How We Talk

While we tend to think of a narrative as a monologue, which by definition involves one person, in everyday conversation the listener makes important contributions throughout. These contributions feed back into the performance of the narrator, implying that the two (or more) people involved are hooked up to each other, together creating the conditions for the conversation machine to run.

One use for “um” and “uh” is at the beginning of turns that are dispreferred responses. We saw in the last chapter that dispreferred responses are typically delayed, and we also saw that this delay is not only created by inserting a silence, but also by inserting audible alternatives to beginning the turn, including “um” and “uh.”

In nearly half of the dispreferred responses, the first sound one hears is not a word at all, but an in-breath (or click, that is, a “tut” or “tsk” sound). By contrast, preferred actions mostly do not start with these nonword sounds (only 17 percent do). So the reason why dispreferred responses average slightly shorter silences is that people are starting quicker with things that are not words (but which signal that a dispreferred response is forthcoming).

This example shows that a delay in response, as occurs in line 2, is a signal that the respondent is sensitive to the bias of the question and is opting not to produce an answer that goes against this bias. This would account for their delay in response. In turn, the delay may indicate that Person B’s answer is going to be “no.” Person A’s reformulation changes the bias of the question in such a way that “no” becomes the preferred response, and in this context Person B can proceed without delay.
Here is a similar case:2
1. A: What about coming here on the way?
2. (silence)
3. A: Or doesn’t that give you enough time?
4. B: Well no I’m supervising here.

In the short questions, the word “student” was almost always pronounced at a pitch above 360 hertz, while in the long questions, the same word was almost always pronounced at a pitch below that point. Second was a difference in the length of time it took to pronounce the last syllable of “student.” When the word “student” occurred at the end of the turn, as in the short questions, it was lengthened, while in the long questions it was short.

On the measure of how long it took for the fall to occur, the interrupted turns sounded very different from the turns that kept going. The interrupted turns featured a sudden drop in pitch, even more sudden than the turns that were actually complete. By contrast, the turns that kept going had a final pitch fall nearly twice as long in duration.

They discovered that the journalists were not meaning to interrupt. Rather, when it sounded like Mrs. Thatcher was winding down to the end of her sentence, they would launch their next questions, only to find that she wasn’t done. The result was that it sounded like they were interrupting her. Beattie and colleagues did a phonetic analysis of Thatcher’s speech and revealed that she was actually playing by slightly different rules from her interviewers. Her intonation would feature a sharp and rapid drop toward the end of certain phrases, and this appeared to send the signal “I’m finishing my turn now,” when in fact she had more to say.

Most speaker transitions are timed such that there is little if any gap or overlap between speakers. Gaps and overlaps are common enough, but the vast majority of them run for less than half a second, suggesting that no-gap-no-overlap is a norm or standard. This is surprising because following the rules is not something people always do in language. Many of the rules found in grammar books—think “Don’t finish a sentence with a preposition”—are broken all the time. But the one-at-a-time rule is not like this.

The caller announces that they want to ask something. But they do not then immediately ask it. They first give the background to the question. The host gives feedback, showing that they are listening, until the question itself comes. Then the host answers. The example shows that by using these kinds of preliminaries, we can reserve or block out time for our own input in conversation, and ramp up the degree of commitment that we demand of others in conversation. Here, not only is the host obliged to answer the question, but prior to that they are required to sit tight and pay attention while the questioner gives the relevant background.

As we have seen from the above cases, a person who is asked a question is obliged to provide a response, preferably an informative answer. But as any politician knows, when it comes to the content of an answer, there is leeway. As former US Secretary of Defense Robert McNamara put it, “Don’t answer the question you were asked. Answer the question you wish you were asked.”