Word-level intelligibility of time-compressed speech: prosodic and segmental factors

Abstract

In this study we investigate whether speakers, in line with the predictions of the Hyper-and Hypospeech theory, speed up most during the least informative parts and less during the more informative parts, when they are asked to speak faster. We expected listeners to benefit from these changes in timing, and our main goal was to find out whether making the temporal organisation of artificially time-compressed speech more like that of natural fast speech would improve intelligibility over linear time compression. Our production study showed that speakers reduce unstressed syllables more than stressed syllables, thereby making the prosodic pattern more pronounced. We extrapolated fast speech timing to even faster rates because we expected that the more salient prosodic pattern could be exploited in difficult listening situations. However, at very fast speech rates, applying fast speech timing worsens intelligibility. We argue that the non-uniform way of speeding up may not be due to an underlying communicative principle, but may result from speakersÕ inability to speed up otherwise. As both prosodic and segmental information contribute to word recognition, we conclude that extrapolating fast speech timing to extremely fast rates distorts this balance between prosodic and segmental information.

Word perception in fast speech: artificially time-compressed vs. naturally produced fast speech

“…Selective compression, based on natural fast speech timing, might in fact have improved intelligibility at the speech rate at which this timing pattern was observed (about 1.5 times normal rate), but not at the very fast rate employed in (Janse et al, 2003) perception study (about three times normal rate). If, on the other hand, the changed timing in natural fast speech is indeed due to articulatory restrictions, and not to a communicative strategy, then word perception of artificially time-compressed speech is not helped by making its temporal pattern closer to that of natural fast speech, not even at a rate of speech that speakers can attain.…”

“…Prosody has been shown to be an important source of information in lexical processing (Cutler and Clifton, 1984;Cutler and van Donselaar, 2001;Cutler and Koster, 2000;van Heuven, 1985;Slowiaczek, 1990), and listeners rely even more on prosodic information in case of adverse listening conditions (van Donselaar and Lentz, 1994;Wingfield, 1975;Wingfield et al, 1984). However, for this heavily time-compressed speech, making the timing pattern more similar to that of natural fast speech turned out to have a negative effect on intelligibility, relative to linear time compression (Janse et al, 2003).…”

“…When people speak faster, consonant durations are reduced less, relatively, than vowel durations (Gay, 1978;Max and Caruso, 1997). When speech rate is increased, durations of unstressed syllables in polysyllabic words are reduced more than those of stressed syllables, relatively, which makes the prosodic pattern more prominent (Janse et al, 2003). Furthermore, at sentence level, durations of sentencestressed syllables are reduced less, relatively speaking, than durations of unstressed syllables (Peterson and Lehiste, 1960;Port, 1981).…”

“…However, this nonlinear type of speed-up need not be beneficial for the listener. Janse et al (2003) started from the assumption that the enhanced prosodic pattern found in naturally produced fast speech might improve the intelligibility of artificially time-compressed speech: if speakers time-compress speech in a selective way for the sake of the listener, listeners should benefit from such nonlinear time compression. The duration study showed that, when speakers speed up their speech rate, unstressed syllables in disyllabic words were affected more in duration, relatively, than stressed syllables.…”

“…Three objections may be raised against the experimental set-up in (Janse et al, 2003) that might explain why imitating the rules of natural fast speech timing showed no improvement over linear time compression.…”