If you want to know how to translate turn-taking and interruptions in conversation, begin with a deceptively simple question: who has the right to speak at this exact moment? A faithful translation of an interview, video call, novel, podcast, meeting or subtitle sequence must preserve more than the words each person says. It must also show who starts a turn, who invites another person to speak, who talks over whom, who yields, and whether an interruption is supportive, competitive, urgent or accidental.
The central mistake in translating overlapping dialogue is to turn a living exchange into an orderly sequence of perfect sentences. A translator may make every line grammatical and every speaker polite while quietly changing the relationship between them. An interruption that shuts someone down can become a friendly addition; a collaborative completion can look like an attempt to seize control; an interrupted warning can be made to sound like a completed instruction. These are meaning errors, not merely stylistic differences.
This guide explains a practical turn-taking translation method for learners and working translators. It distinguishes overlap from interruption, explores pauses, handovers, speaker labels, cut-off words and subtitle limitations, and shows how to preserve conversational agency without filling the page with unreadable notation. For the wider system, see Master Art of Translation, the guide to backchannels and acknowledgment signals and the guide to dialogue and subtext.
What turn-taking means and why it is part of meaning
A conversation is a jointly managed sequence of actions. Participants make statements, ask questions, tell stories, object, request help and provide responses. Each contribution occurs within an interactional arrangement: someone currently has the floor; another person may be waiting; a pause may invite entry; and a speaker may choose to keep or give up a turn. Turn-taking is the set of visible and audible practices that make this possible.
In a written translation, people notice lexical fidelity first. They ask whether the target words correspond to the source words. Yet imagine a witness saying, “I saw the car turn—” and an interviewer saying, “Before the bridge?” If the target text reads, “I saw the car turn before the bridge,” it does not merely smooth the dialogue. It transfers a geographical claim from the interviewer to the witness. The witness never completed that statement. The interviewer proposed the location. The resulting translation changes the evidence.
Ordinary conversations carry similar distinctions at lower stakes. A teenager begins, “I could try—” and a friend bursts in with, “You should audition!” The friend may be encouraging, but the encouragement is not the teenager’s decision. Writing “I could try auditioning” under the teenager’s name merges two people into one. The translator needs a model that records not only sentence meaning but the distribution of agency.
A quick working answer: map the floor before rewriting a line
Start by making a rough conversation map. Identify each speaker, the beginning and end of each turn, where speakers overlap, and the status of unfinished utterances. Then classify the interaction: an invited handover, a normal next turn, an interruption, a cooperative completion, a backchannel or a true simultaneous start. Finally choose the least obtrusive target-language presentation that keeps these relationships visible.
The method has two passes. During the listening or reading pass, annotate what happened without judging anyone’s intentions. During the translation pass, recreate the same interactional function using natural target-language words, punctuation, pauses or subtitle conventions. Keep the distinction between observation and inference. Overlapping voices can be heard; an intention to dominate may be uncertain unless tone, context and repeated behavior support it.
This also explains why punctuation alone is not a complete strategy. A dash might mean an interrupted word, a speaker’s self-correction, a sudden topic change or a typographical convention. If the same symbol is used for every event, readers cannot tell whether another person cut in or whether the speaker stopped voluntarily. Maintain a small, consistent transcription key appropriate to the audience.
Overlap is not always interruption
Two voices sounding at once do not necessarily mean one person is violating the other’s turn. A listener might murmur approval while a friend continues a story. A colleague may supply a missing word at exactly the moment the original speaker searches for it. Two people may start responding after the same question. All are overlap, but their conversational functions differ.
Consider a speaker explaining, “We met on the—” and a colleague saying, “Thursday,” while the original speaker replies, “Thursday, exactly.” A translation that makes the colleague sound aggressive loses the cooperative nature of the sequence. The same outward timing could mean something different if the colleague repeatedly finishes the speaker’s sentences despite signs of discomfort. The surrounding response tells the translator whether the completion was welcome.
A competitive interruption attempts to redirect or close down another person’s contribution. But the transcript alone may not establish intent. When the evidence is incomplete, preserve the overlap and the speaker boundaries rather than deciding that one participant was rude. If the original text does explicitly frame the event as an interruption, retain that characterization. Accuracy includes restraint: translate what the source permits you to conclude, not what a familiar stereotype suggests.
