signals the moment comprehension arrives. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a recognition cue proportionate to the exchange. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the listener hear a change of understanding? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Omitting the discovery and making understanding sound previously settled. The acceptable middle ground is the one that keeps the observed function—The interjection signals the moment comprehension arrives.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 77: Oh, I am sorry to hear that.

Source utterance: “Oh, I am sorry to hear that.” Situation: A friend learns of disappointing news. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Treating oh as excitement. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The word prepares an empathetic response. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Choose a sympathy-compatible recognition form. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would the recipient feel heard? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Treating oh as excitement. The acceptable middle ground is the one that keeps the observed function—The word prepares an empathetic response.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 78: Oh? I thought the meeting was later.

Source utterance: “Oh? I thought the meeting was later.” Situation: An attendee reacts to an unexpected timetable. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Rendering the question as anger. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The brief interjection registers mild surprise and invites clarification. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Retain uncertainty and room for correction. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the other person explain the schedule without defensiveness? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Rendering the question as anger. The acceptable middle ground is the one that keeps the observed function—The brief interjection registers mild surprise and invites clarification.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 27: hmm

The central job of hmm in this group is to signals thought, doubt, a search for words, or reserved disagreement. Its meaning cannot be chosen solely from a bilingual headword. First examine whether the sound is a pause in reasoning or a negative evaluation. A common risk is translating any hesitation into refusal. A workable first approach is to calibrate the target language’s written representation and contextual meaning. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 79: Hmm, that might work.

Source utterance: “Hmm, that might work.” Situation: A designer considers an unfamiliar alternative. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Presenting the idea as already accepted. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The sound conveys ongoing assessment. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use an idiomatic thinking cue or a measured qualified response. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the proposal remain under consideration? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Presenting the idea as already accepted. The acceptable middle ground is the one that keeps the observed function—The sound conveys ongoing assessment.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 80: Hmm. The totals are different.

Source utterance: “Hmm. The totals are different.” Situation: An accountant notices a discrepancy. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Rendering the hesitation as approval. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The sound precedes concerned scrutiny. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Choose a brief caution or reflective cue without fabricating a verdict. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Will a reader expect further checking? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Rendering the hesitation as approval. The acceptable middle ground is the one that keeps the observed function—The sound precedes concerned scrutiny.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 81: Hmm, let me think.

Source utterance: “Hmm, let me think.” Situation: A friend is asked to recall an old address. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Transforming reflection into unwillingness to help. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker creates a moment for memory retrieval. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a natural pause or response formula. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the speaker still intend to search memory? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Transforming reflection into unwillingness to help. The acceptable middle ground is the one that keeps the observed function—The marker creates a moment for memory retrieval.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 28: uh-huh

The central job of uh-huh in this group is to may be a continuer, a sign of agreement, or sceptical irony. Its meaning cannot be chosen solely from a bilingual headword. First examine prosody and whether the other speaker has finished their account. A common risk is promoting an attentive listener’s minimal cue to substantive endorsement. A workable first approach is to separate receipt of information from acceptance of its content. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 82: Uh-huh. Go on.

Source utterance: “Uh-huh. Go on.” Situation: An interviewer encourages a witness to continue. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Treating the brief sound as belief in the entire account. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: This is primarily a continuer. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a minimal listening acknowledgement that does not ratify facts. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the witness continue without assuming endorsement? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Treating the brief sound as belief in the entire account. The acceptable middle ground is the one that keeps the observed function—This is primarily a continuer.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 83: Uh-huh, that’s the one.

Source utterance: “Uh-huh, that’s the one.” Situation: A person recognises an item shown on a screen. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Undermining a clear confirmation with unjustified scepticism. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The cue confirms identity in context. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use a recognition equivalent rather than mere passive listening. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would the colleague select the identified item? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Undermining a clear confirmation with unjustified scepticism. The acceptable middle ground is the one that keeps the observed function—The cue confirms identity in context.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 84: Uh-huh. And you expect that to work?

