Constraints on statistical computations at 10 months of age: the use of phonological features

Recently, several studies have argued that infants capitalize on the statistical properties of natural languages to acquire the linguistic structure of their native language, but the kinds of constraints which apply to statistical computations remain largely unknown. Here we explored French‐learning infants' perceptual preference for labial‐coronal (LC) words over coronal‐labial words (CL) words (e.g. preferring bat over tab) to determine whether this phonotactic preference is based on the acquisition of the statistical properties of the input based on a single phonological feature (i.e. place of articulation), multiple features (i.e. place and manner of articulation), or individual consonant pairs. Results from four experiments revealed that infants had a labial‐coronal bias for nasal sequences (Experiment 1) and for all plosive sequences (Experiments 2 and 4) but a coronal‐labial bias for all fricative sequences (Experiments 3 and 4), independently of the frequencies of individual consonant pairs. These results establish for the first time that constellations of multiple phonological features, defining broad consonant classes, constrain the early acquisition of phonotactic regularities of the native language.     

Implementations are not conceptualizations: revising the verb learning model.

In a recent issue of this journal, Pinker and Prince (1988) and Lachter and Bever (1988) presented detailed critiques of Rumelhart and McClelland's (1986) connectionist model of the child's learning of the phonological form of the English past tense. In order to address these criticisms, a new connectionist model was constructed using the back-propagation algorithm, a larger input corpus, a fuller paradigm, and a new phonological representation. This new implementation successfully addressed the criticisms of the phonological representation used by Rumelhart and McClelland. It did a much better job of learning the past tense using a fuller input set with realistic frequencies of occurrence. Ancillary simulations using the same network were able to deal with the homonymy problem and the generation of forms like "ated" from "ate". The one feature not provided by the new model was a way of modeling early correct production of irregular forms. The success of the new model can be used to help clarify the extent to which the published critiques apply to a particular connectionist implementation as opposed to fundamental principles underlying the broader connectionist conceptualization.     

Learning words’ sounds before learning how words sound: 9-Month-olds use distinct objects as cues to categorize speech information

One of the central themes in the study of language acquisition is the gap between the linguistic knowledge that learners demonstrate, and the apparent inadequacy of linguistic input to support induction of this knowledge. One of the first linguistic abilities in the course of development to exemplify this problem is in speech perception: specifically, learning the sound system of one’s native language. Native-language sound systems are defined by meaningful contrasts among words in a language, yet infants learn these sound patterns before any significant numbers of words are acquired. Previous approaches to this learning problem have suggested that infants can learn phonetic categories from statistical analysis of auditory input, without regard to word referents. Experimental evidence presented here suggests instead that young infants can use visual cues present in word-labeling situations to categorize phonetic information. In Experiment 1, 9-month-old English-learning infants failed to discriminate two non-native phonetic categories, establishing baseline performance in a perceptual discrimination task. In Experiment 2, these infants succeeded at discrimination after watching contrasting visual cues (i.e., videos of two novel objects) paired consistently with the two non-native phonetic categories. In Experiment 3, these infants failed at discrimination after watching the same visual cues, but paired inconsistently with the two phonetic categories. At an age before which memory of word labels is demonstrated in the laboratory, 9-month-old infants use contrastive pairings between objects and sounds to influence their phonetic sensitivity. Phonetic learning may have a more functional basis than previous statistical learning mechanisms assume: infants may use cross-modal associations inherent in social contexts to learn native-language phonetic categories.     

Psycholinguistic evidence for the underspecification of morphosyntactic features

This paper investigates the paradigmatic relations between inflected word forms (or their affixes) and the feature specifications of these elements. In two sentence-matching experiments German speakers had to decide whether sentence pairs involving inflected adjectives or determiners were identical or not. In both experiments, there was a delay when an inflected form contained positive feature specifications for grammatical features that did not match the feature specifications of the grammatical context in which it appeared. No delay, however, occurred when an incorrectly inflected form had mismatching negative specifications, whereas its positively specified features matched the respective positive features of the context. This result provides evidence for a different status of positively and negatively specified morphosyntactic features. It supports the idea of radical underspecification according to which only positive feature specifications are part of the representations of morphologically complex forms or affixes, whereas negative feature specifications are assigned on the basis of paradigmatic contrasts.     

