A grounding criterion is a task-sensitive standard for when people in an interaction have established sufficient mutual evidence that an intended meaning has been understood. Two students preparing a poster may need to identify the same current file before dividing the work. One says ‘Use version four’; the other replies, ‘The copy with the corrected diagram, right?’ Once the reference is clarified and mutually accepted, they may be able to proceed. They do not need to demonstrate identical knowledge about every part of the project.
This idea turns vocabulary into a technology for reliable coordination. Words such as received, understood, confirmed, approved and completed describe different achievements. A grounding criterion helps us decide what evidence of understanding is enough for the next action instead of assuming that a nod, a read receipt or an ‘okay’ guarantees success. It also prevents endless checking when the task needs only one clear answer.
The concept comes from Herbert H. Clark and Susan E. Brennan’s research on grounding in communication. This guide explains it through original parent–student–tutor conversations, school projects, reading, Mathematics and digital messages. It distinguishes shared understanding from factual accuracy, legitimate permission and successful task completion. The right level of grounding depends on purpose, consequences and communication medium—not a universal script that every person must follow.
Editorial scope: All school and family cases are original hypothetical examples. The grounding criterion is an interactional concept, not a universal test of truth, consent, personal ability or academic performance.
Contents: understanding enough for the task
Open 57 chapters and cases
- The simple example: which worksheet?
- Clark and Brennan’s account
- Grounding is a process; the criterion is the target
- The relevant unit is ‘enough for current purposes’
- Mutual belief is not mind-reading
- Hearing is not the same as understanding
- Understanding is not the same as agreement
- Understanding is not the same as truth
- Understanding is not permission
- Understanding is not task completion
- What counts as evidence of understanding?
- An appropriate acknowledgement
- An insufficient acknowledgement
- Repeating the whole message may be unnecessary
- The principle of least collaborative effort
- Why a small clarification can save rework
- When a brief instruction is enough
- Grounding in face-to-face conversation
- Grounding by pointing
- Grounding in telephone calls
- Grounding in text messages
- Read receipts are not understanding receipts
- Grounding in video meetings
- The wrong referent can remain hidden
- Grounding and repair are closely linked
- Grounding and turn taking are different
- Grounding and adjacency pairs
- A simple task needs a small criterion
- A complex task needs a stronger criterion
- A criterion must not silently change
- Grounding a Mathematics variable
- Grounding a Science measurement
- Grounding an English paragraph instruction
- Grounding a shared project status
- The false certainty of a friendly ‘okay’
- Permission to ask for clarification
- Different readers have different prior information
- A group can have uneven common understanding
- The role of shared artefacts
- When to stop clarifying
- Grounding can fail even after agreement
- Grounding is not a consent framework
- Grounding and privacy
- Worked case: a parent–student message
- Worked case: a tutor explains an equation
- Worked case: science-group measurement
- Worked case: a digital script handover
- What would a stronger confirmation add?
- What would an unnecessary confirmation look like?
- A practical four-question grounding check
- The total effort of speaker and listener
- A five-line status record
- Diagnostic: three different communication failures
- Diagnostic: apparently successful but actually insufficient
- Ten questions with short answers
- Frequently asked questions
- Source notes and the next practical step
1. The simple example: which worksheet?
A learner hears ‘Complete the later worksheet’ and finds two documents with similar titles. One is a practice sheet; the other contains corrected questions. The learner asks which one is intended.
The teacher replies, ‘Use Worksheet C, revision two, Questions 1 to 5.’ The learner repeats the target and begins the work. The exchange has supplied stronger evidence that both parties refer to the same assignment.
The criterion is local: enough shared understanding to choose the right worksheet. It does not prove the student knows how to answer every question on it.
2. Clark and Brennan’s account
Herbert H. Clark and Susan E. Brennan developed the grounding account in their 1991 chapter Grounding in Communication. A useful Stanford research overview by Clark explains the idea of establishing mutual belief that an utterance has been understood sufficiently for the current purpose.
Clark’s chapter Grounding in Using Language develops this account as collaborative coordination across levels of joint activity.
The theory is not an instruction that every listener must repeat every sentence. It asks what work the participants need to do to establish an adequate basis for continuing together.
3. Grounding is a process; the criterion is the target
Grounding describes participants’ efforts to establish and maintain shared understanding. The grounding criterion describes how much of that understanding is sufficient for the purpose at hand.
