How we learn to pronounce the sounds of speech: (1) Infants

2008

Timing patterns and the qualities of speech sounds are two important aspects of pronunciation. It is generally believed that imitation from adult models is the mechanism by which a child replicates them. However, this account is unsatisfactory, both for theoretical reasons and because it leaves the developmental data difficult to explain. I describe two alternative mechanisms. The first explains some timing patterns (vowel length changes, 'rhythm', etc.) as emerging because a child's production apparatus is small, immature and still being trained. As a result, both the aerodynamics of his speech and his style of speech breathing differ markedly from the adult model. Under their constraints the child modifies his segmental output in various ways which have effects on speech timing but these effects are epiphenomenal rather than the result of being modelled directly. The second mechanism accounts for how children learn to pronounce speech sounds. The common, but actually p...

Journal of Chemical Physics, 2006

We consider two theories of laryngeal representation, one using a single feature [voice] generalizing across prevoicing languages and aspiration languages, and the other using multiple features: [voice] for pre-voicing languages and [spread glottis] for aspiration languages. We derive predictions for children's early productions, and test these for three Germanic languages. Children acquiring Dutch, a prevoicing language, show de-voicing of stops, while available data from German, an aspiration language, show de-aspiration. Although the difference might simply reflect intrinsic properties of children's early production and perception systems, we argue that a representational account is in order, based on multiples features. The case is made for English, an aspiration language, based on the early productions of a single child. A laryngeal harmony pattern is found which spreads voicelessness from coda to onset, which is argued to involve activity of [spread glottis]. This is interpreted as evidence for a laryngeal representation involving multiple features.

Language, 1994

Adults and older children learning the sound system of a second language (L2) differ from young children acquiring the sound system of their native language (LI) in two important respects. The L2 learners are better able to control their speech apparatus than the young children ilcquiring an Ll, ilnd they alreildy possess a phonetic systcm for producing speech. As il result, far more errors in production ilre likely to arise from the inilppropriilte use of !Jrc(liollsly Ilcqllired structllres in L2 learning than in Ll acquisition.' It is well known that adult learners are mrely, if ever, completely successful almaslering the sound system of iln L2. A widely held view is Ihilt when ildults encounter iln L2 word, they attempt to "decompose" it into the phonemic units of the Ll, ilnd then produce the L2 word as if it consisted of phonic elements (illiophones, phonemes) from the LI (I'o/ivilnov 19:11). This chapter provides a brief overvicw of n.'seiHch Ihill hilS de-'We use the term L2tCll/llil/,'i r.lther than /1c'1l1isili"l/b,'C.HISI• 01 till' view th ••t I'honelic systel11s, e\'l'n those of adults, undergo constant change in till' laC(' 01 new phonetic input. Thus, speech is never fully /1cqllired. The terl11 sl't't'dllcnmi'J,'i as used here refers to all aspects 01 learning that affect the production and I,,'reeption of the sounds 111.1 king up words. II is used in preference to 1,11I11/(1/o,'ii((/III'/1/11ill,'i because lI1ueh 01 our res(',uch to date has focused on phonetic-level processes. The preparation of this chal'tl'r was supported by NIII grant IJC00257. TI1<' author thanks wlkagues (0.-5. Bohn, W. Eplting, M. Munro). graduate students (K. Fletcher, S. Jang, C. Wang), and research assistants (L. Cueva, C. 1\1l'na, L. Skelton) who h'l\'e p,utieipated in work reported here. Thanks arc also extended to O.-S. lIohn. C. Ferguson, and 1\. Walle)' lor editorial C0Il11111'nts.