Phonemes matter: the role of phoneme-level awareness in emergent Chinese readers

The importance of phonological awareness for learning to read may depend on the linguistic properties of a language. This study provides a careful examination of this language-specific theory by exploring the role of phoneme-level awareness in Mandarin Chinese, a language with an orthography that, at its surface, appears to require little phoneme-level insight. A sample of 71 monolingual Mandarin-speaking children completed a phonological elision task and a measure of single-character reading. In this sample, 4- and 5-year-old preschoolers were unable to complete phoneme-level deletions, whereas 6- to 8-year-old first graders were able to complete initial, final, and medial phoneme-level deletions. In this older group, performance on phoneme deletions was significantly related to reading ability even after controlling for syllable- and onset/rime-level awareness, vocabulary, and Pinyin knowledge. We believe that these results reopen the question of the role of phonological awareness in reading in Chinese and, more generally, the nature of the mechanisms underlying this relationship.     

Perception of synthetic nasal consonants in initial and final syllable position

Identification and discrimination of synthesized syllable-initial and syllable-final nasal consonants (/mæ-næ-ηæ/ and æm-æn-æη) by adult American subjects were assessed to determine (1) whether place-of-articulation contrasts in nasals, cued only by second and third formant transition variations, were perceived categorically, and (2) if linguistic experience affected_ the perception of this acoustic dimension. In two experiments, subjects produced consistent identification functions with sharp boundaries between familiar phoneme categories. Corresponding discrimination functions showed “peaks” of relatively accurate perception for cross-category comparison pairs, indicating categorical perception. Identification consistency and discrimination accuracy were inferior for the/n/-/η/ contrast in the unfamiliar (and phonologically inappropriate) syllable-initial condition compared to the familiar syllable-final condition. No such difference was found in identification and discrimination of the acoustically comparable oral stop consonant contrast/d/-/g/in syllable-initial and syllable-final position. These results provide evidence that perception of linguistically relevant acoustic dimensions by adults is constrained, at least in part, by their familiarity with those acoustic (and phonetic) contrasts in specific phonological contexts.     

Vowel bias in Danish word‐learning: processing biases are language‐specific

The present study explored whether the phonological bias favoring consonants found in French‐learning infants and children when learning new words (Havy & Nazzi, 2009; Nazzi, 2005) is language‐general, as proposed by Nespor, Peña and Mehler (2003), or varies across languages, perhaps as a function of the phonological or lexical properties of the language in acquisition. To do so, we used the interactive word‐learning task set up by Havy and Nazzi (2009), teaching Danish‐learning 20‐month‐olds pairs of phonetically similar words that contrasted either on one of their consonants or one of their vowels, by either one or two phonological features. Danish was chosen because it has more vowels than consonants, and is characterized by extensive consonant lenition. Both phenomena could disfavor a consonant bias. Evidence of word‐learning was found only for vocalic information, irrespective of whether one or two phonological features were changed. The implication of these findings is that the phonological biases found in early lexical processing are not language‐general but develop during language acquisition, depending on the phonological or lexical properties of the native language.     

Rules, representations, and the English past tense

The significance of the English past tense in current cognitive science is that it offers a clear contrast between a potentially rule-based system-the procedures for forming the regular past tense-and an unpredictable and idiosyncratic set of irregular forms. This contrast has become a focus for a wide-ranging debate about whether mental computation requires the use of symbols. Highly regular combinatorial phenomena, such as the regular past tense, are prime candidates for rule-based symbolic computation. Earlier research concentrated on the evidence for this during language acquisition, looking at how children learned the English regular and irregular verb systems. Over the last five years attention has shifted towards the properties of the adult system, and we review here some recent research into the neural correlates of the two types of procedure. The evidence suggests that there are divergences in the neural systems underlying the generation and perception of regular and irregular forms. Regular inflected forms seem to involve primarily combinatorial processes, while irregular forms appear to have a hybrid status, sharing their semantic properties with the regular forms but diverging in the phonological domain, where their form representations are stored as complete units. This indicates that the regular and irregular past tenses may not, after all, provide a clean contrast in the types of mental computation they implicate.     

Perception Abilities of L1 Cypriot Greek Listeners - Types of Errors Involving Plosive Consonants in L2 English

This paper investigates the difficulties adult second language (L2) users of English encounter with plosive consonants in the L2. It presents the results of a task examining the acquisition of plosive voicing contrasts by college students with Cypriot Greek (CG) linguistic background. The task focused on the types of errors involving plosive consonants indicating that performance was significantly better in the voiceless plosive category. Participants were able to perceive voiced plosives but they treated such instances as a /nasal + voiced plosive/ sequence. Therefore, the question raised concerns different phonological contrasts realised through similar phonetic cues. The patterns observed suggested that this gap between phonetic cues and phonological contrast might explain why CG users have difficulties perceiving voiced English plosives. In this context, voice onset time (VOT) differences between the L1 and L2 are of crucial importance. In English, voiced plosives are characterised by short lag VOT while their voiceless counterparts fall within the long lag VOT continuum. The same phonetic contrast is used in CG to differentiate between single and geminate voiceless plosives. The results are discussed in relation to the frameworks of second language phonology and speech perception suggesting that the difficulties faced by the L2 listeners support the operation of a phonetic-phonological challenge.     

