How forensic linguistic analysis can reveal the assumptions built into legal and institutional questioning

In legal and administrative proceedings, disputed answers receive considerable attention. They are transcribed, quoted, segmented, compared, and interpreted. A hesitation may be treated as uncertainty. A reformulation may become a contradiction. An unusual word may raise questions about knowledge or credibility.

The question that produced the answer, however, often receives far less scrutiny.

It is usually treated as a neutral instrument: a procedural device designed to obtain information already residing in the respondent’s mind. The answer is regarded as evidence; the question merely provides the occasion for its production.

But answers are not independent linguistic objects. They are turns in an interaction. They respond to something, resist something, accept something, correct something, or attempt to change the direction of the exchange. To interpret them adequately, it is necessary to examine not only what was said, but what the speaker was being asked to do.

Two cases illustrate why this distinction matters.

A request produced during an interrogation

During a police interrogation in Louisiana, Warren Demesme produced the statement that would later become famous:

“If y’all, this is how I feel, if y’all think I did it, I know that I didn’t do it so why don’t you just give me a lawyer dog cause this is not what’s up.”

The Louisiana courts considered whether those words amounted to a clear and unequivocal request for counsel. Justice Scott Crichton described the expression as an “ambiguous and equivocal reference to a ‘lawyer dog.’” The interrogation continued.

Much of the public discussion concentrated on the final two words. Was dog part of the noun phrase—a nonexistent lawyer dog—or was dawg a colloquial form of address: give me a lawyer, dawg?

That question is linguistically relevant, but it begins too late.

Demesme’s statement was not produced as an isolated sentence. It appeared within an interrogation, as part of an unfolding exchange between police officers and a person denying an accusation. Even within the fragment preserved in the judicial decision, the turn displays several simultaneous actions.

He responds to what he understands as the officers’ suspicion:

“If y’all think I did it…”

He asserts his own epistemic position:

“I know that I didn’t do it…”

He asks for a lawyer:

“Why don’t you just give me a lawyer…”

And he rejects the interaction as it is currently proceeding:

“Cause this is not what’s up.”

The utterance may therefore be heard not merely as a sentence containing the word lawyer, but as an attempt to alter the participation framework of the interrogation. Demesme is denying the allegation, resisting the direction of the questioning, and introducing counsel as the condition under which the exchange should continue.

The judicial analysis, however, primarily examined whether the wording satisfied an institutional threshold of clarity. The response remained visible; the interaction that made it relevant largely disappeared.

The transcription and legal treatment of this expression are discussed in greater detail in “Give Me a Lawyer, Dog”: A Case of Linguistic Ambiguity.

A lexical test produced during a refugee hearing

A different interaction occurred during the Canadian refugee hearing of Nastran Yeganeh, an Iranian midwife who sought protection after performing clandestine hymenoplasty procedures.

During the hearing, the adjudicator asked Yeganeh to provide the “technical, anatomical term” for the membrane involved in the procedure. She answered in Farsi through an interpreter. The expressions that entered the English-language record were virginity curtain and virginity tissue.

Those expressions were treated as evidence against her.

The underlying reasoning appeared straightforward: a professionally experienced midwife should know the anatomical term hymen. If she could not produce it, her claimed experience—and therefore her account—could be doubted.

Yet, interactionally, the exchange was not simply a request for information.

The adjudicator’s question introduced an unannounced lexical test. It presumed that authentic professional knowledge should be demonstrable through the production of a particular terminological equivalent. It also presumed that this equivalent should be available on demand across languages, through an interpreter, and under the conditions of a refugee hearing.

The question therefore contained a prior model of the credible speaker:

A genuine midwife would know this term.
A genuine midwife would express that knowledge through the expected lexical label.
Failure to produce that label would indicate failure of knowledge.

Yeganeh was not merely being asked what anatomical structure she meant. She was being required to perform her professional authenticity in the vocabulary anticipated by the adjudicator.