Where a turn actually ends
The end of a grammatical sentence is not always the end of a conversational turn. A speaker can complete a clause and then continue with a second part. Conversely, “Because of the traffic” may be a complete answer even though it looks like a sentence fragment. Conversational completion depends on syntax, intonation, context, gaze and the action being performed.
Suppose someone says, “The second option is cheaper, but—” and another person starts, “But the delivery date is later?” The first speaker’s “but” projects a qualification that was not yet given. The responder supplies a possible qualification; the translator must not collapse it into the original speaker’s position. By contrast, “The second option is cheaper.” followed by a pause and another participant’s “What about delivery?” is an ordinary next turn. The timing and projected grammar change the event.
Different languages signal completion differently. Some routinely place verbs or particles at the end of clauses, and some use sentence-final material that marks stance or politeness. A translation may need to move word order substantially to sound natural. That movement must not make a speaker finish a thought that the source leaves suspended. Preserve the practical incompleteness, even when a literal cut-off at the same word would be unnatural.
Interrupted words versus interrupted ideas
There are at least three levels of interruption. A word can break off halfway through; a clause can stop after a complete word; or a speaker can finish a sentence while another participant overlaps to capture the next turn. These produce different reading experiences and should not be marked indiscriminately.
When the speaker says something like “I wa— I wasn’t asking,” the first fragment may reveal a self-repair rather than another person’s interruption. When a second voice cuts in after “I wanted to explain—”, the unfinished explanation matters more than the last phoneme. In a target language with different word order, preserve the fact of a halted action, not an arbitrary collection of letters that happen to correspond to the English source.
In court transcripts, research interviews and spoken-language corpora, explicit transcription policy may require precise timing and partial-word detail. In ordinary narrative prose, a dash and a well-placed speaker change may be enough. The correct density is determined by the document’s purpose. Too little notation hides the event; too much specialized notation overwhelms general readers and turns a novel into a research transcript.
Cooperative completions and the ownership of a suggestion
Some conversational partners build an utterance together. A person searches for a place name and another supplies it. A learner hesitates over a technical term and a teacher offers a candidate. A family member begins a familiar phrase and another finishes it affectionately. In each case, the added words are attributable to the second participant even when the first adopts them.
For example: “The form goes to the, um—” “Admissions office?” “Yes, admissions.” The question mark after the offered phrase is important. The second speaker is making a candidate completion, not issuing a definitive correction. The first speaker ratifies it. A translation that reads “The form goes to the admissions office” deletes both the search and the confirmation. That may be acceptable in a heavily edited summary explicitly labeled as one, but it is not a faithful transcript of the exchange.
A target audience may not need every filled pause, particularly in accessible subtitles. Even then, an offered candidate and its acceptance may carry the central relationship. Preserve who proposed the answer and who adopted it. If a summarizing editor removes these details, the result should not be presented as a verbatim translation.
Interruptions that protect safety or urgency
Interruption is not automatically hostile. Someone may cut into a conversation because a vehicle is approaching, food is burning, a patient is deteriorating or a deadline is about to pass. The translation should make the urgency intelligible without rewriting the interruption as ordinary conversation.
Imagine: “Walk through the next doorway and—” “Stop! That step is broken.” The second voice changes the action at a critical moment. If a translation politely postpones “Stop” until after the directions, readers may imagine the warning came too late. In subtitles, the immediate imperative should remain prominent. In a novel, a line break and an interruption dash can show the shift of control. In an interpreting context, the interpreter should make the urgent warning audible rather than waiting for the first speaker to finish an irrelevant sentence.
The opposite risk is adding alarm. An ordinary timekeeping interruption, “Sorry, we need to close the meeting,” need not be dramatized as a command. The target must preserve the urgency level supported by voice, words and setting. Do not assume that all overlaps are rude or all warnings are emergencies.
Invited handovers: when the first speaker gives the floor away
Some turn changes are explicitly invited: “What do you think, Mira?”, “Would you explain the next point?”, or “You go first.” The invitation can be phrased as a question, imperative or social cue, but it allocates the next opportunity to speak. A translation that makes the invitation sound like a command may change the relationship between participants.
Role matters. A teacher may ask a pupil to elaborate; a moderator may pass to a panelist; a patient may ask a doctor a question; a chair may recognize someone formally. In some languages, pronouns, honorifics, names or verbal endings carry the expected respect. Choose forms that preserve relative authority and cordiality rather than copying the literal shape of an English expression.