iMinerva: A Mathematical Model of Distributional Statistical Learning

Statistical learning refers to the ability to identify structure in the input based on its statistical properties. For many linguistic structures, the relevant statistical features are distributional: They are related to the frequency and variability of exemplars in the input. These distributional regularities have been suggested to play a role in many different aspects of language learning, including phonetic categories, using phonemic distinctions in word learning, and discovering non‐adjacent relations. On the surface, these different aspects share few commonalities. Despite this, we demonstrate that the same computational framework can account for learning in all of these tasks. These results support two conclusions. The first is that much, and perhaps all, of distributional statistical learning can be explained by the same underlying set of processes. The second is that some aspects of language can be learned due to domain‐general characteristics of memory.     

Vowel categorization during word recognition in bilingual toddlers

Toddlers’ and preschoolers’ knowledge of the phonological forms of words was tested in Spanish-learning, Catalan-learning, and bilingual children. These populations are of particular interest because of differences in the Spanish and Catalan vowel systems: Catalan has two vowels in a phonetic region where Spanish has only one. The proximity of the Spanish vowel to the Catalan ones might pose special learning problems. Children were shown picture pairs; the target picture’s name was spoken correctly, or a vowel in the target word was altered. Altered vowels either contrasted with the usual vowel in Spanish and Catalan, or only in Catalan. Children’s looking to the target picture was used as a measure of word recognition. Monolinguals’ word recognition was hindered by within-language, but not non-native, vowel changes. Surprisingly, bilingual toddlers did not show sensitivity to changes in vowels contrastive only in Catalan. Among preschoolers, Catalan-dominant bilinguals but not Spanish-dominant bilinguals revealed mispronunciation sensitivity for the Catalan-only contrast. These studies reveal monolingual children’s robust knowledge of native-language vowel categories in words, and show that bilingual children whose two languages contain phonetically overlapping vowel categories may not treat those categories as separate in language comprehension.     

A neuronal model for syllable representation

A speculative neuronal template, equivalent to canonical syllable forms and independent of segmental representations, is offered to help account for (1) the inviolate nature of phonotactic constraints in aphasic speech output, and (2) left hemisphere specialization for speech sound access and output. The model, which attempts to relate plausible neuronal systems to linguistic function, is based on cell assemblies that are thought to develop by way of genetic predisposition and ontogenetic language experience, into configurations that can represent canonical slot positions for the consonants and vowel comprising a syllable. The syllable is assumed to be the basic organizational rhythmic unit for serial concatenation of sublexical segments. A scheme for neurological differentiation of vowels and consonants is offered. Phonotactic constraints can become "hard-wired" to help create the automaticity underlying phonological sound organization. Testable predictions are offered to substantiate the claims of the model.     

Learning General Phonological Rules From Distributional Information: A Computational Model

Phonological rules create alternations in the phonetic realizations of related words. These rules must be learned by infants in order to identify the phonological inventory, the morphological structure, and the lexicon of a language. Recent work proposes a computational model for the learning of one kind of phonological alternation, allophony (Peperkamp, Le Calvez, Nadal, & Dupoux, 2006). This paper extends the model to account for learning of a broader set of phonological alternations and the formalization of these alternations as general rules. In Experiment 1, we apply the original model to new data in Dutch and demonstrate its limitations in learning nonallophonic rules. In Experiment 2, we extend the model to allow it to learn general rules for alternations that apply to a class of segments. In Experiment 3, the model is further extended to allow for generalization by context; we argue that this generalization must be constrained by linguistic principles.     

A Connectionist Model of Phonological Representation in Speech Perception

A number of recent studies have examined the effects of phonological variation on the perception of speech. These studies show that both the lexical representations of words and the mechanisms of lexical access are organized so that natural, systematic variation is tolerated by the perceptual system, while a general intolerance of random deviation is maintained. Lexical abstraction distinguishes between phonetic features that form the invariant core of a word and those that are susceptible to variation. Phonological inference relies on the context of surface changes to retrieve the underlying phonological form. In this article we present a model of these processes in speech perception, based on connectionist learning techniques. A simple recurrent network was trained on the mapping from the variant surface form of speech to the underlying form. Once trained, the network exhibited features of both abstraction and inference in its processing of normal speech, and predicted that similar behavior will be found in the perception of nonsense words. This prediction was confirmed in subsequent research (Gaskell & Marslen-Wilson, 1994).     

