The Berkeley Phonetics & Phonology Forum ("Phorum") is a weekly talk and discussion series featuring presentations on all aspects of phonology and phonetics.
Based on multiple distinct phonological processes in Guébie (Kru), I propose a model of constraint-based phonology that has access to specific morphosyntactic information to account for both affixal and process morphology. Specifically, I describe an instance of scalar tone shift and a process of phonologically determined noun class agreement, both which can be modeled in Cophonology Theory, where the phonology has access to morphosyntactic features. While any single morphophonological process in Gu ebie could be accounted for by one of many phonological models, looking across phenomena in a language allows us to find a single model that best accounts for many distinct processes.
If a child is to produce adult-like speech, then that child’s spectrogram should be a scaled version of that of an adult. This means that a child, whose vocal tract is scaled up by the ratio of its vocal tract length to that of an adult, should, at the very least, produce a sequence of area functions that resembles those of the adult, because formant frequencies are determined by area functions. We explore two types of difficulties that very young children (9 to 24 month old) could have in producing adult-like speech according to the scaled spectrogram criterion, described above. We state the first type of difficulty in terms of hypotheses regarding the curvature of the tongue surface. In two dimensions, curvature can be conceived to be a measure of “waviness” of a curve, and this measure can be applied to sagittal and lateral outlines, such as those of the tongue and palate surfaces. The hypotheses regard limitations on highly curved tongue outlines for children. It will be shown that these hypotheses are consistent with observations of children’s production of syllable-initial, strong /ɹ/, undifferentiated tongue gestures during velar stop, and the two sibilant fricatives of English with diminished acoustic distinction. The other kind of difficulty regards the fact that formant frequencies and some spectral properties of aerodynamic noise sources do not scale in the same way. This is illustrated with oft reported fronting of velar stop consonants in children. It is argued that the child cannot simultaneously produce the expected formant frequency pattern and the expected time evolution of the noise spectrum after the release of an aspirated velar consonant. If time permits, we speculate on the relation between back vowel variability in children and limits on tongue curvature, as well as findings on the relative sizes of the mouth and pharynx in the developing child. We conclude by emphasizing that physical detail should be considered when we study children’s speech development.
Speech perception is subject to critical/sensitive period effects, such that the acquisition of non-native (L2) speech sounds is far more difficult in adulthood than in childhood. Although adults can be trained to perceive differences among speech sounds that are not part of their native language, success is (1) variable across individuals, (2) variable across specific sounds to be learned, and (3) training may or may not generalize to untrained instances. Any theory of L2 speech perception must explain these three phenomena. Accounts of the L2 speech learning process have drawn from traditions in linguistics, psychology, and neuroscience, yet a full description of the barriers to perceptual learning of L2 sounds remains elusive. Evidence from our lab suggests that training on non-native sounds produces plastic effects in the brain regions involved in native-language perception, and that consolidation during sleep plays a large role in the degree to which training is maintained and generalizes to new talkers. Further, similar mechanisms may be at play when listeners learn to perceive non-standard tokens in the context of accented speech. Taken together, these findings suggest that speech perception is more plastic than critical period accounts would predict and that individual variability in brain structure and sleep behavior may predict some of the variability in ultimate L2 sound acquisition success.
It is well known that the way monolingual listeners discriminate non-native speech sounds is strongly influenced by their native (L1) sound system (Best, McRoberts, & Goodell, 2001). Such perceptual constraints also apply to bilingual speakers, as several studies have shown (Antoniou, Best, & Tyler, 2013; Sebastián-Gallés & Soto-Faraco, 1999). However, the influence of language-specific categories on speech perception is less well-understood in the case of bilinguals, and the question of whether bilinguals simultaneously access both of their phonologies during non-native contrast discrimination has not been systematically examined. My study aimed to do so by testing the ability of early bilinguals to discriminate non-native phonetic contrasts consisting of sounds that exist in either their L1 or L2, but not in both. Using forced-choice discrimination and identification tasks, I compared the perceptual performance of Spanish-English bilinguals and English monolinguals on several dental-alveolar stop contrasts found in Nepali. Results showed that despite showing some sensitivity to phonetic differences between sounds, the bilinguals did not significantly differ from the English control group.
An ongoing sound change in American English is /s/-retraction, the process by which /s/ is produced approaching /ʃ/ in the context of /r/. Individuals vary greatly in degree, frequency and distribution of observed retraction. This study seeks to better understand the role of retraction through a manipulation of prosodic position of /str/ /skr/ and /spr/ clusters. The results of this study suggest a general dampening in centroid frequency phrase-initially, with some individuals showing additional effects for the interaction of prosodic position and phonological context. For these individuals, /s/ is most retracted in phrase-initial /str/ clusters, potentially increasing the possibility of target reinterpretation by listeners. These findings suggest a possible role of prosodic position in the actuation of sound change, both in production and possible effects in perception.
