Volume 72, Issue 2-3

The p honeme /ʝ/ in Spanish is known to have a wide range of phonetic realizations, even within a single dialect. Whereas the influence of different contextual factors in conditioning the allophony of /b d ɡ/ has been the object of much recent acoustic investigation, somewhat surprisingly there are no comparable studies of the voiced palatal obstruent. We report on an acoustic investigation of this consonant in Peninsular Spanish. Both preceding context and stress appear to be relevant factors. We also find an effect of style, with more constriction in more formal styles. The presence of a word boundary also affects the degree of constriction, although to a lesser extent. Although we have identified several conditioning factors in the degree of constriction of this consonant, compared to /b d ɡ/, there appears to be greater variability in a given context, including very constricted, stop-like realizations of /ʝ/ in conversational styles even within the intervocalic context. Whereas our results largely confirm previous nonquantified description, they offer a more nuanced view of the continuum of the degree of constriction in the realization of this consonant and the factors that affect it, going beyond earlier views that postulated distinct allophones.

The origin and nature of ‘irregular', ‘sporadic' sound changes have been debated by different theories of phonological change since at least the Neogrammarians. They are often attributed to non-phonological factors, as analogy or borrowing, or to the non-(purely)-phonological mechanism of lexical diffusion. The aim of this paper is to show that an irregular sound change in the historical phonology of Tuscan - namely the voicing of some intervocalic voiceless stops - is not due to borrowing (as often argued), but to a variable and allophonic voicing rule, whose output was only partially subject to phonological recategorization. The most likely causes for this irregular outcome are the variable strength intervocalic voicing had in different environments, the interaction with another lenition process and the perceptual ambiguity of the lenited stops.

This is a study of final poststressed vowel devoicing following /s/ in Brazilian Portuguese. We contradict the literature describing it as deletion by arguing, first, that the vowel is not deleted, but overlapped and devoiced by the /s/, and, second, that gradient reduction with devoicing may lead to apocope diachronically. The following results support our view: (1) partially devoiced vowels are centralized; (2) centralization is inversely proportional to duration; (3) total devoicing is accompanied by lowering of the /s/ centroid; (4) the /s/ noise seems to be lengthened when the vowel is totally devoiced; (5) aerodynamic tests reveal that lengthened /s/ has a final vowel-like portion, too short to be voiced; (6) lengthened /s/ favors vowel recovery in perceptual tests. This seems to be a likely path from reduction to devoicing to listener-based apocope.

Thi s paper investigates similarities between lexical consonant clusters and CVC sequences differing in the presence or absence of a lexical vowel in speech perception and production in two Portuguese varieties. The frequent high vowel deletion in the European variety (EP) and the realization of intervening vocalic elements between lexical clusters in Brazilian Portuguese (BP) may minimize the contrast between lexical clusters and CVC sequences in the two Portuguese varieties. In order to test this hypothesis we present a perception experiment with 72 participants and a physiological analysis of 3-dimensional movement data from 5 EP and 4 BP speakers. The perceptual results confirmed a gradual confusion of lexical clusters and CVC sequences in EP, which corresponded roughly to the gradient consonantal overlap found in production.

This paper investigates the production of English unstressed vowels by two groups of early (ESp) and late Spanish (LSp) bilinguals and a control group of native English (NE) monolinguals. Three acoustic measurements were obtained: duration and intensity ratios of unstressed to stressed vowels, normalized vowel formants and euclidean distances. Both groups of bilinguals showed significantly fewer differences in duration between stressed and unstressed vowels than the NE monolinguals. Intensity differences depended on whether the stress pattern of the target English words matched the stress pattern of their Spanish cognates. As for vowel quality, the early bilinguals reduced the unstressed vowels, which clustered around the midcenter area of the vowel space, in the same fashion as the NE monolinguals, suggesting that vowel reduction might be operating at the phonological level. However, the late bilinguals showed a context-dependent, phonetic-level pattern with vowels that were more peripheral in the vowel space.

The present study explored the production and perception of the /i/-/I/ vowel contrast in second language (L2)-dominant early learners of American English who no longer fluently speak their first language (L1, Spanish). The production task analyzed the extent to which the early learner group differed from controls (native English speakers and L1-Spanish late-onset learners of English) with regard to duration and spectral centroids. The perception experiment examined how these early learners classified resynthesized stimuli drawn from the /i/-/I/ contrast using distinct acoustic cues - spectral and temporal - in a 2-alternative forced choice identification task. The first experiment revealed that the early learners produced the contrast in a native-like manner in terms of the spectral envelope and duration use. The second experiment found that early learners differed from both control groups in how they categorized the /i/-/I/ continua based on spectrum and duration, and the extent to which they rely on these two cues. The effects of linguistic experience on L2 phonetic behavior are discussed.