For a casual greeting, no elaborate confirmation is usually necessary. For a consequential change to a school submission, an exact identifier, date or status may need checking.
The distinction prevents the word grounded from becoming a vague compliment. Ask which understanding was established, by what evidence and for which action.
4. The relevant unit is ‘enough for current purposes’
Suppose two friends are choosing a book for a brief reading activity. They may only need to agree on the title. A librarian performing a formal loan transaction may also need the correct physical copy and borrower record.
Both interactions concern a book, but their grounding criteria differ because the tasks differ. Identifying the title is not always sufficient for identifying the item in an inventory.
Always name the next action before deciding how much mutual understanding is enough. Without a purpose, ‘everyone understands’ is too broad to evaluate.
5. Mutual belief is not mind-reading
Participants cannot directly inspect one another’s minds. They rely on responses, visible actions, corrections, questions and shared records as evidence that communication has worked.
A listener might nod while still misunderstanding a technical term. Another might ask a precise question yet understand the wider task very well. These behaviours need contextual interpretation.
The research model concerns the interactional establishment of sufficient mutual belief, not an infallible test of internal knowledge.
6. Hearing is not the same as understanding
A teacher states, ‘Use the denominator from the second fraction.’ A pupil hears every word but does not know what denominator means. The acoustic message arrived; the mathematical instruction remains unclear.
Repeating the same sentence louder may not solve the concept problem. A labelled example can identify the denominator and connect it to the intended operation.
The grounding criterion is not met simply because the pupil says ‘I heard you’. The task requires understanding the part of the instruction needed for the next calculation.
7. Understanding is not the same as agreement
Two students understand a proposed answer and disagree about whether it is correct. They may have successfully grounded the proposition while still debating its truth.
A learner can say ‘I understand your claim, but I think the evidence is insufficient.’ That is not a failure of communication; it makes the disagreement inspectable.
Do not insist on agreement as a measure of understanding. Shared reference and acceptance of a conclusion are different achievements.
8. Understanding is not the same as truth
A group could mutually understand an outdated deadline and plan around it. Their common interpretation would not make the date accurate.
An authoritative current notice may be needed before acting. Grounding establishes what the participants take one another to mean; verification concerns whether the content matches reality.
The Common Ground guide explores why shared assumptions can be useful, mistaken or provisional.
9. Understanding is not permission
A student fully understands how a locked equipment cabinet opens. That knowledge does not grant authority to access it.
An appropriate school procedure may require permission, supervision or a designated staff member. Those are separate conditions from understanding the instruction.
Teach pupils to distinguish ‘I know what you mean’ from ‘I am allowed to do it’. Grounding should not be treated as a substitute for legitimate rules.
10. Understanding is not task completion
A learner correctly repeats ‘Submit Worksheet B through the class portal’. The student has shown understanding of an instruction; the worksheet has not yet been submitted.
A later receipt may establish successful transfer of the file, while another review may establish that its contents meet the task requirements.
Use separate words for understood, attempted, received, reviewed and completed. Collapsing them can create the very workflow errors the grounding process was meant to prevent.
11. What counts as evidence of understanding?
A relevant reply, correct paraphrase, appropriate clarification or successful reference to the intended object may support the participants’ belief that they understood one another.
The value of the evidence depends on the task. Echoing a question can show hearing without demonstrating subject understanding; completing a new example may show deeper use of a concept.
Ask what misunderstanding would matter for the next action and choose a proportionate check.
12. An appropriate acknowledgement
A teacher says, ‘The meeting begins at four in Room 2.’ A parent replies, ‘Four o’clock, Room 2—thank you.’ The repeated information provides evidence of receiving the practical details.
That is sufficient for many ordinary scheduling exchanges, assuming the notice itself is correct. It does not establish the parent’s route, transport or later attendance.
A grounding check should protect the specific action, not pretend to settle every future uncertainty.
13. An insufficient acknowledgement
A tutor asks a pupil to use the corrected version of a diagram. The pupil replies ‘Okay’ but begins editing the older file. The visible action shows that the earlier acknowledgement did not establish the intended reference.
The group can stop, identify the version and correct the handover. The pupil need not be described as careless merely because the first message was ambiguous.
A failed action can supply evidence that the grounding criterion has not been met or that another prerequisite is missing.
