Sandra Schwab

Regional variation and articulation rate in French

The present investigation deals with regional variation and articulation rate in French. The articulation rate in read speech and in spontaneous speech was examined in seven variants of French: Paris and Lyon in France; Tournai and Liege in Belgium; Geneva, Neuchâtel and Nyon in Switzerland. Results showed that Swiss speakers articulate at a lower syllable rate than French speakers (especially Parisian speakers) and Belgian speakers, independently of the speaking style (reading or conversation). This finding confirms that articulation rate varies regionally. Moreover, results revealed that extra-linguistic and linguistic factors, such as the speakers age and gender, the speaking style, the utterance length and the articulation rate of the adjacent inter-pause chunk, also affect articulation rate.

Effect of Speaking Rate on the Identification of Word Boundaries

Two experiments were conducted to determine whether listeners’ ability to use allophonic variation to identify word boundaries is influenced by speaking rate. Listeners in both experiments were presented two-word sequences (such as great eyes) spoken by naturally fast and naturally slow talkers; in one experiment the sequences were presented in quiet and in the other they were presented in noise. The listeners’ task was to identify the intended sequence from among four choices with alternative segmentations (e.g. great eyes, gray ties, great ties, gray eyes). In both experiments performance was worse for the sequences produced by the naturally fast talkers than for those produced by the naturally slow talkers. This finding suggests that the extent to which allophonic variation contributes to the identification of word boundaries may depend on the rate at which the speech was produced.

Relationship between Speech Rate Perceived and Produced by the Listener

The present research examines the impact of listeners’ own rate of speech production on their perception of speech rate. We recorded 28 native French speakers reading a passage at normal, fast and slow rates. The same speakers then assessed the rate of production of these samples by all speakers and at all rates (normal, fast and slow), using a magnitude estimation task. We show that there is an inverse relationship between listeners’ own rate of production and their judgment of rate, so that listeners with slower rates tend to overestimate sample rates relative to faster speakers. This effect of listeners’ own rate was present in the magnitude estimation results at slow and normal rates, but not at fast rates. Furthermore, listeners with a slow rate are more sensitive to rate differences within a given rate category (normal, fast and slow). The listener’s own rate should therefore be considered as a critical factor in speech rate perception.

La perception du débit en langue seconde

Non-native speakers of a second language often report that the speech rate in that language is faster than the rate in their own language. So as to compare speech rate perception by native (French) and non-native (Swiss German) speakers, and to determine if rate estimation by non-native speakers is correlated with their level of comprehension, we asked two groups of 96 participants, native and non-native speakers of French, to listen to short stories read at slow, medium and fast rates. They were asked to answer a few comprehension questions and to give an estimate of the speech rate. The results obtained show that there is indeed a difference between the two groups: the faster the physical speech rate, the greater the impression of speed in the non-native speakers as compared with the native speakers. In addition, when speech rate is slow and normal, there is a significant negative correlation between oral comprehension and perceived rate: the lower the comprehension, the higher the estimated rate.

The perception of speech rate in non-manipulated, reversed and spectrally rotated speech reveals a subordinate role of amplitude envelope information

Previous research suggests that the broad-band amplitude envelope (ENV) of speech is crucial for the perception of speech rate and timing. The present experiment tested this claim using non-manipulated and spectrally rotated speech (rotated around 2.5 kHz) with a bandwidth of 5 kHz which both contain identical ENV and reversed speech in which the temporal organisation of ENV is distorted. 44 listeners of Swiss German rated perceived speech tempo on a continuous scale reaching from “rather slow” to “rather fast” in 48 stimuli (4 sentences × 4 speakers × 3 signal conditions). Results revealed a significant effect of signal condition. Both reversed and spectrally rotated speech were perceived as significantly faster than clear speech but there was no difference between spectrally rotated and reversed speech. Results were consistent for all sentences and speakers. Results suggest that the intelligibility of the signal plays a higher role in the perception of speech rate than the presence of the ENV.Previous research suggests that the broad-band amplitude envelope (ENV) of speech is crucial for the perception of speech rate and timing. The present experiment tested this claim using non-manipulated and spectrally rotated speech (rotated around 2.5 kHz) with a bandwidth of 5 kHz which both contain identical ENV and reversed speech in which the temporal organisation of ENV is distorted. 44 listeners of Swiss German rated perceived speech tempo on a continuous scale reaching from “rather slow” to “rather fast” in 48 stimuli (4 sentences × 4 speakers × 3 signal conditions). Results revealed a significant effect of signal condition. Both reversed and spectrally rotated speech were perceived as significantly faster than clear speech but there was no difference between spectrally rotated and reversed speech. Results were consistent for all sentences and speakers. Results suggest that the intelligibility of the signal plays a higher role in the perception of speech rate than the presence of the ENV.