Search

Several studies found that word duration is predicted by its informativity (average past predictability) above and beyond predictability in the current context, suggesting retrieval of phonetically-specific tokens from memory. We show that a significant effect of informativity can emerge from noise in predictability estimates. We generate durations from a model in which 38% of log duration is predicted by log probability, as in our actual data, but the rest is normally-distributed noise. Estimated probability for each word in each context is then generated from a binomial distribution with success probability from the real sample and size matching context frequency. We compute informativity and fit the regression model we fit to the real data. Informativity is significant in 100% of simulations, even though probability is the only true predictor, although the effect of informativity is smaller in simulation (0.7 < b < 0.10) than in the actual corpus (b = 0.12).

There is evidence that the process of executing a plannedutterance involves the use of both preceding-context and top-down cues. Utterance-initial words are cued only by the top-down plan. In contrast, non-initial words are cued both bytop-down cues and preceding-context cues. Co-existence ofboth cue types raises the question of how they interact duringlearning. We argue that this interaction is competitive: itemsthat tend to be preceded by predictive preceding-context cuesare harder to activate from the plan without this predictivecontext. A novel computational model of this competition isdeveloped. The model is tested on a corpus of repetitiondisfluencies and shown to account for the influences onpatterns of restarts during production. In particular, this modelpredicts a novel Initiation Effect: following an interruption,speakers re-initiate production from words that tend to occurin utterance-initial position, even when they are not initial inthe interrupted utterance.

In production, frequently used words are preferentiallyextended to new, though related meanings. In comprehension,frequent exposure to a word instead makes the learnerconfident that all of the word’s legitimate uses have beenexperienced, resulting in an entrenched form-meaningmapping between the word and its experienced meaning(s).This results in a perception-production dissociation, where theforms speakers are most likely to map onto a novel meaningare precisely the forms that they believe can never be usedthat way. At first glance, this result challenges the idea ofbidirectional form-meaning mappings, assumed by all currentapproaches to linguistic theory. In this paper, we show thatbidirectional form-meaning mappings are not in factchallenged by this production-perception dissociation. Weshow that the production-perception dissociation is expectedeven if learners of the lexicon acquire simple symmetricalform-meaning associations through simple Hebbian learning.

Several studies examined cue competition in human learningby testing learners on a combination of conflicting cuesrooting for different outcomes, with each cue perfectlypredicting its outcome. A common result has been thatlearners faced with cue conflict choose the outcomeassociated with the rare cue (the Inverse Base Rate Effect,IBRE). Here, we investigate cue competition including IBREwith sentences containing cues to meanings in a visual world.We do not observe IBRE. Instead we find that position in thesentence strongly influences cue salience. Faced with conflictbetween an initial cue and a non-initial cue, learners choosethe outcome associated with the initial cue, whether frequentor rare. However, a frequent configuration of non-initial cuesthat are not sufficiently salient on their own can overcome acompeting salient initial cue rooting for a different meaning.This provides a possible explanation for certain recurringpatterns in language change.

We report on rapid perceptual learning of intonation contour categories in adults and 9- to 11-year-old children. Intonation contours are temporally extended patterns, whose perception requires temporal integration and therefore poses significant working memory challenges. Both children and adults form relatively abstract representations of intonation contours: Previously encountered and novel exemplars are categorized together equally often, as long as distance from the prototype is controlled. However, age-related differences in categorization performance also exist. Given the same experience, adults form narrower categories than children. In addition, adults pay more attention to the end of the contour, while children appear to pay equal attention to the beginning and the end. The age range we examine appears to capture the tail-end of the developmental trajectory for learning intonation contour categories: There is a continuous effect of age on category breadth within the child group, but the oldest children (older than 10;3) are adult-like.

In the 1990s, language acquisition researchers and theoretical linguists developed an interest in learning mechanisms, and learning theorists rediscovered the verbal learning tradition. Nonetheless, learning theory and language acquisition continued to develop largely independently, which has stymied progress in both fields. However, exciting progress is happening in applying learning theory to language, and, more recently, in using language learning data to advance domain‐general learning theory. These developments raise hopes for a bidirectional flow of information between the fields. The importance of language data for learning theory and of learning theory for understanding language is briefly discussed. [ABSTRACT FROM AUTHOR]

Language learning often involves predicting categorical outcomes based on a set of cues. Error in predicting a categorical outcome is the difference between zero or one and the outcome's current level of activation. The current activation level of a categorical outcome is argued to be a non-linear, logistic function of activation the outcome receives from the cues. Crucially, the logistic activation function asymptotically approaches zero and one without ever reaching or overshooting them. This allows error-driven learning to avoid settling on spurious associations between cues and outcomes that never co-occur ("spurious excitement"). In an artificial language experiment, humans are also not observed to show spurious excitement. The logistic activation function is compared to alternative solutions to spurious excitement, and shown to have important advantages. It enables one-shot learning and steep, S-shaped learning curves, and explains why cue competition in language learning can be overcome with additional training. [ABSTRACT FROM AUTHOR]

In spontaneous speech, speakers sometimes replace a word they have just produced or started producing by another word. The present study reports that in these replacement repairs, low-frequency replaced words are more likely to be interrupted prior to completion than high-frequency words, providing support to the hypothesis that the production of high-frequency words is more automatic than the production of low-frequency words (Bybee, 2002; Logan, 1982). Frequency appears to have an effect on interruptibility even when word duration is statistically controlled. In addition, the frequency of the replaced word is positively correlated with the frequency of the word it is replaced by, supporting the hypothesis that high-frequency words are easier to access in production (Kittredge, Dell, Verkuilen, & Schwartz, 2008): the more frequent the target, the more frequent an inappropriate word needs to be to overcome the target and be uttered, only to be replaced. (Contains 1 table, 12 figures, and 14 footnotes.)

This article proposes and tests an experimental method to assess the psychological reality of hierarchical theories of constituent structure in particular domains. I show that a hierarchical theory of constituent structure necessarily makes the prediction that an association between constituents should be easier to learn than an association between strings that cross constituent boundaries, whereas dependency-based theories of constituency predict that constituents cannot have associations that their parts do not have. Previous research has shown that the major division within an English syllable is between the onset and the nucleus, rather than between the nucleus and the coda. Thus, the nucleus forms a constituent (the rime) with the coda rather than the onset. The hierarchical theory of constituency thus proposes that English speakers should learn rime-affix associations more easily than body-affix associations and that the rime can have associations that its parts do not have. These predictions are confirmed in the present study. Applications of the experimental method and its variants to linguistic constituency in other domains are discussed.