14. Repeating the whole message may be unnecessary
If a speaker asks for a pencil and the listener immediately hands over the intended pencil, the action itself can provide sufficient evidence of understanding for that small request.
Requiring the listener to recite the complete request before acting could add effort without improving the outcome. In a higher-stakes procedure, explicit verification may still be required.
Match the grounding method to the cost of potential misunderstanding and the information available through the medium.
15. The principle of least collaborative effort
Clark and Brennan discuss a principle of least collaborative effort: participants coordinate contributions while attending to the effort required from both speaker and listener over the exchange.
A very short ambiguous instruction may appear efficient for the speaker but create several clarification messages for the listener. A longer initial instruction with a clear filename can reduce joint effort.
The principle is not a formula saying all conversations should be brief. Efficiency concerns the total cost of achieving the needed understanding, including possible repair.
16. Why a small clarification can save rework
A project message says ‘Use the final version’. Four files contain the word final. A student spends time comparing them and still chooses the wrong one.
One extra sentence—’Use version five, approved after Tuesday’s source check’—could have made the intended file easier to identify if accurate.
Audience-aware detail can reduce downstream correction. The relevant test is the actual information need, not a blanket rule against longer messages.
17. When a brief instruction is enough
A class has just placed one labelled red notebook on the shared desk. The teacher asks, ‘Pass me the red one’. The shared context may make the referent obvious.
Repeating a long file identifier in this setting could be unnecessary. Common ground and the current physical situation support a shorter expression.
The next question is whether the participants actually identified the same object, not whether the message contained a predetermined minimum number of words.
18. Grounding in face-to-face conversation
Face-to-face participants can sometimes use speech, pointing, gaze, shared objects and immediate responses to coordinate reference.
The available cues do not guarantee identical understanding. Some forms of communication are inaccessible or unsuitable for particular people; the interaction must allow an appropriate route to clarification.
Ask which cues were available and which part of the intended message they actually helped establish.
19. Grounding by pointing
A teacher points to one of two graphs and says, ‘Compare this slope’. If the gesture and sightline identify one graph unambiguously, the learner may understand the referent.
If the learner is reading a transcript later, the pointing cue has disappeared. A durable instruction should name the graph or provide another accessible reference.
Grounding can depend on a shared situation that does not travel automatically into a written document.
20. Grounding in telephone calls
Telephone speakers cannot ordinarily rely on visible pointing to the same objects. They may specify names, repeat numbers and use explicit verbal acknowledgements.
A momentary audio failure can interrupt the grounding process even when both people know the intended task. A brief request to repeat the affected detail can restore it.
Do not treat communication-medium limits as evidence that a listener lacks intelligence or attention.
21. Grounding in text messages
Text retains visible words but may allow delays and interleaved topics. A reply ‘Yes’ after several active questions can have an unclear target.
Quoting the relevant question or naming the exact file version can help the participants identify which issue the response settles.
The written medium supports later inspection, but a message being delivered is not proof that it was read or interpreted correctly.
22. Read receipts are not understanding receipts
A platform shows Delivered or Read beside a message. That status can describe a technical event according to the platform, not a full demonstration that the recipient understands or agrees with the contents.
For a school project requiring approval, an explicit authorised decision may be necessary. A read indicator cannot substitute for it.
Name the different evidence sources: delivery status, recipient reply, task performance and formal approval.
23. Grounding in video meetings
Participants may have shared screens and visual reactions while also experiencing audio delay or difficulty identifying which document is current.
A speaker can establish context with a title, page number and version label. Another participant can ask for clarification if the display differs.
Choose grounding checks that do not rely solely on unverified facial expressions or a particular technology feature.
24. The wrong referent can remain hidden
Two people use the phrase ‘the diagram’ while viewing different versions. They both respond confidently, so the mismatch remains unnoticed.
Only a later action—such as editing different files—reveals the discrepancy. A short version check at the beginning may prevent the error.
The Reference Chain guide explains how names, descriptions and pronouns point to the intended entity.
25. Grounding and repair are closely linked
When a listener asks ‘Page fourteen or forty?’, they signal that part of the original contribution is not yet sufficiently understood. The speaker supplies the missing distinction.
After the clarification, the listener may answer the original question or perform the requested action. The repair helps participants approach their grounding criterion.
The Conversational Repair guide focuses on this sequence of identifying and resolving trouble.