Source utterance: “Uh-huh. And you expect that to work?” Situation: A critic reacts to an implausible proposal. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Making the first sound sound enthusiastic. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The following question reveals sceptical irony. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Translate the stance using the whole turn and suitable punctuation. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can readers perceive doubt without adding a personal insult? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Making the first sound sound enthusiastic. The acceptable middle ground is the one that keeps the observed function—The following question reveals sceptical irony.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 29: like

The central job of like in this group is to marks approximation, introduces a sample, or signals represented speech. Its meaning cannot be chosen solely from a bilingual headword. First examine whether like is a comparison, a quotative marker, or conversational filler. A common risk is removing semantic approximation or preserving gratuitous filler. A workable first approach is to identify the syntactic role and associated uncertainty. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 85: It took like twenty minutes.

Source utterance: “It took like twenty minutes.” Situation: A speaker gives a rough travel estimate. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Turning the number into an exact measured duration. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: Like marks approximation in this informal utterance. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve approximate time without awkward literal comparison. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Will a reader know the duration was estimated? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Turning the number into an exact measured duration. The acceptable middle ground is the one that keeps the observed function—Like marks approximation in this informal utterance.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 86: She was like, ‘We need another day.’

Source utterance: “She was like, ‘We need another day.’” Situation: A narrator represents someone else’s gist rather than a certified quotation. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Presenting the wording as an exact transcript. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The quotative introduces enacted or approximate reported speech. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Select a natural report form and avoid overclaiming verbatim accuracy. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would a reader mistake an impression for a precise quote? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Presenting the wording as an exact transcript. The acceptable middle ground is the one that keeps the observed function—The quotative introduces enacted or approximate reported speech.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 87: It’s like a small workshop, but quieter.

Source utterance: “It’s like a small workshop, but quieter.” Situation: A visitor describes an unfamiliar studio to a friend. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Treating the comparison as categorical identity. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The speaker uses resemblance to aid understanding. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Keep the analogy and its qualification. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can the reader imagine likeness without equating the places? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Treating the comparison as categorical identity. The acceptable middle ground is the one that keeps the observed function—The speaker uses resemblance to aid understanding.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Mechanism 30: sort of

The central job of sort of in this group is to hedges category membership, softens judgment, or acknowledges partial resemblance. Its meaning cannot be chosen solely from a bilingual headword. First examine whether approximation concerns factual classification or social stance. A common risk is forcing a binary answer not present in the source. A workable first approach is to identify the precise proposition being weakened. Those instructions are function-level; they do not assert an identical particle exists in every language.

Diagnostic 88: It was sort of a rehearsal.

Source utterance: “It was sort of a rehearsal.” Situation: A performer describes a trial event with some but not all rehearsal features. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Classifying the session as an official rehearsal. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The phrase signals uncertain category membership. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Preserve the near-but-not-quite status. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Would the reader avoid assigning formal rehearsal obligations? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Classifying the session as an official rehearsal. The acceptable middle ground is the one that keeps the observed function—The phrase signals uncertain category membership.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 89: I sort of agree.

Source utterance: “I sort of agree.” Situation: A reviewer recognises merit but retains reservations. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Recording the comment as full approval. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The marker narrows agreement to part of a proposal. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Use partial endorsement or explicit limitation. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Can a decision-maker see that consensus is incomplete? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Recording the comment as full approval. The acceptable middle ground is the one that keeps the observed function—The marker narrows agreement to part of a proposal.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Diagnostic 90: That sounded sort of rude.

Source utterance: “That sounded sort of rude.” Situation: A friend gives careful feedback on a remark. The key is not to count whether the target has one discourse particle or two. The question is what conversational transition a reader hears at this exact point, and what relation that creates between the people involved.

