Cross-language differences in cue use for speech segmentation

Abstract

Two artificial-language learning experiments directly compared English, French, and Dutch listeners' use of suprasegmental cues for continuous-speech segmentation. In both experiments, listeners heard unbroken sequences of consonant-vowel syllables, composed of recurring three-and four-syllable "words." These words were demarcated by ͑a͒ no cue other than transitional probabilities induced by their recurrence, ͑b͒ a consistent left-edge cue, or ͑c͒ a consistent right-edge cue. Experiment 1 examined a vowel lengthening cue. All three listener groups benefited from this cue in right-edge position; none benefited from it in left-edge position. Experiment 2 examined a pitch-movement cue. English listeners used this cue in left-edge position, French listeners used it in right-edge position, and Dutch listeners used it in both positions. These findings are interpreted as evidence of both language-universal and language-specific effects. Final lengthening is a language-universal effect expressing a more general ͑non-linguistic͒ mechanism. Pitch movement expresses prominence which has characteristically different placements across languages: typically at right edges in French, but at left edges in English and Dutch. Finally, stress realization in English versus Dutch encourages greater attention to suprasegmental variation by Dutch than by English listeners, allowing Dutch listeners to benefit from an informative pitch-movement cue even in an uncharacteristic position.

The Use of Prosodic Cues in Learning New Words in an Unfamiliar Language

“…For example, Warner, Otake, and Arai ( 2010 ) argued that the crosslinguistic use of rising F0 found in Japanese and French (Welby, 2007 ) as well as Korean (Kim, 2004 ) is due to the universally applicable auditory-perceptual salience of F0 rise. Phrasefi nal lengthening has also been claimed to be a universally applicable segmentation cue, as English, Dutch, and French listeners use the fi nal lengthening cue in lexical segmentation (Tyler & Cutler, 2009 ).…”

Phonetic richness can outweigh prosodically-driven phonological knowledge when learning words in an artificial language

“…In particular, as discussed in the introduction, the fact that Korean listeners perceive (phonetically poorer) unreleased stops better than (phonetically richer) released stops in processing non-native speech has provided a concrete example of a situation where phonological experience overrides phonetic richness in non-native speech perception (Cho & McQueen, 2006). The language-specificity of speech perception has also been observed in lexical segmentation of unfamiliar artificial languages or non-native languages-e.g., the use of language-specific phonotactics (Weber & Cutler, 2006) and rhythmic/prosodic cues (Cutler & Otake, 1994;Kim et al, 2012;Tyler & Cutler, 2009) in processing non-native speech. The results of the present study therefore appear to be in contrast to the generally observed dependency that listeners have on their phonological knowledge in the perception of unfamiliar or non-native speech.…”

“…Listeners had to identify the words in the test phase. This paradigm has been shown to be effective in testing effects of specific cues in the speech signal without recourse to any prior lexical knowledge and it has also revealed listeners' use of native segmentation cues in lexical segmentation of these unfamiliar, non-native languages (e.g., Bagou et al, 2002;Kim et al, 2012;Saffran et al, 1996;Tyler & Cutler, 2009).…”

“…Cho and McQueen's (2006) phoneme monitoring study, for example, found that Korean listeners, who are always exposed to unreleased word-final stops in their native language, detected unreleased stops (phonologically viable in their native language yet phonetically poorer) more efficiently than released stops (phonologically unviable in Korean yet phonetically richer) in processing speech in Dutch and English, while Dutch listeners showed the opposite pattern. Recent studies which employed an artificial language learning paradigm have also shown that listeners are not able to use an acoustically salient cue (i.e., high pitch) in lexical segmentation when the cue appeared in a position where listeners would not normally expect that cue (Kim et al, 2012;Tyler & Cutler, 2009).…”

“…Listeners can use the acoustic correlates of prosodic structure to decode those structures, facilitating segmentation of words at prosodic boundaries (e.g., Christophe, Peperkamp, Pallier, Block, & Mehler, 2004). For instance, word-and phrase-final lengthening appears to facilitate segmentation of words not only in continuous native-language speech (Kim & Cho, 2009;Salverda, Dahan, & McQueen, 2003), but also in artificial-language speech streams (Bagou, Fougeron, & Frauenfelder, 2002;Kim, Broersma, & Cho, 2012;Saffran, Newport, & Aslin, 1996;Tyler & Cutler, 2009). Likewise, it is well-known that stress patterns (cued e.g., by longer duration of stressed syllables in English and Dutch) are exploited in lexical segmentation (Cutler & Norris, 1988;Quené , 1993;Sluijter & van Heuven, 1996).…”