26. Grounding and turn taking are different
Turn-taking cues help people anticipate when a speaking opportunity becomes available. Grounding concerns establishing that contributions have been understood sufficiently for the joint activity.
A student can take a turn correctly yet answer a different question. Another can understand the task but need time before responding.
See Turn-Taking Cues for the organisation of speaking opportunities.
27. Grounding and adjacency pairs
A question projects an answer, but the first answer may ask for clarification. That inserted sequence can help identify what was meant before the original question is answered.
An adjacency pair describes the social organisation of related actions, while grounding describes the collaborative work of achieving sufficient shared understanding.
The Adjacency Pair guide details that distinction.
28. A simple task needs a small criterion
An invented library game asks pupils to place one picture card beside the matching word. The immediate grounding criterion may be that each pupil knows which picture and word the instruction refers to.
An appropriate check can be to ask the pupil to identify the pair. Requiring a long formal definition of every related concept would be disproportionate to the task.
The criterion should be sufficient for the action, without replacing later opportunities for deeper learning.
29. A complex task needs a stronger criterion
An older student must interpret a graph, select the appropriate quantity and justify a conclusion. A nod after the teacher’s explanation is weak evidence that the required reasoning can be performed.
A more suitable grounding check may ask the learner to explain what each axis means and demonstrate one relevant calculation with a new data point.
The stronger evidence matches the task’s complexity. It is not a judgement that the learner is less cooperative.
30. A criterion must not silently change
A teacher begins by asking pupils merely to identify a diagram, then later expects them to explain its causal mechanism. Both may be valid learning goals, but the grounding criteria differ.
If the criteria change, state the new target clearly. A pupil who met the earlier identification task should not be retrospectively described as failing it because the later explanation is incomplete.
Accurate vocabulary makes progress and new challenges visible as separate stages.
31. Grounding a Mathematics variable
An invented task uses x to represent the number of stickers per packet. One learner thinks x means the total number of stickers.
Before solving, ask ‘What does x stand for in this problem?’ If the pupils provide the intended unit and role, the variable’s referent is better grounded.
The next equation can then be checked independently. Shared meaning of x is a prerequisite for a valid collaborative solution, not proof the arithmetic will be correct.
32. Grounding a Science measurement
Two pupils agree to measure ‘plant height’ but one begins at the pot base and another at the soil surface. The phrase is shared, yet the measurement convention is not.
An explicit labelled diagram and a worked example can establish the reference points for their records. Then the values become more meaningfully comparable.
The Operational Definition guide shows how a term can be linked to a specific observable procedure.
33. Grounding an English paragraph instruction
A tutor asks for a ‘supported inference’. The pupil interprets supported as ‘a confident answer’ while the tutor means ‘a conclusion connected to relevant textual evidence’.
A brief contrast between an unsupported guess and a justified interpretation can make the target clearer. The pupil then identifies an appropriate detail in a new sentence.
The grounding criterion concerns understanding what the task asks for. Further practice may still be needed to perform it independently.
34. Grounding a shared project status
A writer says a document is ‘done’. The editor hears ‘approved for publication’, while the writer means ‘first draft finished’.
The group can separate Draft Complete, Review Pending and Approved. An exact status plus current version reduces the scope for misunderstanding.
Grounding the status word protects the workflow. The State Transition guide explains how the documented state can change after a review.
35. The false certainty of a friendly ‘okay’
A group member replies ‘Okay’ to an assignment request. The response might acknowledge hearing, signal willingness or merely indicate that the message was noticed.
Where responsibility for a consequential task matters, ask for the specific accepted action: ‘Can you review Questions 1–4 by Friday?’ A reply naming those conditions provides stronger evidence.
Do not require elaborate confirmations for every small social exchange. Use the check when the ambiguity would alter what happens next.
36. Permission to ask for clarification
A student may hesitate to ask a question because others seem ready to proceed. The missing understanding can then become a hidden source of errors.
A tutor can normalise focused questions such as ‘Which file do you mean?’ or ‘What does this symbol represent?’. These are legitimate parts of coordinating a task.
Do not equate a request for clarification with inability or lack of preparation. It can be the most efficient way to meet the relevant grounding criterion.
37. Different readers have different prior information
An experienced class member recognises ‘Revision C’ instantly; a newcomer has not seen earlier versions. The speaker may need to identify the subject, file and current location for the newcomer.