A further complication is that being addressed does not guarantee compliance. If someone says, “Your turn,” and the addressee remains silent, the silence is part of the exchange. Do not supply an answer in the translated transcript merely to complete the apparent invitation. Translate the offered floor, the absence of response when relevant, and any subsequent change in speaker.
Pauses that invite entry and pauses that do not
A pause is not simply a gap between words. It can be a place where another person may enter, a deliberate withholding of information, a search for vocabulary, a moment of emotion, an opportunity for correction or an ordinary breath. Timing may be relevant, but interpretation requires context.
A person saying, “I need to tell you something,” followed by a long silence may be choosing how much to disclose. If the next person asks, “Are you all right?”, the silence helped prompt that question. Omitting it can make the response seem disconnected. In another dialogue, a pause before a name may be caused by memory search, not dread. Words such as “hesitantly” or “ominously” should not be added unless supported by the source.
When a project requires measured pauses, adopt a declared convention such as “[pause]” or “[three-second pause]” and do not invent measurements. For general prose, natural punctuation or a brief narrative tag can indicate a meaningful gap. The goal is never to reproduce every breath mechanically. It is to retain those gaps that alter inference, interpersonal stance or sequence.
Backchannels are not attempts to take over
A listener may say “mm-hm,” “right,” or “I see” while the principal speaker continues. These listener tokens often acknowledge hearing or understanding without claiming a full turn. Treating every backchannel as an interruption produces choppy translations and can falsely portray the listener as intrusive.
Compare “I was at the station—” “Mm-hm.” “—when the train stopped” with “I was at the station—” “No, listen, we have to leave.” The first exchange keeps the principal speaker on the floor. The second turns the action away. A target language may use different acknowledgment particles, gestures or short phrases, but the floor pattern should remain comparable.
Not every “yes” is an agreement either. In some settings it functions almost like a signal that someone is still listening. When summarizing, the translator should resist merging attention, understanding, commitment and consent. A reader deciding who agreed to a proposal needs the distinction even if it seems too fine for casual conversation.
Simultaneous starts and false attribution
After a question or silence, two people may begin at once. Neither has interrupted an established turn; both are claiming an available opening. If the translator always assigns the first complete sentence to one person and postpones the other, the reader may miss that both volunteered at the same time.
In a meeting, a chair says, “Who can take the early shift?” and two colleagues both begin, “I can—”. The shared start may indicate enthusiasm, competition or simple coincidence. If one then says, “You take it,” the allocation that follows depends on the original simultaneity. A good transcript can use overlapping lines, a timestamp, or an editorial note such as “[both begin speaking]” when house style allows.
For subtitles, simultaneous starts pose a spatial problem: two distinct voices cannot always fit comfortably in the same short display. The subtitler may prioritize the speech that changes the story or the required accessibility cue. Such prioritization should follow a consistent policy and should not hide the source of a decisive answer.
Turn-taking in subtitles and captions
Subtitles impose constraints unlike those of a book. Viewers have limited time to read, and text must stay coordinated with sound and image. Yet speaker change, interruption and simultaneity remain meaningful. The challenge is choosing which conversational features are essential under the format’s space and timing limits.
Speaker labels may be needed when the speaker is off-screen, not visually identifiable or speaking simultaneously with another person. A dash at the start of separate subtitle lines can sometimes identify a change, depending on the selected style guide. Do not assume there is one universal subtitle convention: broadcaster, platform and accessibility specifications differ. The translator should follow the project’s documented style and retain source-speaker attribution.
Take the sequence “Wait, I didn’t—” / “You did.” If the subtitle completes the first sentence as “I didn’t do it,” it invents a denial. If “You did” appears too early, it may make the second speaker seem to anticipate words never spoken. The translation, timing and visual segmentation must work together. Compression is permissible; changing speech acts or assigning unreached meanings is not.
Turn-taking in simultaneous and consecutive interpreting
An interpreter has to process meaning while managing a live conversational floor. In consecutive interpreting, speakers usually take longer turns and pause for rendition; the interpreter must communicate turn boundaries and may need a transparent way to request shorter segments. In simultaneous interpreting, speech streams overlap by design, and listeners rely on audio routing, speaker identification and context.