The tempting but misleading version: Converting tentative social judgment into an absolute accusation. That result could be fluent at sentence level while materially changing the interaction. It would cause a reader to assign the wrong strength to a commitment, to overlook a correction, or to treat a conversational aside as new propositional evidence. In a brief subtitle or a cleaned-up record, the temptation is particularly strong because short tokens look dispensable even when the structure they signal remains necessary.

Function diagnosis: The hedge softens interpersonal criticism. Here the analyst should separate what is being asserted from how the assertion is positioned. If the marker is deleted, the source’s underlying facts may survive, but that is not by itself an adequate test. The translated line must still fit the preceding turn, indicate the same kind of participation or movement, and leave a sensible place for the next response.

Transfer decision: Translate a cautious impression rather than a categorical verdict. A good target candidate may relocate the pragmatic work to a different position, distribute it across a short phrase, or retain it through a structurally clear juxtaposition. A literal rendering is preferable only if native readers understand the same function at this point in the dialogue. When no safe reading is available from the text, a translator should query context rather than manufacture emotional certainty.

Diagnostic question: Does the hearer retain room to clarify intent? For review, read the two turns before and after the chosen version, remove any explanation of the translator’s intention, and ask a competent target reader what happened socially. Compare that answer with the source: who agreed, who merely listened, what was corrected, what remains optional, and what was promised. If the likely action changes, revise the cue even if the translation sounds polished.

Contrast and transfer: Try a controlled counter-version that omits the marker entirely, then a second version that overstates it. Record the specific interpretive difference between those extremes. For this case the crucial warning is: Converting tentative social judgment into an absolute accusation. The acceptable middle ground is the one that keeps the observed function—The hedge softens interpersonal criticism.—without adding a speaker attitude unsupported by the scene. This three-version exercise is useful for translator training because it makes an otherwise invisible interpersonal error audible.

Failure modes: why plausible translations go wrong

Failure 1: The automatic-zero error

Failure pattern: Deleting every particle because it contributes no new object, event or number. Repair: Compare a bare proposition with the original interaction. If the bare version sounds more certain, harsher, less caring or more approving, it is not functionally equivalent. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 2: The automatic-literal error

Failure pattern: Assigning one dictionary equivalent to every occurrence of a common marker. Repair: Sort uses by speaker action, not spelling. For example ‘right’ can be a tag, an acknowledgement or an agenda pivot. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 3: The agreement inflation error

Failure pattern: Treating backchannels such as ‘uh-huh’ as agreement with every claim made by the speaker. Repair: Observe whether the listener takes substantive responsibility for the proposition; a continuer often invites further speech only. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 4: The scepticism inflation error

Failure pattern: Rendering ‘hmm’ or ‘oh?’ as an accusation. Repair: Look for an actual contrastive follow-up, confrontation or dismissive intonation before selecting an antagonistic form. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 5: The false correction error

Failure pattern: Rendering ‘actually’ as a strong contradiction when it is only an unexpected disclosure. Repair: Identify the assumption being revised and how firmly the source challenges it. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 6: The lost correction error

Failure pattern: Deleting ‘actually’ when readers need to know a previous date or classification was wrong. Repair: Show the repaired item visibly enough that recipients can update an instruction or record. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 7: The procedural approval error

Failure pattern: Translating ‘okay’ as an authorisation instead of comprehension or readiness. Repair: Distinguish ‘I understand’ from ‘I approve’. Review what action would be justified after the target line. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 8: The categorical promise error

Failure pattern: Converting ‘well, I can try’ or ‘sort of agree’ into an unconditional commitment. Repair: Audit modal strength, conditions and hedging across the complete turn. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 9: The literal-attention error

Failure pattern: Treating ‘look’ or ‘listen’ as a physical instruction when the speaker is taking the floor. Repair: Ask if a real object or sound is present. Without one, the form may be interactional. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 10: The missing-physical-instruction error

Failure pattern: Dropping a literal look or listen command as if it were filler. Repair: If the following complement names a label or alarm, preserve the required sensory action. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 11: The topic-trajectory error