That difference calls for audience-sensitive explanation rather than a broad judgement about competence.
The Audience Design guide explains how to supply detail without altering facts.
38. A group can have uneven common understanding
In a three-person project, two participants may know which model is under discussion while the third has missed the reference. A group-wide ‘we all understand’ may therefore be premature.
Ask the participant who needs the information to identify the object or question, with an accessible way to respond. Do not assume one confident reply represents the entire group.
Grounding can be sensitive to the actual set of participants who must coordinate the next action.
39. The role of shared artefacts
A clearly labelled worksheet, map or current version record can make reference easier across speakers. The artefact may provide information that nobody needs to recite from memory.
But the document might be outdated or interpreted differently. It should remain current, accessible and appropriately connected to the task.
The concept of a Boundary Object concerns how shared representations can support groups with different perspectives.
40. When to stop clarifying
Two learners have identified the same correct worksheet and the first question to attempt. Repeating the entire instruction five more times may add effort without resolving a remaining ambiguity.
A sufficient grounding criterion gives a reason to move forward. If a new inconsistency appears later, the interaction can reopen the relevant issue.
The Stopping Rule guide helps distinguish justified completion from premature abandonment.
41. Grounding can fail even after agreement
A teacher and pupil agree that a source paragraph supports a claim. Later, they discover the paragraph came from a different edition with different wording.
The earlier agreement accurately describes what the pair believed, but it does not settle whether the source actually supported the intended conclusion.
The responsible response is to revise the shared understanding in light of the new evidence rather than conceal the error.
42. Grounding is not a consent framework
Mutual understanding about what a proposal means does not automatically create valid consent, permission or authorisation for a consequential action.
Those activities may have legal, ethical or institutional conditions beyond conversational grounding. A clear response from the appropriate person remains necessary under the real rules.
Teach the distinction between ‘I understand what is requested’ and ‘I agree to it’. Clear understanding supports better decisions without replacing the decision itself.
43. Grounding and privacy
A teacher does not need every detail of a pupil’s private life to explain a grammar correction. The grounding criterion should be tied to the task’s legitimate information need.
An instruction that demands unnecessary disclosure can be inappropriate even if it would make the listener’s intentions easier to infer.
Appropriate communication uses enough information, not every possible piece of information. Privacy and dignity remain independent considerations.
44. Worked case: a parent–student message
A fictional parent receives ‘The practice is tomorrow’ from the student. The parent does not know whether this means the scheduled lesson, an extra revision task or a club meeting.
An appropriate question is ‘Which activity, and what time?’ The student identifies the lesson and its verified schedule. The parent repeats the actionable details.
The exchange may now be sufficiently grounded for travel planning, provided the schedule is current. It does not guarantee attendance or that every school requirement has been completed.
45. Worked case: a tutor explains an equation
A tutor says ‘Apply the same operation to both sides’. A learner asks ‘Do you mean subtract four from each side in this equation?’ The tutor confirms and asks why that preserves equality.
The clarification establishes which operation is intended. Explaining the equivalence demonstrates a different layer of mathematical understanding.
This separates grounding an instruction from mastering the concept and helps the tutor target the next learning action accurately.
46. Worked case: science-group measurement
Two students plan to record a model’s width. One measures from its outer edge; the other means the internal opening.
They identify the target quantity and draw a labelled measurement route. Their agreement makes the later readings more comparable under the stated procedure.
The criterion is not a particular number of acknowledgements but sufficiently shared knowledge of what is being measured and how.
47. Worked case: a digital script handover
An invented video team has three files: Draft 2, Revised 3 and Approved 3. The editor hears ‘Use version three’ and could choose either of the latter two.
The writer clarifies ‘Use Approved 3, with the checked narration’. The editor reads the filename back and confirms the recorded approval status.
The grounding criterion is met for selecting a file when the relevant identifier and authority record align. The actual export and upload remain separate workflow actions.
48. What would a stronger confirmation add?
If the cost of choosing the wrong file is small and easily reversible, a quick acknowledgement may be appropriate. If a publication will be difficult to correct after release, the team may require a verified version link and explicit approval.
Additional confirmation should protect a meaningful risk rather than grow merely because a process looks more formal.
Ask which failure the next check prevents and whether the source of that check is reliable.
