How a spoken utterance became a legal problem—and what it reveals about meaning in institutional contexts

A question of legal rights, in this case, begins with a question of punctuation.

In 2017, the The New York Times reported on a case before the Louisiana Supreme Court under a deceptively simple premise: whether a comma could determine if a suspect had requested a lawyer. The phrase at the center of the dispute—“give me a lawyer dog”—was not, in itself, unusual. It emerged in the middle of a police interrogation, produced in the kind of spontaneous, unstructured language that characterizes real speech.

What followed, however, was not a clarification of meaning, but a cascade of interpretations.

Was the suspect addressing the officer as “dog,” in a colloquial vocative—give me a lawyer, dog—or was the phrase to be read literally, as a request for a “lawyer dog”? The distinction, almost trivial at first glance, became legally decisive. The court ultimately treated the statement as ambiguous, and therefore insufficient to invoke the right to counsel.

The case is often retold as a curiosity—an example of how a missing comma can produce absurdity. But this framing obscures the real issue. The ambiguity did not originate in punctuation. It emerged from the interaction between spoken language, institutional expectations, and the interpretive choices of those tasked with assigning meaning.

What is at stake here is not a comma, but a model of language.

The question, then, is not only what was said, but who said it—and under what conditions.

The speaker, Warren Demesme, was interrogated by police in Louisiana in connection with a serious criminal investigation. Like many interrogations, the exchange unfolded in a register far removed from the controlled clarity that legal doctrine tends to assume. The language was informal, fragmented, and shaped by the pressures of the situation.

At one point, in the course of denying involvement, he produced the now-famous line:

“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.”

It is difficult to read this sentence without hearing it.

Its meaning does not reside in its punctuation—there is none—but in its rhythm, its intonation, its context. The phrase dog, in many varieties of spoken English, functions as a vocative: a way of addressing the interlocutor, not of modifying the noun that precedes it. In speech, this distinction is rarely problematic. In writing, stripped of prosody, it becomes suddenly visible.

What the court was asked to determine, however, was not how the sentence sounded, but whether it met a legal threshold: whether it constituted a clear and unequivocal request for counsel.

And here, the linguistic question becomes institutional.

Why does it matter whether the phrase counted as a request for a lawyer?

In U.S. law, once a suspect clearly invokes the right to counsel, the rules governing interrogation change immediately. Questioning must cease until a lawyer is present. Any statement obtained after that point, in the absence of counsel, risks being excluded. The determination, therefore, is not procedural in a minor sense—it defines the admissibility of everything that follows.

This requirement, however, comes with a condition. The request must be sufficiently clear.

In Davis v. United States (1994), the U.S. Supreme Court established that ambiguous or equivocal references to a lawyer do not obligate officers to stop questioning. Statements such as “Maybe I should talk to a lawyer” were deemed insufficient. The burden, in effect, is placed on the speaker to articulate the request in a way that leaves no room for interpretation.

At first glance, this may appear reasonable. Legal procedures depend on thresholds, and thresholds require criteria. But the criterion here assumes a particular model of language: one in which speakers can, under pressure, produce statements that are not only meaningful, but unambiguous according to institutional standards.

The difficulty is that natural language rarely operates under those conditions.

Spoken interaction—especially in high-stress contexts such as police interrogations—is fragmented, indirect, and context-dependent. Meaning is distributed across tone, timing, shared assumptions, and social cues. What counts as a request in ordinary interaction does not necessarily align with what counts as a request in law.

The question, then, is not simply whether the suspect asked for a lawyer, but what kind of language the system is prepared to recognize as such.

The ambiguity in “give me a lawyer dog” does not begin where it appears to.

It is not, strictly speaking, a property of the utterance as produced in speech. It emerges along the way—through a sequence of transformations that are usually treated as neutral, but are anything but.

The first transformation is from speech to text.

Spoken language unfolds in time. It carries rhythm, pauses, stress, and intonation. These features are not decorative; they are part of how meaning is constructed. A vocative—dog, in this case—is typically marked prosodically. It may be separated by a pause, a shift in pitch, or a change in intensity. In interaction, it is rarely ambiguous.

None of that survives transcription.

What remains is a linear string of words:

give me a lawyer dog

At this point, the utterance is no longer heard. It is read. And reading imposes a different set of expectations. Without prosody, the reader must decide how to segment the phrase. Is dog attached to lawyer, forming a compound noun? Or does it stand apart, addressing the interlocutor?

This is the second transformation: from prosodic interpretation to lexical segmentation.

The court’s reading implicitly adopts the first option. It treats dog as part of the noun phrase, yielding the now-infamous “lawyer dog.” But this interpretation is not derived from the speaker’s production alone. It depends on a prior decision: to treat the written string as if its meaning could be resolved without reference to how it was spoken.

A third transformation follows. Once the phrase is framed as ambiguous at the level of words, that ambiguity is attributed back to the speaker. The uncertainty introduced by transcription and interpretation is reclassified as a lack of clarity in the original request.

At that point, the linguistic problem becomes a legal one.

Under the standard articulated in Davis v. United States (1994), only clear and unequivocal requests for counsel trigger the obligation to stop questioning. If the statement is deemed ambiguous, the interrogation may proceed.

What appears, then, as a simple failure of expression is in fact the endpoint of a chain of interpretive decisions.

The ambiguity is not merely in the sentence. It is produced.

What this case ultimately reveals is not a failure of punctuation, nor even a failure of expression, but a mismatch between two models of language.

On one side, there is language as it is used: situated, adaptive, dependent on context, shaped by interaction. Meaning is not contained in words alone, but distributed across timing, tone, and shared understanding. Requests are often indirect, partial, or embedded in larger turns of speech, yet they are nonetheless recognizable in ordinary interaction.

On the other side, there is language as it is processed institutionally: segmented, stabilized, and evaluated against predefined thresholds. Meaning must be recoverable from the verbal surface, independently of the conditions in which the utterance was produced. Ambiguity, in this model, is treated as a deficiency of the speaker rather than as a feature of the medium.

The difficulty arises when these two models are forced to align.

The legal standard does not ask whether a request was made in a way that would be understood in ordinary interaction. It asks whether the request was articulated in a form that satisfies institutional criteria of clarity. In doing so, it transforms a problem of interpretation into a problem of expression.

Seen from this perspective, the question is no longer whether the defendant asked for a lawyer, but what counts, for the purposes of the system, as having asked.

That distinction may appear subtle. In practice, it determines when legal protections begin—and when they do not.

Daniel Avilán

Sources

• The New York Times (2017). “Was That a Request for a Lawyer? Depends on the Comma.”
• The Washington Post (2017). “The suspect told police, ‘give me a lawyer dog.’ The court says he wasn’t asking for a lawyer.”
• State v. Demesme, 228 So. 3d 1206 (La. 2017). Louisiana Supreme Court.
• Davis v. United States, 512 U.S. 452 (1994). U.S. Supreme Court.