Failure pattern: Rendering ‘anyway’ as a contrast even though it resumes an interrupted task. Repair: Trace the earlier topic, digression and return. Use a resumption cue if the target needs one. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 12: The deadline-invention error

Failure pattern: Interpreting agenda-opening ‘now’ as a temporal command. Repair: Replace the cue with ‘turning to’ as a test. If meaning remains, the marker likely organises discourse. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 13: The evidence-overreach error

Failure pattern: Using ‘in fact’ to turn a specific supporting statistic into a claim that everything is settled. Repair: Keep examples, counts and evidence bounded to what was reported. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 14: The example-exhaustion error

Failure pattern: Turning ‘for example’ into ‘namely’ or a closed list. Repair: Ask whether other members of the category remain possible. If yes, preserve non-exhaustiveness. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 15: The concession-cancellation error

Failure pattern: Translating ‘mind you’, ‘still’ or ‘that said’ as total rejection of the previous claim. Repair: Underline the part of the first claim that remains true after the qualification. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 16: The causal-overreading error

Failure pattern: Making ‘moreover’ or a turn-opening ‘so’ assert causation without textual support. Repair: Separate additive reasoning, inferred consequence and conversation management. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 17: The false-paraphrase error

Failure pattern: Allowing ‘in other words’ to introduce a stronger obligation than the first clause. Repair: Test whether the second formulation is entailed by the first under the local policy and context. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 18: The register distortion error

Failure pattern: Using formal argumentative connectors for intimate dialogue or conversational particles in a formal report. Repair: Specify the expected social setting before selecting a formulation. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 19: The character-voice erasure error

Failure pattern: Standardising every speaker’s interaction cues to the same neutral prose. Repair: Preserve distinctive interpersonal habits where they are narratively meaningful and supported by the source. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 20: The character-voice exaggeration error

Failure pattern: Adding slang, dialect markers or insolence to simulate an attitude the source does not license. Repair: Maintain character differentiation without relying on invented stereotypes. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 21: The false-verbatim error

Failure pattern: Treating a ‘like’ quotative as proof of exact words. Repair: Identify whether the source offers approximate enacted speech or a certified quotation. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 22: The factive-stance confusion error

Failure pattern: Treating social cues as evidence claims, and vice versa. Repair: Separate ‘I hear you’ from ‘I confirm that statement is true’. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 23: The isolated-string error

Failure pattern: Translating a marker in a user-interface resource without the preceding state or next action. Repair: Retrieve UI state, trigger, button function and expected response before writing the string. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 24: The overannotation error

Failure pattern: Explaining every omitted marker inside a public-facing translation. Repair: Keep the text natural; store technical rationale in the translator’s review notes when necessary. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 25: The underreview error

Failure pattern: Proofreading only spelling, punctuation and factual nouns. Repair: Add a dedicated pragmatic review in which a reader reconstructs stance and conversational moves. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 26: The source-quality error

Failure pattern: Inferring original dialogue intent from a paraphrased or edited transcript. Repair: Distinguish source evidence from reconstruction and preserve unresolved ambiguity. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 27: The subtitle-timing error

Failure pattern: Retaining every filler mechanically and obscuring the actual communicative event on screen. Repair: Budget characters and time for functional meaning; reduce surface material only with explicit purpose and review. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 28: The euphemism error

Failure pattern: Softening a necessary warning merely because the speaker used a conversational preface. Repair: Separate politeness framing from operational content; do not make the substantive hazard disappear. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 29: The adversarial-error error

Failure pattern: Overcorrecting a mildly frank statement into moral accusation. Repair: Use the least forceful target form that reliably communicates the observed challenge. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Failure 30: The blind-post-edit error

Failure pattern: Accepting machine output because the vocabulary is accurate and the sentences read smoothly. Repair: Evaluate the target as a human exchange: who now believes what, who agreed to what, and what the next person would infer. In practice the reviewer should create a concrete test by comparing an ordinary recipient’s likely inference from the source with that from the target. This turns vague comments such as ‘the tone feels wrong’ into a falsifiable concern: the text has created or removed consent, dismissal, certainty, causal logic, politeness, urgency or an open choice. When two natural candidates remain, record why the chosen one offers lower interpretive risk in this communication setting.

