Variability and Detection of Invariant Structure

Abstract

Two experiments investigated learning of nonadjacent dependencies by adults and 18-month-olds. Each learner was exposed to three-element strings (e.g., pel-kicey-jic) produced by one of two artificial languages. Both languages contained the same adjacent dependencies, so learners could distinguish the languages only by acquiring dependencies between the first and third elements (the nonadjacent dependencies). The size of the pool from which the middle elements were drawn was systematically varied to investigate whether increasing variability (in theform of decreasing predictability between adjacent elements) would lead to better detection of nonadjacent dependencies. Infants and adults acquired nonadjacent dependencies only when adjacent dependencies were least predictable. The results point to conditions that might lead learners to focus on nonadjacent versus adjacent dependencies and are importantfor suggesting how learning might be dynamically guided by statistical structure.

Do as I say, not as I do: A lexical distributional account of English locative verb class acquisition

“…Other verb bias alternations such as the noun phrase/sentence complement alternation (e.g., the man wrote [that] the story [was interesting]; Garnsey, Pearlmutter, Myers & Lotocky, 1997) also involve variability in the post-verbal position (e.g., optional that), suggesting that such variation may play a role in verb class acquisition more generally. Humans also find it difficult to learn non-adjacent dependencies in artificial grammar learning tasks and variation can modulate this ability (Gómez, 2002;Gómez & Maye, 2005;Newport & Aslin, 2004). This is particularly relevant given that SRN models provide some of the best accounts of learning in these tasks (Chang, Janciauskas & Fitz, 2012).…”

What's statistical about learning? Insights from modelling statistical learning as a set of memory processes

“…Learners are sensitive not just to conditional relations among adjacent elements, but also to relations among non-adjacent elements (e.g. [85,86]). That is, when presented with sequences like AXB, AYB and AZB, learners detect the relationship between A and B even though these items are not directly adjacent.…”

Rapid generalization in phonotactic learning

“…We believe, however, that it is unlikely that participants remembered a significant fraction of the exposure words. Words were never repeated more than once in exposure; the high variability of the vowel patterns (and therefore the particular words) is likely to have encouraged learning of the consonant patterns rather than learning of particular words (Gómez 2002). Indeed, although it is probably more difficult to remember 64 different words (in the Eight Sets group) than eight different words (in the One Set group), participants in the Eight Sets group showed better learning outcomes than those in the One Set group.…”