De Gruyter eBooks, 2009

In this article, I discuss a linguistic model for explaining second-language (L2) sound perception, which is a phenomenon that has commonly been modelled within the disciplines of phonetics and psycholinguistics. This linguistic model will be applied to the learning of SIMILAR L2 sounds. This L2 learning scenario refers to the acquisition of the knowledge involved in the perception of L2 sounds that are phonologically equivalent but yet phonetically different from the acoustically closest sounds in the learner's first language (L1). In the introduction, I argue, based on phonetic and psycholinguistic grounds, that speech perception is a language-specific phenomenon that should be brought into the domain of phonological modelling. Additionally, I propose a number of characteristics which incorporate phonological, phonetic, and psycholinguistic modelling and which should be found in a comprehensive and explanatory adequate model for sound perception. In § 2, I demonstrate that the Linguistic Perception (LP) model complies with these criteria. Crucially, I show that the L1 acquisition component of the LP model is shown to constitute a successful proposal for the mechanisms involved in learning to perceive L1 sounds. In § 3, I show how the L2 version of the LP model successfully describes, explains, and predicts the learning of SIMILAR L2 sounds. Specifically, the model predicts that listeners are optimal perceivers of their native language and that beginning L2 learners start with a copy of their L1 optimal perception. These two predictions are confirmed by the perception of /ae/ and // by monolingual Canadian English (CE) and Canadian French (CF) listeners and by the L2 perception of beginning CE learners of CF. Further, the model predicts that learners will adjust their initial L2 perception by means of the same mechanism used by L1 learners. This developmental prediction is confirmed by the gradual shifting of the category boundary between /ae/ and // in CE learners of CF. Finally, the L2LP model hypothesizes that both L1 and L2 can be optimal because they are handled by two separate grammars. The data demonstrate that CE learners of CF have differential perception systems for their L1 and L2. In sum, it is shown that this model provides the currently most comprehensive description, explanation, and prediction of L2 sound perception. It successfully incorporates an L2 phenomenon which was commonly regarded as phonetic or psycholinguistic within the domain of phonology, a modelling proposal which follows the tradition started by .

Proceedings of the 26th Annual Boston …, 2002

2021

The acquisition of a sound system is an integral component of second language (L2) communication, yet it is one of the most difficult skills to teach and is therefore largely ignored in L2 classrooms (Derwing, 2010). In laboratory settings, phonetic training studies have typically examined syllables, rather than words, with no referential meaning. Support for this decontextualization of the stimuli and the subsequent distributional learning that takes place has come from findings that word learning impedes phonetic learning, especially when involving minimal pairs (Feldman, Griffiths, Goldwater, & Morgan, 2013; Hayes-Harb & Masuda, 2008). This advantage for distributional learning over minimal pair learning, however, has not been demonstrated with regard to generalizing to untrained, analogous contexts. To investigate the role of the lexicon in L2 phonetic learning, the experiment in this dissertation trained participants on an artificial language with a different VOT category boundary than that of their first language, English. The experiment featured a between-subjects design in which participants were exposed to one of two training conditions. In the explicit, minimal pairbased condition, participants learned fine-grained VOT differences through voiced-voiceless minimal pairs (bilsu/pilsu) that illustrated a VOT category boundary of around 0 ms, providing explicit evidence for the phones' contrasting phonological statuses. The implicit condition served to determine whether distributional learning could take place without explicit information from minimal pairs. The implicit condition's lexicon contained no minimal pairs to illustrate a direct meaningful relationship between voiced/voiceless pairs of phones (binsu/pilsu). Lexical learning

Frontiers in psychology, 2014

1970

This paper takes issue with the position that children's phoneme acquisition schedule is dictated primarily by auditory perceptual factors and suggests the alternative position that ease of production accounts for age of acquisition. It is felt that perceptual theory cannot adequately explain phonological development, e.g. three-year-olds produce certain sounds which they will not accurately perceive until much later and vice versa. Three psychological scaling tests are described in which adults were asked to designate certain phonemes as harder or easier to produce. These judgments were compared with the phonemes acquired by three-year-olds. A highly significant correlation between the adult ratings and children's phoneme acquisition was found. Featural analysis tended to support the hypothesis, and further confirmation was seen in the patterning of articulatory errors among children. The author stresses that these findings do not negate the importance of perceptual factors, emphasizing that he is trying to explain not the dynamics of phoneme acquisition but rather the schedule which it follows. It is, however, emphasized that there is no compelling evidence for perceptual theory, and certain observations which either militate against perceptual theory or encourage alternative speculation are discussed. (FWB)

Perspectives on Behavior Science, 2023

Language researchers have historically either dismissed or ignored completely behavioral accounts of language acquisition while at the same time acknowledging the important role of experience in language learning. Many language researchers have also moved away from theories based on an innate generative universal gram- mar and promoted experience-dependent and usage-based theories of language. These theories suggest that hearing and using language in its context is critical for learning language. However, rather than appealing to empirically derived princi- ples to explain the learning, these theories appeal to inferred cognitive mechanisms. In this article, I describe a usage-based theory of language acquisition as a recent example of a more general cognitive linguistic theory and note both logical and methodological problems. I then present a behavior-analytic theory of speech per- ception and production and contrast it with cognitive theories. Even though some researchers acknowledge the role of social feedback (they rarely call it reinforce- ment) in vocal learning, they omit the important role played by automatic reinforce- ment. I conclude by describing automatic reinforcement as the missing link in a par- simonious account of vocal development in human infants and making comparisons to vocal development in songbirds.