Interpreters should not become invisible decision-makers about who deserves the floor. If multiple people speak at once and accuracy becomes impossible, the safer professional response is to request orderly repetition where the setting permits. Guessing which claim belonged to which participant can be particularly harmful in healthcare, legal and high-stakes administrative discussions.
In a multilingual public meeting, one participant may direct a question to the interpreter rather than the intended speaker. The interpreter should distinguish a request for clarification of the language from a substantive reply. The precise protocol depends on setting and professional guidance, but the principle is stable: do not silently convert the interpreter’s own coordination into another person’s testimony or decision.
Meetings, classrooms and interviews
Institutional conversations carry explicit rules about turns. A moderator may recognize speakers; a teacher may nominate a pupil; an interviewer may ask and follow up; a committee chair may stop someone because allocated time has expired. Those constraints are not peripheral. They explain why one person can interrupt without appearing personally aggressive and why another person may protest.
Suppose a chair says, “Your two minutes are over,” before a delegate has finished. The intervention is a procedural cut-off. Translating it as “I disagree with you” would change the entire event. In a classroom, “Let her finish” can protect a student’s opportunity to explain. In an interview, “Did you actually see that?” may interrupt because the interviewer is testing whether an answer is direct evidence or inference.
The translator should record the rule being enforced, when it is evident, and preserve the tone of its enforcement. A procedural intervention can still be brusque; a correction can still be courteous. Neither institution nor job title alone determines politeness. Words, timing and surrounding reactions provide the evidence.
Novel dialogue and the rhythm of characterization
Literary translators face a particular temptation to make dialogue beautiful. An awkward turn can be made elegant; a barrage of short interruptions can be recast as balanced paragraphs; a hesitant character can suddenly become articulate. Even when each individual proposition survives, characterization changes.
A character who frequently talks over others may be anxious, exuberant, domineering or impatient. The translator need not decide which interpretation is right, but should preserve a recognizable pattern. A character who rarely claims the floor may be shy, strategic, deferential or simply given little opportunity. Flattening those asymmetries erases part of the narrative.
Literal punctuation is not always the answer. A language may have different conventions for quotation and dialogue; a dash in one edition may not function like a dash in another. The target should use established narrative conventions to generate a comparable experience. Readers should feel when a voice is cut off or unexpectedly enters without needing a transcript legend on every page.
Group chats, voice notes and remote calls
Digital conversations create their own turn-taking problems. In group messaging, two replies may be composed simultaneously but arrive in a different order. A user may reply to a specific earlier message rather than the most recent one. In a video call, latency can make speakers appear to interrupt each other, and muted microphones create turns no one hears.
A translation of a group-chat export should preserve reply threading, timestamps and sender identity when those fields affect meaning. If “Yes” is a reply to an earlier question rather than the current proposal, rearranging the messages can create false consent. In a support chat, an automated acknowledgement may interrupt a person’s explanation on screen but is not a human attempt to seize control.
Voice-note translations should distinguish gaps caused by recording from pauses chosen by a speaker when evidence permits. Do not infer malice from latency. In remote-meeting minutes, the aim may be a readable record of decisions rather than verbatim overlap; the document should clearly identify itself as minutes or a summary instead of silently claiming transcript-level precision.
High-stakes evidence: keep the provenance of every word
For research interviews, depositions, safeguarding records and complaints, the ownership of words matters acutely. A leading question can supply an assumption; an interrupted answer can leave an allegation unresolved; a witness’s correction can change a factual claim. Translation must not eliminate that provenance in the name of fluency.
Consider the sequence: “Did he threaten you with the tool?” “He—” “With the metal bar?” “I never said a metal bar.” A flattened sentence about a metal bar would misrepresent the witness’s knowledge. Accurate translation retains the interviewer’s proposal and the witness’s rejection. Editorial decisions about transcript notation should be documented and, where relevant, checked against the original recording.
This does not mean every ordinary subtitle or social conversation requires forensic transcription. Risk determines rigor. Where the text can affect rights or safety, use a suitable qualified translator, independent review and the source material. Do not treat a polished AI transcript as conclusive evidence without verifying audio, speaker identities and difficult overlaps.
How to mark overlap without making text unreadable
A usable translation system needs a small notation policy. At minimum, decide how to show a speaker change, an unfinished utterance, meaningful silence and simultaneous speech. Then apply those choices consistently. A dash for cut-off speech and a distinct tag for meaningful overlapping speech may be sufficient for a general audience, while research transcription may require aligned brackets or timestamps.