Biased generalization of newly learned phonological alternations by 12-month-old infants

Previous work has suggested that learners are sensitive to phonetic similarity when learning phonological patterns (e.g.,  and ). We tested 12-month-old infants to see if their willingness to generalize newly learned phonological alternations depended on the phonetic similarity of the sounds involved. Infants were exposed to words in an artificial language whose distributions provided evidence for a phonological alternation between two relatively dissimilar sounds ([p ∼ v] or [t ∼ z]). Sounds at one place of articulation (labials or coronals) alternated whereas sounds at the other place of articulation were contrastive. At test, infants generalized the alternation learned during exposure to pairs of sounds that were more similar ([b ∼ v] or [d ∼ z]). Infants in a control group instead learned an alternation between similar sounds ([b ∼ v] or [d ∼ z]). When tested on dissimilar pairs of sounds ([p ∼ v] or [t ∼ z]), the control group did not generalize their learning to the novel sounds. The results are consistent with a learning bias favoring alternations between similar sounds over alternations between dissimilar sounds.     

From melody to lexical tone: Musical ability enhances specific aspects of foreign language perception

Previous research shows that music ability provides positive effects on language processing. This study aims at better clarifying the involvement of different linguistic subdomains in this cross-domain link, assessing whether or not musicality and music expertise enhance phonological and lexical tone processing of Mandarin Chinese. In two experiments different groups of adults and children with no previous experience in tonal languages, were invited to perform a same–different task trying to detect phonological and tonal variations in pairs of sequences of monosyllabic Mandarin Chinese words. Main results show that all subjects perform significantly better in detecting phonological variations rather than tonal ones. They also show that both melodic proficiency and music expertise are good predictors for a better tonal, but not phonological identification. Data lead to a model of music-to-language transfer effect in which musicality selectively affects linguistic intonation while leaving phonological processing substantially unaffected.     

Validity of the salience asymmetry interpretation of the implicit association test: comment on Rothermund and Wentura (2004)

The Implicit Association Test (IAT) requires responding to category contrasts such as young versus old, male versus female, and pleasant versus unpleasant. In introducing the IAT, A. G. Greenwald, D. E. McGhee, and J. L. K. Schwartz (1998) proposed that IAT measures reflect mental structures involving the nominal features of the IAT's categories (e.g., age, gender, or valence features). In contrast, K. Rothermund and D. Wentura proposed that IAT performance is dominated by salience asymmetries of the IAT's pairs of contrasted categories. To assess relative contributions of nominal feature contrasts versus salience asymmetries, the authors (a) briefly summarize the extensive evidence now available to support construct validity of the IAT as a measure based on nominal category features and (b) present 2 new experiments that yielded results problematic for the salience asymmetry interpretation.     

Learning words and learning sounds: Advances in language development

Phonological development is sometimes seen as a process of learning sounds, or forming phonological categories, and then combining sounds to build words, with the evidence taken largely from studies demonstrating ‘perceptual narrowing’ in infant speech perception over the first year of life. In contrast, studies of early word production have long provided evidence that holistic word learning may precede the formation of phonological categories. In that account, children begin by matching their existing vocal patterns to adult words, with knowledge of the phonological system emerging from the network of related word forms. Here I review evidence from production and then consider how the implicit and explicit learning mechanisms assumed by the complementary memory systems model might be understood as reconciling the two approaches.     

Phonological processing and lexical access in aphasia

This study explored the relationship between on-line processing of phonological information and lexical access in aphasic patients. A lexical decision paradigm was used in which subjects were presented auditorily with pairs of words or word-like stimuli and were asked to make a lexical decision about the second stimulus in the pair. The initial phonemes of the first word primes, which were semantically related to the real word targets, were systematically changed by one or more than one phonetic feature, e.g., cat-dog, gat-dog, wat-dog. Each of these priming conditions was compared to an unrelated word baseline condition, e.g., nurse-dog. Previous work with normals showed that even a nonword stimulus receives a lexical interpretation if it shares a sufficient number of phonetic features with an actual word in the listener's lexicon. Results indicated a monotonically decreasing degree of facilitation as a function of phonological distortion. In contrast, fluent aphasics showed priming in all phonological distortion conditions relative to the unrelated word baseline. Nonfluent aphasics showed priming only in the undistorted, related word condition relative to the unrelated word baseline. Nevertheless, in a secondary task requiring patients to make a lexical decision on the nonword primes presented singly, all aphasics showed phonological feature sensitivity. These results suggest deficits for aphasic patients in the various processes contributing to lexical access, rather than impairments at the level of lexical organization or phonological organization.