A connectionist model of the retreat from verb argument structure overgeneralization

Abstract

A central question in language acquisition is how children build linguistic representations that allow them to generalize verbs from one construction to another (e.g., The boy gave a present to the girl → The boy gave the girl a present), whilst appropriately constraining those generalizations to avoid non-adultlike errors (e.g., I said no to her → *I said her no). Although a consensus is emerging that learners solve this problem using both statistical and semantics-based learning procedures (e.g., entrenchment, pre-emption, and semantic verb class formation), there currently exist few - if any - proposals for a learning model that combines these mechanisms. The present study used a connectionist model to test an account that argues for competition between constructions based on (a) verb-in construction frequency, (b) relevance of constructions for the speaker's intended message, and (c) fit between the fine-grained semantic properties of individual verbs and individual constructions. The model was able not only (a) to simulate the overall pattern of overgeneralization-then-retreat, but also (b) to use the semantics of novel verbs to predict their argument structure privileges (just as real learners do), and

Second Language Users’ Restriction of Linguistic Generalization Errors: The Case of English Un– Prefixation Development

“…To be specific, these three mechanisms have been found to play different roles at distinct stages of L1 development. This motivated Ambridge and his colleagues (Ambridge, ; Ambridge & Blything, ; Ambridge, Pine, & Rowland, , ) to propose a hybrid account to explain the generalization‐then‐restriction process during which language speakers gradually retreat from overgeneralization errors as language input increases. The present study tested whether the mechanisms that explain errors in the generalization‐then‐restriction process in L1 learning can also be applied to L2 learning.…”

“…Notes 1 Other accounts, such as Bayesian inference (Alishahi & Stevenson, 2008), lexical distributional cues (Twomey et al, 2014), and a combination of entrenchment, preemption, and verb semantics (Ambridge & Blything, 2016), have been proposed to explain children's rejection or avoidance of overgeneralization errors.…”

“…Considering that the first two of these proposals are not directly relevant to this study, only Ambridge and Blything's (2016) entrenchment + preemption + verb semantics model was targeted here. 2 After reanalyzing his 2013 results with updated mixed-effect models, Ambridge found that entrenchment was no longer effective (B.…”

“…Results show that these mechanisms have differential effects on L1 development of verb argument constructions (Ambridge, Pine, & Rowland, 2011, 2012Ambridge, Pine, Rowland, & Chang, 2012b;Ambridge, Pine, Rowland, Freudenthal, & Chang, 2014;Ambridge, Pine, Rowland, & Young, 2008), which refer to pairings of lexical information about the arguments of a verbal predicate and their semantic and syntactic properties, such as locatives (e.g., Lisa sprayed water onto the flowers) and datives (e.g., Lisa gave the book to Bart), and on L1 word learning, including un-prefixation (Ambridge, 2013;Blything, Ambridge, & Lieven, 2014). Also, the combined effects of both statistical measures (preemption + entrenchment) and verb semantics have been observed for L1 dative development (Ambridge & Blything, 2016). 1 In the present study, I hope to contribute to this line of inquiry by examining whether and to what extent entrenchment, preemption, and verb semantics influence the process by which learners learn to reject or avoid overgeneralization errors in the L2 acquisition of English un-prefixation.…”

“…Other accounts, such as Bayesian inference (Alishahi & Stevenson, ), lexical distributional cues (Twomey et al., ), and a combination of entrenchment, preemption, and verb semantics (Ambridge & Blything, ), have been proposed to explain children's rejection or avoidance of overgeneralization errors. Considering that the first two of these proposals are not directly relevant to this study, only Ambridge and Blything's () entrenchment + preemption + verb semantics model was targeted here.…”