Transfer workshops: adapting the method across media

Transfer workshop 1: Subtitles with tight reading time

A two-speaker scene contains repeated ‘well’, ‘you know’, and ‘right’. Preserve the one cue that explains a turning point and sacrifice nonessential repetition only if the relationship remains clear. Compare the meaning conveyed through pacing and visible expression with the target line. Never assume silent facial context that the source does not show.

What to audit: Timing, turn attribution, dramatic intent, reader speed and the next utterance. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Transfer workshop 2: Documentary interview transcript

A witness responds ‘uh-huh’ repeatedly while an interviewer describes events. Do not rewrite these as affirmative factual statements. Mark turn-taking and choose a transcription convention for minimal responses. If the assignment requires a clean transcript, keep an auditable distinction between acknowledged speech and confirmed claims.

What to audit: Verbatim level, speaker identity, punctuation, whether the cue is receipt or endorsement. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Transfer workshop 3: Customer-service conversation

A representative says ‘Okay, I see the issue’ before checking entitlement. Avoid translating the line as ‘approved’. Preserve acknowledgement and note that eligibility remains unverified. The operational consequence matters more than the surface particle.

What to audit: Understanding versus authorisation, next procedural step, promise boundaries. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Transfer workshop 4: Workplace meeting

A chair says ‘Right, next item’ after a heated discussion. Translate the chairing transition rather than a judgment about who won. In minutes, the marker may be omitted while the topic boundary is documented; in a transcript, the transition may remain overt.

What to audit: Agenda boundary, vote status, recorded decision and medium. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Transfer workshop 5: Literary character dialogue

A character habitually says ‘look’ before challenging other characters. Do not convert every occurrence into the same rude imperative, but do not erase a recurring speech habit that shapes character. A translation may distribute the force across sentence openings and rhythm.

What to audit: Voice consistency, power relation, scene tempo, pattern across chapters. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Transfer workshop 6: Instructional video

The narrator says ‘Now, look at the display’ while demonstrating a step. The first word may mark sequence and the second is a literal action instruction. A rushed subtitle that removes both loses step timing and the required visual target.

What to audit: Action order, image synchrony, literal versus discourse use. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Transfer workshop 7: Written editorial

A writer uses ‘on the other hand’, ‘moreover’, and ‘nevertheless’ to separate a trade-off, added evidence and concession. Substitution with one generic connector flattens reasoning. Build an argument map before rephrasing.

What to audit: Reason type, counterpoint survival, audience expectation. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Transfer workshop 8: Machine-translation post-editing

An output deletes all ‘I mean’ turns and presents revised deadlines as if no correction occurred. The post-editor should reconstruct the repair chain and restore the changed-information status. A comparison of token matches alone will not detect the pragmatic error.

What to audit: Context window, revisions, source and target turn sequence. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Transfer workshop 9: Voiceover for an explanatory film

A guide uses ‘so’, ‘anyway’ and ‘for example’ to move between stages. Target prosody can carry some transitions, but the script still needs identifiable topics and illustration boundaries. Read the version aloud with the visuals.

What to audit: Pace, transition logic, item exhaustiveness and oral naturalness. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Transfer workshop 10: Accessibility and captions

A speaker’s ‘hmm’ conveys thoughtful reservation before a cautious answer. Depending on caption specification, use a natural interjection or a meaningful notation if the sound cannot be recovered from text. Avoid inventing emotional labels.

What to audit: Access to sound, caption convention, epistemic degree and viewer inference. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Transfer workshop 11: Research interview translation

An interviewee says ‘sort of’ when describing uncertain category membership. The researcher may be studying the extent of agreement or experience, so the hedge matters. Treating the answer as definite can bias thematic coding.

