Lu-Feng Shi

Relative Weighting of Semantic and Syntactic Cues in Native and Non-Native Listeners' Recognition of English Sentences

Objective: Non-native listeners do not recognize English sentences as effectively as native listeners, especially in noise. It is not entirely clear to what extent such group differences arise from differences in relative weight of semantic versus syntactic cues. This study quantified the use and weighting of these contextual cues via Boothroyd and Nittrouer’s j and k factors. The j represents the probability of recognizing sentences with or without context, whereas the k represents the degree to which context improves recognition performance. Design: Four groups of 13 normal-hearing young adult listeners participated. One group consisted of native English monolingual (EMN) listeners, whereas the other three consisted of non-native listeners contrasting in their language dominance and first language: English-dominant Russian-English, Russian-dominant Russian-English, and Spanish-dominant Spanish-English bilinguals. All listeners were presented three sets of four-word sentences: high-predictability sentences included both semantic and syntactic cues, low-predictability sentences included syntactic cues only, and zero-predictability sentences included neither semantic nor syntactic cues. Sentences were presented at 65 dB SPL binaurally in the presence of speech-spectrum noise at +3 dB SNR. Listeners orally repeated each sentence and recognition was calculated for individual words as well as the sentence as a whole. Results: Comparable j values across groups for high-predictability, low-predictability, and zero-predictability sentences suggested that all listeners, native and non-native, utilized contextual cues to recognize English sentences. Analysis of the k factor indicated that non-native listeners took advantage of syntax as effectively as EMN listeners. However, only English-dominant bilinguals utilized semantics to the same extent as EMN listeners; semantics did not provide a significant benefit for the two non-English-dominant groups. When combined, semantics and syntax benefitted EMN listeners significantly more than all three non-native groups of listeners. Conclusions: Language background influenced the use and weighting of semantic and syntactic cues in a complex manner. A native language advantage existed in the effective use of both cues combined. A language-dominance effect was seen in the use of semantics. No first-language effect was present for the use of either or both cues. For all non-native listeners, syntax contributed significantly more to sentence recognition than semantics, possibly due to the fact that semantics develops more gradually than syntax in second-language acquisition. The present study provides evidence that Boothroyd and Nittrouer’s j and k factors can be successfully used to quantify the effectiveness of contextual cue use in clinically relevant, linguistically diverse populations.

Acoustic–Phonetic Versus Lexical Processing in Nonnative Listeners Differing in Their Dominant Language

PURPOSE Nonnative listeners have difficulty recognizing English words due to underdeveloped acoustic-phonetic and/or lexical skills. The present study used Boothroyd and Nittrouers (1988)j factor to tease apart these two components of word recognition. METHOD Participants included 15 native English and 29 native Russian listeners. Fourteen and 15 of the Russian listeners reported English (ED) and Russian (RD) to be their dominant language, respectively. Listeners were presented 119 consonant-vowel-consonant real and nonsense words in speech-spectrum noise at +6 dB SNR. Responses were scored for word and phoneme recognition, the logarithmic quotient of which yielded j. RESULTS Word and phoneme recognition was comparable between native and ED listeners but poorer in RD listeners. Analysis of j indicated less effective use of lexical information in RD than in native and ED listeners. Lexical processing was strongly correlated with the length of residence in the United States. CONCLUSIONS Language background is important for nonnative word recognition. Lexical skills can be regarded as nativelike in ED nonnative listeners. Compromised word recognition in ED listeners is unlikely a result of poor lexical processing. Performance should be interpreted with caution for listeners dominant in their first language, whose word recognition is affected by both lexical and acoustic-phonetic factors.

Voicing, devoicing, and noise measures in Shanghainese voiced and voiceless glottal fricatives

Shanghainese has a rather rare voicing distinction between the glottal fricatives /h/ and /ɦ/. We evaluate the acoustic characteristics of this contrast in ten male and ten female speakers of urban Shanghainese dialect. Participants produced 20 CV words with a mid/low central vowel in a short carrier phrase. All legal consonant-tone combinations were used: /h/ preceded high, low, and short tones whereas /ɦ/ preceded low and short tones. Preliminary analyses suggested that the traditional “voiced” and “voiceless” labels for these sounds are not always phonetically accurate; hence we measure the duration of any voicing break relative to the entire phrase, as well as the harmonics-to-noise ratio (HNR) over the time. We expect longer relative voiceless durations and lower HNR measures for /h/ compared to /ɦ/. A question of interest is whether any gender differences emerge. A previous study on American English [Koenig, 2000, JSLHR 43, 1211–1228] found that men phonated through their productions of /h/ more oft...

Measures of spectral tilt in Shanghainese stops and glottal fricatives

Shanghainese differs from other major Chinese dialects in having a three-way contrast among stop consonants. Although phonological descriptions often label these sounds as voiceless aspirated, voiceless unaspirated, and voiced, in contemporary Shanghainese the “voiced” category is not typically produced with closure voicing in initial position. Rather, the vowel onset appears to be characterized by breathy voicing, as demonstrated by acoustic, aerodynamic, and transillumination data. Past studies have been limited to few speakers. This study presents preliminary data of a large-scale study of the Shanghainese stop system as well as the voicing contrast in glottal fricatives, another unusual feature of the language. Data have been collected from 20 male and female native speakers. Measures are made of voice onset time (for the stops), and three measures of spectral tilt (for stops and glottal fricatives): The relative amplitudes of (a) the first two harmonics (H1-H2); (b) H1 and the first formant (H1-A1); ...