Structure and Substance in Artificial‐Phonology Learning, Part II: Substance

Abstract

Artificial analogues of natural-language phonological patterns can often be learned in the lab from small amounts of training or exposure. The difficulty of a featurallydefined pattern has been hypothesized to be affected by two main factors, its formal structure (the abstract logical relationships between the defining features) and its phonetic substance (the concrete phonetic interpretation of the pattern). This paper, the second of a two-part series, reviews the experimental literature on phonetic substance, which is hypothesized to facilitate the acquisition of phonological patterns that resemble naturally-occurring phonetic patterns. The effects of phonetic substance on pattern learning turn out to be elusive and unreliable in comparison with the robust effects of formal complexity (reviewed in Part I). If natural-language acquisition is guided by the same inductive biases as are found in the lab, these results support a theory in which inductive bias shapes only the form, and channel bias shapes the content, of the sound patterns of the worlds languages.

Evidence for a learning bias against saltatory phonological alternations

“…It has been argued that phonological generalizations are easier to learn when they can be described using fewer features (e.g., Moreton & Pater, 2012). Because examples ending in voiced stops and voiceless fricatives were withheld from training, targeting a more general class of sounds would (a) be equally consistent with the input, (b) require fewer features for grouping the sounds, and (c) explain the extension to untrained sounds in the Potentially Saltatory condition.…”

“…Velar stops are more acoustically and perceptually similar to palato-alveolar affricates before high vowels than before mid vowels (e.g., [ki] and [tʃi] are more similar than where the alternating sounds were more similar to each other. With the support of a computational model, Wilson suggested that phonetic similarity was responsible for the observed asymmetry in degree of generalization (but see Moreton & Pater, 2012, for potential problems with the interpretation of these results).…”

“…There is extensive evidence that humans prefer simple solutions to more complex solutions within a number of cognitive domains, and indeed, the results of many artificial grammar studies showing differences in phonological learning can potentially be attributed to a preference for simpler patterns (for a review, see Moreton & Pater, 2012).…”

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

“…One bias with ample support in the literature is that complex patterns are more difficult to learn, and less readily generalized, than simpler patterns (adults: Pycha et al, 2003;infants: Saffran & Thiessan, 2003;Cristia & Seidl, 2008;Chambers et al, 2011; for an overview see Moreton & Pater, 2012a). A more controversial proposal (see Moreton & Pater, 2012b) is that learners prefer patterns with an underlying phonetic motivation, sometimes called a substantive bias (e.g., Wilson, 2006).…”

Can infants learn phonology in the lab? A meta-analytic answer

“…Admittedly, how and when listeners generalize from a set of sounds used in the exposure to other sounds is an open topic -see e.g., Moreton and Pater (2012) for a recent review of artificial grammar learning work.…”