What to audit: Uncertainty, methodological documentation, consent boundaries and data interpretation. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Transfer workshop 12: Live conversation preparation

An interpreter or bilingual facilitator hears ‘actually’ before a polite correction. The communicative priority is to prevent a wrong shared understanding while keeping the speaker’s interpersonal intent. A later clarifying question may be preferable to a speculative harsh phrasing.

What to audit: Speaker intention, immediate recipient action, correction target and timing. Write a two-column mini-log headed ‘source function’ and ‘target function’. Identify the decisive signal and the risk if it is lost. Then draft three solutions: explicit cue, structural cue, and cue omission with functional compensation. Read each as the audience would, without giving them your notes. If any draft changes agreement, uncertainty, the reason relationship, or the social role of a speaker, reject it. This is a practical comparison exercise, not a claim that a single form is universally correct.

Four-week practical training and quality-control plan

Week 1 — Identify before translating

Collect 25 short exchanges from rights-cleared or original material. Mark each candidate token as literal content, pragmatic marker, unresolved, or multifunctional. For each marker list the preceding turn and following response. Revisit the same item without its neighbours and record how often classification changes. The aim is to discover which context cannot safely be omitted.

At the end of the week, retain a compact decision record for the most difficult cases: source excerpt, preceding and following turns, proposed target, intended communicative action, alternative that was rejected, and a short test explaining why. This corpus of decisions becomes an accountable training resource. The goal is fewer avoidable misunderstandings, not an artificial target percentage of markers preserved.

Week 2 — Compare target functions

For 20 examples, draft a literal candidate, a natural functional candidate and a deliberately overstrong candidate. Ask a proficient target-language reviewer which reply or action follows most naturally. Document disagreement. Build a small function-level glossary with notes, but do not enter a one-to-one particle equivalent as a universal rule.

Week 3 — Review real production constraints

Apply the mapping to three different media: a subtitle excerpt, a meeting transcript and a written explanation. Note where timing, formal register and documentary fidelity require different realisations. Check every consent, correction, commitment, example and qualification for scope drift. Where source audio or context is unavailable, flag an uncertainty rather than inventing tone.

Week 4 — Independent validation and transfer

Exchange anonymised extracts with an independent reviewer. Ask them to reconstruct the pragmatic action of each turn without seeing the source analysis. Compare their account with the source evidence. Classify errors by type, revise the function map, and maintain examples of justified overt translation and justified omission. Document the decision that produced better understanding, not just a preference for smoother prose.

Scoring a pragmatic-fidelity review

Use a five-axis review rather than a single fluency score. Mark content accuracy (are the facts unchanged?), speaker action (is this still an offer, refusal, acknowledgement or repair?), discourse relation (is the following statement an example, cause, concession or new topic?), stance and register (are confidence, warmth, friction and formality supported?), and recipient effect (would a reasonable reader respond or act in the intended way?). A translation that fails any critical axis should be revised even if the remaining sentences are elegant.

For repeatable audits, score each axis from 0 to 2: 0 is a material functional distortion; 1 is a recoverable or uncertain difference; 2 is adequately supported. A total is a working internal diagnostic, not an industry-standard certification metric. More important than the sum is the nature of a zero: a misreported approval or safety instruction is more consequential than a small loss of conversational warmth. Review priority therefore follows the risk of the use case, not the mere frequency of the marker.