The policy should also state what you omit. Everyday hesitations and breathing are frequently left out of edited prose. This can be reasonable when no consequential meaning is lost. But deleting a failed answer, an interruption, a contested correction or the exact point where a witness stopped can change interpretation. The editor should distinguish harmless cleaning from substantive rewriting.
Reader comfort matters. A symbol means little unless readers can infer or learn its function. Use natural wording where a symbol is unnecessary, and define specialist notation briefly when it is. Never invent a false precision of timing. If no reliable audio is available, do not add “[interrupts sharply]” or “[overlap: 0.8 seconds]” as if you had measured it.
A worked example: the planning meeting
Imagine a team discussing a launch. Nia says, “We can release the page on Thursday if the translations—”. Tomas overlaps: “Friday. The legal review ends Thursday night.” Nia responds, “Right, Friday after legal.” The exchange contains a provisional proposal, an interruption with a constraint, and a revised plan ratified by the original speaker.
A weak translated summary might say, “Nia confirmed that the page would be released Friday after legal review.” As meeting minutes written after the decision, this could be a serviceable outcome sentence. As a translation of dialogue, it is inadequate. It erases Nia’s initial Thursday condition, Tomas’s intervention and the agreement process. A different weak version could make Tomas’s “Friday” sound like an unexplained command.
A faithful dialogue version keeps three turns separate. Nia’s incomplete condition remains incomplete; Tomas’s correction is clearly his; and Nia’s final agreement reflects a new shared outcome. If the project’s purpose is a decision log, label the resulting text as a summary. If the purpose is transcript fidelity, retain the turn order. Choose the genre before choosing how much dialogue to compress.
A worked example: the family conversation
A parent says, “I thought you were going to—”. A child answers immediately, “I did send it.” The parent continues, “—check the address first.” Here the child’s quick response is to a presumed accusation, while the parent was building toward a more specific expectation. Joining the parent’s fragments into a smooth sentence is possible in some literary translations, but the child’s premature defense must remain premature.
The emotional meaning depends on timing. A version that makes the parent finish and only then lets the child reply presents an ordinary response to a completed instruction. The source presents anticipation, perhaps frustration, and a misunderstanding that can be resolved. Preserving the broken turn shows why the characters talk past each other.
Other languages may require a different clause structure to preserve the split. Recreate the interruption at the point where the child could plausibly misinfer the parent’s direction, rather than cutting at the same English word. This is functional fidelity: the target reader should undergo the same partial-information moment.
A worked example: the service counter
A visitor asks, “Can I pay the fee after I—”. The clerk interrupts: “Payment must be completed before the appointment.” The visitor asks, “Even if the portal is down?” The clerk’s response is a policy statement delivered before the condition was disclosed; the visitor’s follow-up reveals that the interruption prevented a complete question.
A translation that turns the exchange into “Can I pay after my appointment? No, payment must be completed before the appointment” invents both the visitor’s intended condition and a neat response. It could lead an administrator to believe a specific question had been answered. Better to preserve the incomplete question and the follow-up.
In customer-support training, this is also a useful diagnostic example. The translation can reveal a service failure: the clerk responded to an assumption instead of listening to the full problem. Whether the clerk is rude or merely rushed is for context to determine. The translator’s job is to preserve the evidence on which readers can make that judgment.
Translating across languages with different conversational structures
Not all languages mark conversational continuity with the same pronouns, particles or sentence shapes. Some allow omitted subjects to remain clear from context. Some use sentence-final forms that signal respect or invite response. Some rely heavily on pitch contour, while others encode important information in morphology. A translator therefore cannot simply copy English punctuation onto a structurally different target language and expect identical effects.
Start with a functional analysis of the source: what did each contribution accomplish, and when did it become available to the next speaker? Then build natural target-language phrasing that permits the same next action. If the second speaker interrupts before a crucial qualifier, preserve a plausible cut-off before that qualifier becomes available. If a speaker trails off deliberately, avoid presenting it as another person’s abrupt intervention.
Language-pair expertise matters. A gesture that signals turn completion in one community may not have a direct written equivalent in another. Do not invent universal rules for every language. Consult authentic comparable dialogue, professional style guides and knowledgeable speakers, and test the target scene aloud to see whether the same interpersonal relationship emerges.
