Colin Wilson

A Maximum Entropy Model of Phonotactics and Phonotactic Learning

The study of phonotactics is a central topic in phonology. We propose a theory of phonotactic grammars and a learning algorithm that constructs such grammars from positive evidence. Our grammars consist of constraints that are assigned numerical weights according to the principle of maximum entropy. The grammars assess possible words on the basis of the weighted sum of their constraint violations. The learning algorithm yields grammars that can capture both categorical and gradient phonotactic patterns. The algorithm is not provided with constraints in advance, but uses its own resources to form constraints and weight them. A baseline model, in which Universal Grammar is reduced to a feature set and an SPE-style constraint format, suffices to learn many phonotactic phenomena. In order for the model to learn nonlocal phenomena such as stress and vowel harmony, it must be augmented with autosegmental tiers and metrical grids. Our results thus offer novel, learning-theoretic support for such representations. We apply the model in a variety of learning simulations, showing that the learned grammars capture the distributional generalizations of these languages and accurately predict the findings of a phonotactic experiment.

Learning Phonology With Substantive Bias: An Experimental and Computational Study of Velar Palatalization

There is an active debate within the field of phonology concerning the cognitive status of substantive phonetic factors such as ease of articulation and perceptual distinctiveness. A new framework is proposed in which substance acts as a bias, or prior, on phonological learning. Two experiments tested this framework with a method in which participants are first provided highly impoverished evidence of a new phonological pattern, and then tested on how they extend this pattern to novel contexts and novel sounds. Participants were found to generalize velar palatalization (e.g., the change from [k] as in keep to [t?∫S] as in cheap) in a way that accords with linguistic typology, and that is predicted by a cognitive bias in favor of changes that relate perceptually similar sounds. Velar palatalization was extended from the mid front vowel context (i.e., before [e] as in cape) to the high front vowel context (i.e., before [i] as in keep), but not vice versa. The key explanatory notion of perceptual similarity is quantified with a psychological model of categorization, and the substantively biased framework is formalized as a conditional random field. Implications of these results for the debate on substance, theories of phonological generalization, and the formalization of similarity are discussed.

When is less more? Faithfulness and minimal links in wh-chains

at Colorado; to Joseph Aoun, Luigi Burzio, Robert Frank, Jane Grimshaw, David Pesetsky and Alan Prince, for extremely helpful conversations; to audiences at Arizona, Brown, Cornell, Delaware, Georgetown, Hopkins, Maryland, UCLA, USC, and the MIT conference, for stimulating ideas and questions. For partial financial support, we gratefully acknowledge NSF grant BS-9209265, and then NSF grant IRI-9213894, both to Smolensky and Legendre; and the Center for Language and Speech Processing at Johns Hopkins.

Transparency, strict locality, and targeted constraints

The claim that feature assimilation is strictly local, applying only between adjacent segments, appears to be contradicted by languages in which, descriptively speaking, vowel harmony passes through so-called ‘transparent’ vowels without affecting them. We illustrate this apparent contradiction with the pattern of vowel harmony and transparency found in Wolof (Niger-Congo). As described by Ka (1988), Archangeli & Pulleyblank (1994), and Pulleyblank (1996), among others, this language has a progressive (left-to-right) process of [ ATR] (Advanced Tongue Root) harmony. Thus the vowel of the suffix -nnn surfaces as [– ATR] after a [–ATR] root vowel in (1a), and as [+ ATR] after a [+ATR] root vowel in (1b).

Optimality and Wh-Extraction

The study of wh-question formation has historically served as the empirical basis for major constructs in Government-Binding (GB) such as the Empty Category Principle (ECP), the existence of Logical Form (LF) as a separate level of representation— motivated in part by the abstract wh-movement at LF analysis of wh-in-situ in languages like Chinese (Huang, 1982)—and the central but controversial issue of which principles apply at which levels of representation. For example, Huang (1982) argues, based on Chinese, that the ECP applies at S-structure and LF while subjacency and his Condition on Extraction Domain (CED) apply only at S-structure.