The Federal Court later set aside the decision. Her answers had demonstrated knowledge of the anatomy through function and description, even though the expected English term had not appeared in the record.

Whether or not one can establish that the adjudicator had psychologically reached a conclusion before posing the question, the question functioned interactionally as a confirmation test for an already available credibility hypothesis. It did not simply investigate what Yeganeh knew. It established the linguistic form in which her knowledge would have to appear in order to be recognized.

The translation and lexical-equivalence issues in this case are discussed more fully in When “Virginity Curtain” Almost Meant a Death Sentence.

Two questions, one methodological problem

The two cases involve different legal systems, different kinds of proceedings, and different linguistic problems.

In Demesme, a disputed answer was extracted from the interaction and evaluated largely as an autonomous verbal object.

In Yeganeh, the question that produced the disputed answer was treated as though it merely tested knowledge, even though it already contained a particular model of what credible knowledge should sound like.

In both cases, interpretive responsibility was placed almost entirely on the respondent.

Demesme was responsible for producing a request sufficiently recognizable to the institution.

Yeganeh was responsible for producing knowledge in the lexical form expected by the institution.

The work performed by the questioner—framing the issue, selecting the relevant category, creating expectations, and restricting the range of acceptable responses—remained comparatively unexamined.

This reflects a broader methodological problem. Forensic and institutional analysis often treats answers as evidence of intention, memory, knowledge, or credibility while treating questions as procedurally neutral prompts.

They are not neutral.

Questions create inferential environments

A question does more than request information. It establishes the conditions under which an answer will be interpreted.

Consider:

Why did you change your story after speaking with your lawyer?

Grammatically, the question requests a reason. Pragmatically, it does considerably more.

It presents as given that the person had an earlier story, that the story changed, and that the change occurred after a conversation with a lawyer. By mentioning the lawyer, it also invites the inference that the conversation is relevant to the change—perhaps even that it caused it.

The respondent must now decide whether to answer the requested question or challenge its premises.

A direct answer may appear to accept that the story changed. A challenge may be treated as evasive. An attempt to explain the premise before answering may be characterized as unnecessarily long or non-responsive.

This is where presupposition and implicature become institutionally consequential.

Presuppositions allow information to enter an exchange as though it were already shared or established. Conversational expectations encourage the respondent to assume that details introduced into a question are relevant. If the question mentions a lawyer, a delay, a cultural practice, or a previous statement, the respondent is encouraged to search for the intended connection.

Ordinary conversation depends on this cooperation. People routinely accommodate incomplete premises and infer relevance so that an interaction can proceed.

In institutional settings, however, conversational cooperation can be retrospectively converted into evidentiary acceptance.

A respondent may answer within the questioner’s frame simply because resisting every premise would make communication nearly impossible. Later, the institutional record may represent that answer as confirmation that the frame itself was correct.

The progression can be subtle:

presupposition → response → accommodation → adverse inference → institutional fact

What began as an assumption embedded in a question may eventually appear in a written decision as something established by the respondent’s own words.

From sentence meaning to responsive meaning

Traditional textual analysis often asks:

What does this sentence mean?

Interactional analysis asks an additional question:

What is this turn doing here, in response to the preceding turn?

The distinction may be described as one between sentence meaning and responsive meaning.

Sentence meaning concerns grammar, vocabulary, reference, and propositional content.

Responsive meaning concerns the action performed by an utterance within a sequence. A response may accept a premise, resist a categorization, correct a misunderstanding, delay an answer, request clarification, challenge authority, or attempt to close the exchange.

The same words can perform different actions depending on what preceded them.

“Yes” may confirm a factual proposition. It may also acknowledge that a question was heard without accepting its wording.

“I don’t know” may express lack of knowledge. It may also resist a demand for speculation.

A long answer may indicate evasion. It may instead be an attempt to correct a question whose available answer choices are inaccurate.

