Volume 61, Issue 2-3

The aim of this study was to evaluate the acquisition of statistical properties ofa second language (L2). Stop consonants are permitted in word-final position inboth English and Korean, but they are variably released in English and invariablyunreleased in Korean. Native Korean (K) adults and children living in North Americaand age-matched native English (E) speakers repeated English words ending inreleased tokens of /t/ and /k/ at two times separated by 1.2 years. The judgments ofE-speaking listeners were used to determine if the stimuli were repeated with audiblerelease bursts. Experiments 1 and 2 revealed fewer final releases for K thanE adults, and fewer releases for /t/ (but not /k/) for K than E children. Nearly all /t/ and/k/ tokens were heard as intended in experiment 3, which evaluated intelligibility.However, the K adults’ /k/ tokens were identified with less certainty than theE adults’. Taken together, the results suggested that noncontrastive (i.e., statistical)properties of an L2 can be learned by children, and to a somewhat lesser extent byadults.

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

This study investigates the extent to which phonetic voicing is maintained inword-final clusters composed of an underlying voiced stop followed by nonsyllabic/l/ or /r/ in Majorcan Catalan. Electropalatographic and acoustic data forfive speakers of this Catalan dialect reveal that, in agreement with the non-syllabicstatus of the liquid, voicing for /l/ is only available if occurring during the precedingstop. The rhotic is always phonetically voiceless. Speakers differ regarding theextent to which they keep the underlying stop voicing distinction and the productionstrategies they use for that purpose. This distinction is highly robust and distributedover the entire syllable in nasal-stop-/l/ clusters for some speakers, butmuch less clear or absent for those speakers who devoice /l/ as a general rule.Underlying stop voicing is cued primarily by stop closure duration and vocal foldvibration, or else by closure duration rather than by voicing. It may be concludedthat the word-final devoicing process operating in Catalan does not apply toMajorcan Catalan tautosyllabic stop clusters with a liquid, and that phonetic voicingmay affect just the stop or both the stop and the liquid. The implications ofthese findings for sound change are discussed.

The evolution from final obstruents to final glottal stop and then to rhymeglottalization (i.e. from /at/ to /aʔ/, then to /aˀ/) is a well-established general trendin the history of the Sino-Tibetan language family and beyond. It has further beenshown by laryngoscopy that in three languages which retain the nonreleasedsyllable-final obstruents /p/, /t/ and /k/ (Standard Thai, and two Chinese dialects),these obstruents are often accompanied by a glottal stop. The present researchraises the issue whether there is another typological possibility: can nonreleasedfinal obstruents be accompanied consistently by modal phonation, without glottalstop? Analysis of electroglottographic recordings of 126 syllables in two carriersentences spoken by 4 speakers shows that, in Hanoi Vietnamese, the final obstruents/p/, /t/ and /k/ are not accompanied by glottalization, and that the open quotientincreases in the course of the syllable rhyme. Obstruent-final rhymes (whichmay carry either of two tones: D1 or D2) are compared with nasal-final rhymeswhich, under one of the tones (tone B2), are confirmed to be glottalized. Our findingis that tones D1 and D2 (i.e. obstruent-final rhymes) are both produced inmodal voice, which shows that the typological paradigm of observed realizationsof syllable-final obstruents must be enlarged. The discussion puts forward thehypothesis that the unusual association of segments and voice quality found inHanoi Vietnamese is a strategy to maintain the opposition between B2-tone andD2-tone rhymes.

Analyses of the perceptual and acoustic characteristics of the Register 1(ʼclearʼ) versus Register 2 (ʼbreathyʼ) distinction have been carried out on the Kuaidialect of Suai, a Mon-Khmer language. The perception results were obtainedfrom five-parameter synthesized stimuli. They showed that the primary parameterunderlying the distinction is the frequency of onset of laryngeal excitation (F0).One other parameter making a significant contribution was the open quotient. TheF0 result was confirmed by an acoustic analysis of eight pairs of natural utterancesproduced by native speakers. We conclude that the Suai language is in a stateof flux with respect to the voice registers, although the distinction has not disappeared.The perceptual data reveal mixed levels of sensitivity, and the productiondata indicate that some speakers maintain a fairly good distinction, while othersdo not. The language seems to be replacing the register distinction with a prosodicone of pitch accent, possibly as a stage leading to tonogenesis.