Stimulus-specific variability in color working memory with delayed estimation

Working memory for color has been the central focus in an ongoing debate concerning the structure and limits of visual working memory. Within this area, the delayed estimation task has played a key role. An implicit assumption in color working memory research generally, and delayed estimation in particular, is that the fidelity of memory does not depend on color value (and, relatedly, that experimental colors have been sampled homogeneously with respect to discriminability). This assumption is reflected in the common practice of collapsing across trials with different target colors when estimating memory precision and other model parameters. Here we investigated whether or not this assumption is secure. To do so, we conducted delayed estimation experiments following standard practice with a memory load of one. We discovered that different target colors evoked response distributions that differed widely in dispersion and that these stimulus-specific response properties were correlated across observers. Subsequent experiments demonstrated that stimulus-specific responses persist under higher memory loads and that at least part of the specificity arises in perception and is eventually propagated to working memory. Posthoc stimulus measurement revealed that rendered stimuli differed from nominal stimuli in both chromaticity and luminance. We discuss the implications of these deviations for both our results and those from other working memory studies.

Cognitive Biases, Linguistic Universals, and Constraint‐Based Grammar Learning

According to classical arguments, language learning is both facilitated and constrained by cognitive biases. These biases are reflected in linguistic typology-the distribution of linguistic patterns across the worlds languages-and can be probed with artificial grammar experiments on child and adult learners. Beginning with a widely successful approach to typology (Optimality Theory), and adapting techniques from computational approaches to statistical learning, we develop a Bayesian model of cognitive biases and show that it accounts for the detailed pattern of results of artificial grammar experiments on noun-phrase word order (Culbertson, Smolensky, & Legendre, 2012). Our proposal has several novel properties that distinguish it from prior work in the domains of linguistic theory, computational cognitive science, and machine learning. This study illustrates how ideas from these domains can be synthesized into a model of language learning in which biases range in strength from hard (absolute) to soft (statistical), and in which language-specific and domain-general biases combine to account for data from the macro-level scale of typological distribution to the micro-level scale of learning by individuals.

Prosodic structure and wh-questions.

This study examines the influence of wh-gaps on the prosodic contour of spoken utterances. A previous study (Nagel, Shapiro, & Nawy, 1994) claimed that the phonological representation of a sentence containing a filler-gap dependency explicitly encodes the location of the syntactic gap. In support of this hypothesis, Nagel et al. presented evidence that the word immediately preceding a gap is lengthened and that there is a reliable increase in pitch excursion across the gap location. Our study challenges Nagel et al.s claim. We argue that their materials confounded the presence/absence of a gap with other factors that are known to affect intonational phrasing independently. We show that, when these factors are separated, the evidence that syntactic gaps are explicitly encoded in the phonological representation of a sentence disappears.

Burst spectrum as a cue for the stop voicing contrast in American English

Voicing contrasts in stop consonants are expressed by a constellation of acoustic cues. This study focused on a spectral cue present at burst onset in American English labial and coronal stops. Spectral shape was examined for word-initial, prevocalic stops of all three places of articulation in a laboratory production study and a large corpus of continuous read speech. Voiceless labial and coronal stops were found to have greater energy at higher frequencies in comparison to homorganic voiced stops, a difference that could not be attributed to aspiration in the voiceless stops or modal phonation in the voiced, while no consistent effect was found for dorsal stops. This pattern was found with various methods of spectral estimation (time-averaged and multitaper spectra) and measures of spectral energy concentration (center of gravity and spectral peak) for both linear and auditorily based frequency scales. Perceptual relevance of the spectral cue was tested in laboratory and online experiments with continua created by crossing burst shape and voice onset time. A trading relation was observed such that voiceless identifications were more likely for tokens with higher frequency bursts. Goodness ratings indicated that burst spectrum influences category typicality for voiceless stops even when voice onset time is unambiguous.

Structure in talker-specific phonetic realization: Covariation of stop consonant VOT in American English

Abstract Variation across talkers in the acoustic-phonetic realization of speech sounds is a pervasive property of spoken language. The present study provides evidence that variation across talkers in the realization of American English stop consonants is highly structured. Positive voice onset time (VOT) was examined for all six word-initial stop categories in isolated productions of CVC syllables and in a multi-talker corpus of connected read speech. The mean VOT for each stop differed considerably across talkers, replicating previous findings, but importantly there were strong and statistically significant linear relations among the means (e.g., the mean VOTs of [pʰ] and [kʰ] were highly correlated across talkers, r>0.80). The pattern of VOT covariation was not reducible to differences in speaking rate or other factors known to affect the realization of stop consonants. These findings support a uniformity constraint on the talker-specific realization of a phonetic property, such as glottal spreading, that is shared by multiple speech sounds. Because uniformity implies mutual predictability, the findings also shed light on listeners׳ ability to generalize knowledge of a novel talker from one stop consonant to another. More broadly, structured variation of the kind investigated here indicates a relatively low-dimensional encoding of talker-specific phonetic realization in both speech production and speech perception.