I present an analysis of derivational forms in Ese’eja using a special type of output-output faithfulness. Accent on word forms with inflectional morphology is conditioned by syllable count, verb transitivity, and the type of inflectional morphemes present (Vuillermet 2012; Rolle & Vuillermet in press; Rolle in press). In contrast, factors conditioning the position of accent in word forms with derivational morphology are contradictory, and only one generalization can be found: the position of accent in derivational forms is identical to the position in inflected forms. I capture this relation through output-output faithfulness (O-OF), in which outputs are evaluated as faithful to inputs as well as to other outputs (the output base). O-OF in Ese’eja is unique in two ways. First, the inflected forms as a whole act as the output base to derivational forms, which is distinct from canonical examples of O-OF involving paradigm affects within inflectional or derivational paradigms, rather than between them. Second, traditional notions of O-OF account for exceptional stress patterns through transderivational faithfulness between a suprasegmental unit and a segmental unit, e.g.
Many theories of sound change argue that the seeds for sound change lie in the transmission of speech, which is by its nature highly variable, from talker to listener. In previous research, I have explored the effects of variation in articulatory magnitude and inter-articulator timing, separately, on the acoustics (Lin et al., 2014) and perception (Lin, 2011) of post-vocalic laterals in English. This talk is an update on a current project examining the relationship between articulatory magnitude and timing, especially when these post-vocalic laterals are followed by consonants that mask the acoustic consequences of continued articulatory motion. In particular, I ask whether consonants occurring after laterals (such as the /p/ in “help“) overlap the alveolar articulations in the laterals, thus contributing to their acoustic and perceptual “darkness”.
Listeners are usually able to provide consistent (but possibly incorrect) estimates of talker height from speech. Typically, these estimates are thought to be influenced by spectral information mostly via speaker vocal-tract length estimates. I will present the results of an experiment where listeners were asked to estimate the height of 10 female talkers based on natural productions of bVd words containing one of /i æ ɑ u ɝ/. Results indicate that listeners do appear to use vocal-tract length estimates in determining apparent-height. However, apparent-height also varies significantly within-speaker based on the inherent spectral and source characteristics of different vowels. Although the use of acoustic information in the determination of apparent-height was highly systematic, it did not necessarily follow from the empirical relationship between speech acoustics and veridical speaker height, and appears to be constrained by knowledge of the distributional characteristics of the heights of adult females.
Affix telescoping (Haspelmath 1994) is the phenomenon whereby two affixes combine, thereby cutting out the middle man (so that now e.g. alphabet-arian is formed from alphabet without the assistance of *alphabet-ary; cf. document → document-ary → document-ari-an). In this talk I will worry about the role that failing to acknowledge such telescoping may have had in encouraging templatic analyses of agglutinative morphology.
Adult second language (L2) learners, like children learning their first language (L1), must acquire a novel phonological system and new lexical items. Unlike children, however, adult L2 learners have a mature linguistic system in their L1 with the potential to facilitate or interfere with L2 acquisition. Using a lexical decision task, three word-specific properties – phonological neighborhood density (PND), lexical Age of Acquisition (AoA), and usage frequency – were investigated to examine phonological and lexical development in L2 learners as contrasted with native speaker performance. Stimuli were also presented in both plain and foreigner-directed speech (FDS) conditions to examine any possible interactions between word-specific properties and clear speech on perception. Performance by the non-native speakers, late Japanese-English bilinguals, was found to mirror that of native English speakers for AoA and frequency in some conditions; however, the effect of PND on non-native word recognition was found to resemble results observed in L1 acquisition. The implications for this finding on L2 lexical and phonological access will be discussed.
Richtsmeier (2010) argues that most perceived neutralization of contrast in child speech is not wholesale phonological substitution, as previously understood, but rather the missing of articulatory targets. The common child-speech contrast-neutralization pattern known as cluster reduction, or cluster simplification, is understood as deletion of one or more segments of an intended consonant cluster. However, there is evidence that this is not totally true: in Russian-speaking children’s nonword repetition, durations of the productions of intended CC clusters and intended CCC clusters that had been reduced to CC were significantly different (Kavitskaya & Inkelas, 2016), indicating a covert durational contrast between reduced clusters and faithfully produced targets of the same phonological length. This finding supports Richtsmeier’s (2010) assertion that contrast neutralization is actually a missed target, because if children were aiming for the CC cluster target, there would be no additional duration in the reduction. In this talk I examine real child speech for evidence of this durational finding, using Arabic data, since Arabic employs a phonological durational consonant contrast and thus provides a benchmark for determining the relative magnitude of the durational finding via comparison to a child's geminate and singleton targets.