Glossary of practical terms

Discourse marker

A word or phrase that organises, manages or colours discourse without always adding a separate truth-conditional proposition. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Pragmatic marker

A cue whose interpretation depends on the interaction, context, speaker and expected audience inference. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Discourse particle

A short functional item used to signal stance, interactional relation or conversational organisation; definitions vary by linguistic tradition. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Turn

A contribution by one speaker within an exchange. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Adjacency pair

A recognisable two-part conversational relationship, such as invitation–response or question–answer. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Backchannel

A brief listener signal that may encourage continued speech without necessarily expressing agreement. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Repair

A move that corrects, narrows or otherwise revises prior wording or understanding. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Stance

A speaker’s attitude, certainty, evaluation or position toward an expressed matter. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Common ground

Information or assumptions treated by participants as mutually available for the exchange. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Concession

A relationship in which a speaker acknowledges one point while retaining another that might initially seem in tension. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Topic resumption

Returning to an earlier subject after an interruption or digression. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Topic shift

Moving to a new conversational subject. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Functional equivalent

A target-language construction that performs a comparable communicative action in a specified context, not a universally interchangeable word. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Zero realisation

A translation choice in which no overt marker appears because its function is recoverable through other target features. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Compensation

Use of another available linguistic or contextual resource to preserve an effect that cannot be transferred at the same surface position. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Back-translation test

A review practice in which the target’s practical meaning and function are paraphrased back to see whether speaker intent has drifted. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Overtranslation

Adding stronger or more specific commitments, emotions or logic than the source supports. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Undertranslation

Omitting functional meaning that is relevant for interpretation or action. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Register

The pattern of formality, relationship and communicative setting reflected in language choice. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Pragmatic fidelity

Retention of the source’s supported interactional action, stance and relationship in the receiving context. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Post-editing

Human revision of automatically generated translation output according to the assignment’s requirements. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Review specification

A documented account of intended audience, language variety, purpose, accuracy requirements and acceptable editorial intervention. A reviewer should interpret this term as a working analytical tool rather than a universal rule for every language. Consider what evidence in the local exchange would confirm or challenge the classification, and whether the reader-facing translation genuinely needs the distinction stated explicitly.

Frequently asked questions

1. Should every discourse marker be translated?

No. A marker’s function must be preserved when it affects meaning, interaction or discourse structure, but the target may express that function through grammar, word order, punctuation or an overt equivalent. A blanket delete or always-keep policy is unreliable.

A useful practical check is to place the answer into a two-turn exchange and ask a proficient reader what commitment, relationship or transition is being signalled. If multiple readings remain plausible, seek additional context rather than imposing a universal translation rule.

2. Is ‘well’ the same as ‘good’ in translation?

Not automatically. In conversation, ‘well’ may prepare a qualified response or correction. The context must determine whether it is a discourse marker, adverb, adjective or other construction.

3. How should ‘you know’ be translated?

Check whether the speaker appeals to shared knowledge, asks for recognition or softens a claim. Do not turn it into proof that the listener already knows the information.

4. Can ‘okay’ mean approval?

Sometimes, but it can also mean comprehension, willingness to proceed or merely a phase change. Approval requires evidence beyond the token when decisions or permissions matter.

5. What is the difference between a filler and a discourse marker?

Some hesitation items are chiefly processing cues, while others structure the exchange or signal stance. Real utterances can perform more than one function, so a strict binary classification may be unhelpful.

6. Can machine translation preserve pragmatic markers?

It can often generate fluent equivalents, but it can also omit or overstate short interactional cues. Human review should compare whole turns and possible recipient interpretations, especially in consequential contexts.

7. Is omitting ‘anyway’ always acceptable?

No. If the marker visibly returns to a suspended topic, closes a digression or supports a courteous exit, the relationship may need a target-language signal even if the source word itself is absent.

8. Does ‘actually’ always correct another person?

No. It can express surprise or introduce a tactful request. Find what assumption, if any, is being revised.

9. Can discourse markers be translated in literary dialogue?

Yes, but voice and scene matter. The translator may need patterns of hesitation, qualification or challenge across a character’s dialogue rather than isolated lexical replacements.

10. How can translators avoid creating false consent?

Review acknowledgements and backchannels separately from promises, approvals and agreements. Ask what action the translated answer would authorise, then compare that with the source.

11. What if audio or context is missing?

Do not invent a confident tone reading. Preserve reasonable ambiguity, flag the decision for review, or query the source owner when a meaningful difference could affect interpretation.