Common errors and how to detect them
The most serious error is speaker merging: words spoken or suggested by one person are silently assigned to another. Next is false completion, in which a translated sentence supplies a claim the speaker never finished. A third is false aggression, where cooperative overlap is presented as an attack. The reverse—turning a forceful cut-off into a gentle contribution—is equally distortive.
Another class of mistake is temporal distortion. A reaction appears before the event that motivated it, or a correction is placed after the wrong claim. In subtitles, misaligned displays can create that effect even when the words are correct. In chat transcripts, lost reply links can reassign a “yes” or “no.” In published interviews, removing self-corrections can conceal how a witness or expert changed an answer.
Audit these errors by asking, line by line, whether the same person says the same thing at the same stage of knowledge. Then listen or read without the source: does the target scene remain coherent, and does it avoid implying unsupported agreement, consent, dominance or certainty? A faithful translation is not simply one that sounds plausible; it has to be plausible for the right reasons.
A practical review workflow for translators
First establish document purpose: verbatim transcript, readable interview, literary dialogue, subtitles, edited minutes or educational paraphrase. Second identify speaker boundaries. Third annotate the small number of overlaps and interruptions that materially affect agency, response, tone or evidence. Fourth translate the actual speech actions, not just isolated words.
Fifth compare the translation against the recording or written source at each marked boundary. Check whose words they are, whether the preceding turn was complete, and whether a cut-off changed a later reaction. Sixth review the target without the source for readability. Finally record editorial conventions where the stakes justify it, especially if multiple translators, transcribers and reviewers share the project.
Do not measure quality by the number of dashes retained. A translation with fewer symbols may be more accurate if its layout makes the exchange clearer. Conversely, an elegant scene with no visible interruptions can be seriously misleading. The best method is purpose-driven: preserve significant turn behavior while choosing the simplest representation that the target audience can actually use.
Questions readers often ask
Should I translate every “um”, “uh” and incomplete syllable?
Not necessarily. A verbatim research transcript may require detailed disfluency records, while an edited article may omit ordinary fillers that do not change meaning. Preserve hesitation when it reveals uncertainty, delayed consent, word search, correction or a consequential interruption. Apply a stated editorial policy rather than deciding opportunistically which speaker should sound more polished.
Does a dash always mean interruption?
No. It may indicate a sudden ending, cut-off word, self-repair or narrative break. The surrounding speaker labels and typography determine what readers infer. If ambiguity would affect a significant claim, choose clearer notation or a short contextual cue. Do not assume the punctuation convention of one language or publisher applies everywhere.
How do I distinguish cooperation from rudeness?
Observe what happens before and after the overlap. Does the first speaker accept the supplied word? Does the second speaker yield quickly or keep the floor? Is the interruption a response to danger? Does someone object to being interrupted? Tone and repetition help, but motives can remain uncertain. Translate the event faithfully without inventing a psychological diagnosis.
Can I rearrange lines for readability?
In a clearly labeled summary or edited minutes, chronology may be compressed according to the document’s function. In a translation presented as dialogue or testimony, rearrangement can change causal relationships and attribution. If rearrangement is required by subtitle constraints, keep the viewer’s understanding aligned with the actual exchange as closely as the format allows.
What should I do when two people are genuinely unintelligible?
Do not guess. Mark the undecipherable span according to the project’s transcript policy, consult audio where available and request clarification when possible. In high-stakes work, record uncertainty openly and arrange qualified review. An honest gap is preferable to a confident statement placed in the wrong person’s mouth.
Final checklist: the floor is part of the message
Before releasing a conversational translation, confirm that every consequential line belongs to the correct speaker; cut-off claims remain incomplete; overlap is not automatically treated as hostility; cooperative completions are not merged into the wrong voice; meaningful pauses remain visible; urgency and procedural interruptions retain their force; backchannels are not mistaken for agreement; and subtitle or transcript conventions are consistent.
Then check the most important question: would a target reader understand who controlled the exchange, who proposed each idea and when a participant actually committed to a statement? If yes, the translation has preserved something deeper than word choice. It has carried the social mechanics of the conversation across languages.
Continue with how to translate opinions and viewpoints, backchannels and acknowledgment signals, the wider translation system and the correct-translation series index.