A hesitation may reflect uncertainty about the event. It may also reflect the difficulty of responding without accepting a false presupposition.

These possibilities cannot be evaluated by examining the answer alone.

The value of interactional discourse analysis

Interactional discourse analysis does not merely add “context” around an otherwise complete piece of evidence. It changes the unit of analysis.

The relevant unit is not the disputed statement in isolation. It is the elicitation sequence through which the statement was produced.

Such an analysis should ask:

What was asked?

The analyst must identify both the explicit request for information and the propositions treated as already established.

How was it asked?

Open questions, closed questions, negative formulations, alternatives, repeated questions, accusations, reformulations, and requests for confirmation create different response environments.

Who asked, and who was required to answer?

A question posed by a friend does not have the same interactional force as one posed by a police officer, lawyer, judge, or refugee adjudicator. Institutional authority affects the respondent’s ability to reject the question, correct its premise, remain silent, or demand reformulation.

Where did the question occur in the sequence?

A response may be unintelligible without the preceding accusation, repetition, interruption, clarification, or failed repair. The order of turns is not incidental. It is part of the evidence.

What linguistic or institutional mediation occurred?

Interpretation, transcription, punctuation, summarization, and quotation can transform an answer before it reaches a decision-maker. Each transformation may remove features that originally helped establish its meaning.

How was the answer later characterized?

The final written record may describe a response as evasive, inconsistent, vague, emotional, implausible, or technically deficient. The analyst must determine whether that characterization belongs to the respondent’s language or emerged from the interactional and institutional process.

This does not mean that every inconsistency disappears when the preceding questions are examined. Nor does it mean that respondents can never be evasive or unreliable.

It means that a discourse feature should not be attributed to the respondent until the analyst has considered whether it was elicited, constrained, amplified, or even created by the questioning process.

Questions as forensic evidence

Questions should therefore be treated as part of the linguistic evidence.

Their wording reveals what the questioner considered relevant. Their presuppositions reveal what was being treated as established. Their sequence reveals how the respondent was guided toward particular subjects and categories. Their reformulations reveal which answers were accepted, resisted, or pursued. Their lexical choices may reveal the profile against which the respondent was being evaluated.

In Demesme, attention to the interaction changes the question from:

Was this sentence sufficiently clear?

to:

What action was the speaker attempting to perform as the interrogation developed?

In Yeganeh, it changes the question from:

Did the claimant know the correct anatomical word?

to:

Why was the production of that particular word constructed as a test of professional knowledge and credibility?

Those are not merely broader questions. They are more accurate forensic questions because they examine how linguistic evidence was produced.

Before the answer

Legal and administrative decisions often depend on what a person’s words are believed to reveal: intention, consent, knowledge, fear, memory, sincerity, or credibility.

But those words were elicited somewhere, by someone, at a particular moment, through a particular kind of question.

The evidence is therefore not contained only in what the person said. It also lies in how the person was led, required, or permitted to say it.

Before treating an answer as evidence, forensic analysis should determine what the question presupposed, what it made relevant, which responses it preferred, and what institutional consequences followed from accepting or resisting its frame.

The power to ask a question is also the power to define which inferences will appear reasonable.

That power should not remain outside the analysis.

Daniel Avilán

Sources

• State v. Demesme, 228 So. 3d 1206 (La. 2017).
• Davis v. United States, 512 U.S. 452 (1994).
• Yeganeh v. Canada (Citizenship and Immigration), 2018 FC 714.
• Grice, H. P. (1975). “Logic and Conversation.”
• Stalnaker, R. (1974). “Pragmatic Presuppositions.”
• Sacks, H., Schegloff, E. A., & Jefferson, G. (1974). “A Simplest Systematics for the Organization of Turn-Taking for Conversation.”
• Drew, P., & Heritage, J. (Eds.). (1992). Talk at Work: Interaction in Institutional Settings.