Your Own Voice Can Influence Your Perception of Speech

First off: Name a person who’s always talking very quickly…

OK, now name a person who’s always talking slowly…

Odds are that you didn’t have to think long before you thought of (at least) two names. Speakers typically differ in their habitual speech rate (some slow, some fast), depending on age, extraversion, regional background, etc. At the same time, we’re also capable of varying our speech rate, slowing down when talking in public, or speeding up when conversing with friends. Considering all this variation in speech rate, it’s quite astounding that our ears (and the brain attached to them) seem to have little trouble understanding speech produced at different rates.

One trick that listeners use to overcome rate variation in the speech signal is to ‘normalize’ sounds for the surrounding speech rate, as illustrated in the example below. First, consider a sound that is ambiguous, between a short b sound and a longer w sound. If you hear this ambiguous b/w sound preceded by a rather fast sentence, suddenly the ambiguous sound sounds more w-like! That is, relative to the short sounds in the fast sentence (vowels and consonants with relatively short durations), the ambiguous b/w sound that follows them stands out as relatively long, meaning that most listeners will report hearing w (and vice versa for slow contexts).

Schematic illustration of rate normalization. The same ambiguous target word (e.g., in between short /b/ and long /w/) is presented in the context of two different speech rates: a slow sentence and a fast sentence. Hearing the ambiguous target word preceded by a fast context makes the target sound sound longer (e.g., more /w/-like).

Source: Hans Rutger Bosker

Consider normal conversation. Our own speech and the speech of our conversational partner follow each other at a very rapid pace. On average, the silent gap in between the speech from two talkers is only 100 milliseconds long (a tenth of a second) meaning that we hear our friend’s speech in the context of our own speech. Now, can our own speech rate influence how we perceive the speech produced by someone else?

In a new psycholinguistic study published in the Journal of Experimental Psychology: Learning, Memory, and Cognition, participants in two experiments were first asked to listen to speech. They heard slow and fast sentences, followed by ambiguous speech sounds (in this case, vowel sounds somewhere between a short /ɑ/ and a long /a:/ in Dutch). Their task was to indicate which word they heard at the end of a sentence (e.g., “tak” with a short /ɑ/ meaning branch vs. “taak” with a long /a:/ meaning task). The outcomes of this first experiment supported the ‘rate normalization’ explained above (see figure below). That is, participants reported hearing more long /a:/ words when the ambiguous sounds were played in the context of a fast sentence (vs. the same ambiguous words in the context of a slow sentence).

Results of the first two experiments of the study. The panel on the left shows the data for Experiment 1, in which participants simply listened to fast and slow sentences, followed by ambiguous words. The solid line is consistently higher than the dashed line, indicating that fast contexts biased listeners to report hearing more long /a:/ words. The panel on the right shows the data for Experiment 2, in which participants produced fast and slow speech themselves, followed by those ambiguous words. Again the solid line is above the dashed line, indicating that talking at a fast rate biased perception towards long /a:/ words.

Source: Hans Rutger Bosker

Now, crucially, in the second experiment, the same participants were asked to produce fast and slow speech themselves. After each sentence they produced, these same ambiguous words were played to them, and they had to indicate which word they heard, “tak” or “taak”, etc. Interestingly, participants’ perception of the ambiguous vowels depended on their own preceding speech rate: They were more likely to say that they heard a long /a:/ word if they had been talking at a fast speech rate a moment earlier.

This study shows that characteristics of our own voice (in this case, rate of speech) can influence how we perceive what others say. In these specific experiments, all participants produced the same fast and slow speech that influenced subsequent speech perception. Nevertheless, one can imagine that these findings may actually carry implications for habitually fast vs. habitually slow talkers. It would be interesting to explore whether habitually fast and slow talkers differ not only in their talking, but also in their listening.