12. What is the fastest useful QA check?

For each risky marker, ask three questions: what did the speaker do, how does it connect to the preceding turn, and what would a reader do next? Those answers reveal many functional mismatches.

13. Do discourse markers matter in written reports?

Yes. Connectors such as ‘moreover’, ‘nevertheless’, ‘for example’ and ‘in other words’ encode argumentative relationships. Changing them can distort evidence, boundaries and conclusions.

14. Can a punctuation change replace a marker?

Sometimes. A dash, pause or paragraph boundary may make a transition clear, but punctuation alone will not always preserve concession, commitment or interpersonal stance.

15. How do you review subtitles with limited character space?

Prioritise the communicative turning point and the possibility of false implications. Use timing, grammar and dialogue structure to carry information when a surface token cannot fit; verify rather than assume equivalence.

16. What makes this different from general grammar translation?

The central object is not grammatical correctness but the function of small expressions within connected talk and writing. A sentence may be grammatically excellent yet signal the wrong attitude or action.

17. Can speakers use the same marker for opposite effects?

Yes. ‘Uh-huh’ can continue a story, confirm a selected item or imply scepticism depending on sequence and prosody. Stable one-word glosses do not capture that variation.

18. How should a reviewer handle two equally natural solutions?

Compare their expected reader inferences and the assignment’s register. Choose the one that makes fewer unsupported assumptions; record the unresolved alternative if consequential.

19. Do standards prescribe literal particle retention?

Do not treat ISO process standards as dictionaries of particle equivalence. They concern translation-service processes and review specifications, not a universal one-to-one rendering of discourse markers.

20. What should the translator learn next?

Study presupposition for hidden assumptions, discourse reference for participant tracking, and the translation root for project-wide specifications and quality evaluation. Vocabulary and English-usage hubs are relevant when the immediate problem is broader language mastery rather than this chapter’s narrow pragmatic task.

Sources and research notes

Research consulted and verified on 5 November 2026. The article’s analytical workflows, scenarios, diagnostics and scoring rubric are original editorial methods; they are not represented as formal requirements of the authorities cited below.

Cambridge Dictionary, English Grammar Today: Discourse markers (so, right, okay) — identifies conversational organisation and attitude functions, including use of well, right and anyway.
Morozova, Discourse markers in English and European Portuguese translations — corpus-based discussion of functional counterparts and omission or addition of markers.
ISO 17100:2015, Translation services — Requirements for translation services — source for translation service processes and specified quality requirements; the 2015 edition remains current while a new edition is under development.
ISO 18587:2017, Post-editing of machine translation output — Requirements — published requirements for full human post-editing of machine translation output; a replacement draft is in development.
W3C Internationalization: Localization vs. internationalization — distinguishes linguistic translation from broader adaptation for an intended audience.
W3C Internationalization Glossary — terminology relating to language metadata, localization and internationalised text.
The role of discourse markers in translation — discusses why omission can retain propositional content while losing conversational effects.

Next learning routes without replacing protected topic owners

Continue at the Master Art of Translation root for the complete system of meaning transfer, translation project specifications and further specialist chapters. The preceding Presupposition Boundary Fidelity chapter deals with assumptions carried into statements, questions and negation; this chapter deals instead with the pragmatic job of small markers in a connected exchange. The two problems interact, but they are not interchangeable.

When the practical problem is English sentence construction, lexical growth or learning to choose accurate words, follow the existing Vocabulary Learning Hub without duplicating its teaching programme. When the problem is how English discourse and translation work more broadly, use the protected How English Works | Translation and Meaning Across Languages route. This page supplies a specialist diagnostic, not a replacement hub.

Final principle: a small expression can change a large inference. Translate the social move, not only the surface word. Keep factual claims, implied commitments, discourse relations and speaker relationships recoverable for the next reader. Where the evidence does not decide the function, acknowledge the uncertainty instead of inventing it.