49. What would an unnecessary confirmation look like?
Imagine two pupils exchanging one clearly visible ruler at a shared desk. They agree which ruler is intended and complete the handover.
Requiring a multi-step written approval process would likely add effort without serving the small task. A richer process may make sense for valuable or restricted equipment, but that is a different situation.
The contrast shows why grounding criteria are purpose-sensitive rather than universal.
50. A practical four-question grounding check
First: what must the other person understand for the immediate task? Second: which word, object or condition might be ambiguous? Third: what response or observable action would give useful evidence? Fourth: what uncertainty remains after the check?
Use the answers to formulate one focused clarification, not a barrage of unrelated questions. If the next action requires a source or authorisation, identify that as a separate need.
This worksheet is an educational aid inspired by the theory; it is not a validated psychological test of mutual understanding.
51. The total effort of speaker and listener
A speaker can make a message short by saying ‘Do the usual thing’. The listener might then need several questions and two false starts, increasing the group’s total effort.
An initially longer instruction that identifies the action, object and deadline may be more efficient overall. On the other hand, spelling out details already unambiguously shared can create needless work.
Clark and Brennan’s least-collaborative-effort perspective invites us to consider the combined work of achieving sufficient understanding, not only how quickly the first speaker finished.
52. A five-line status record
For an invented group task, record Item, Current Version, Intended Action, Confirmation Evidence and Unresolved Issue. A row might read: Poster P-04; version two; review source labels; reviewer identified the correct file; diagram approval still pending.
The status record makes it easier to distinguish understanding of the request from completion of the requested review.
Keep records proportionate and avoid unnecessary private data. The purpose is practical coordination, not surveillance.
53. Diagnostic: three different communication failures
One student hears the wrong number: a hearing or recognition problem. Another hears the right words but misunderstands the task term: a semantic or conceptual problem. A third understands the request but lacks access to the file: a resource problem.
The first may need repetition, the second a clear example and the third access repair. Repeating the identical sentence is not a universal solution.
Name the type of missing condition before deciding what the next speaker should do.
54. Diagnostic: apparently successful but actually insufficient
Case A: a pupil says ‘yes’ after a long explanation but cannot identify the task; understanding is not established. Case B: the pupil chooses the correct page yet cannot solve the problem; the page reference may be grounded, but subject understanding remains open.
Case C: a reader identifies the current file but has no permission to publish; grounding does not provide authorisation. Case D: everyone agrees on an outdated date; shared understanding does not establish a current fact.
Each diagnosis protects a separate boundary that an undifferentiated word such as ‘confirmed’ could hide.
55. Ten questions with short answers
1. Is grounding simply repeating a sentence? No. 2. Is the criterion the same for every task? No. 3. Does a nod prove complete understanding? No. 4. Can an action show useful understanding? Sometimes, for the relevant task.
5. Is understanding the same as agreeing? No. 6. Is mutual understanding proof that a claim is true? No. 7. Can a short clarification reduce overall effort? Yes. 8. Can an excessive check waste effort? Yes.
9. Do communication media affect grounding? Yes. 10. What is the central test? Have participants established sufficient mutual evidence of understanding for their current joint purpose?
56. Frequently asked questions
Is grounding criterion just another name for common ground? No. Common ground concerns the information participants treat as shared; the grounding criterion concerns the level of mutual understanding needed for a particular contribution and purpose.
Does the theory mean everyone must explain every instruction back? No. Appropriate evidence varies with task and medium. Can a speaker be wrong yet understood? Yes. Can someone understand without accepting a request? Yes.
Why include this in a vocabulary series? It helps people distinguish receipt, understanding, verification, consent and completed action while deciding what clarification is genuinely necessary.
57. Source notes and the next practical step
Primary sources: Herbert H. Clark’s Stanford research overview, Clark and Susan E. Brennan’s Grounding in Communication (1991), and Clark’s Grounding chapter in Using Language. They distinguish grounding as a collaborative process from the sufficiency standard for present purposes.
The interactions and outcomes described in this article are fictional teaching examples, not reported field data or experimentally validated checks. The source theory is more detailed than the worksheets offered here.
Pick an instruction two people need to act on today. Identify one fact that must be shared, one reasonable way of confirming it and one outcome not yet established by that confirmation. Stop clarifying when the task-relevant meaning is secure, then perform and verify the actual next action. That is vocabulary working as coordination technology.