An empirical generative framework for computational modeling of language acquisition

Abstract

A B S T R A C TThis paper reports progress in developing a computer model of language acquisition in the form of (1) a generative grammar that is (2) algorithmically learnable from realistic corpus data, (3) viable in its large-scale quantitative performance and (4) psychologically real. First, we describe new algorithmic methods for unsupervised learning of generative grammars from raw CHILDES data and give an account of the generative performance of the acquired grammars. Next, we summarize findings from recent longitudinal and experimental work that suggests how certain statistically prominent structural properties of child-directed speech may facilitate language acquisition. We then present a series of new analyses of CHILDES data indicating that the desired properties [*] During the preparation of this paper, Shimon

The complexity of narrative interferes in the use of conjunctions in children with specific language impairment

“…Therefore, the comprehension of utterances with more elaborate sentence structure requires mastery of this ability associated with enough working memory resources (26) . It is interesting to note that according to some recent researches seeking to understand SLI grammatical impairment by probabilistic knowledge, memory and comprehension impairments in SLI may favor these children to have restricted access to more elaborate linguistic stimuli, considerably reducing their chances to learn language (27,28) . On the other hand, the influence of stimuli complexity in the type of conjunction used was clear and more pronounced in the SLI group.…”

Measuring Mental Entrenchment of Phrases with Perceptual Identification, Familiarity Ratings, and Corpus Frequency Statistics

“…ADIOS has proven to be effective at grammar induction (see Waterfall, Sandbank, Onnis, and Edelman 2010), but still unknown is how closely algorithms like AIDOS match behavioral data on frequency effects. It may may be premature at present to model behavioral data because of lack of knowledge about boundary conditions.…”

“…That is, we don't know when frequency effects stop appearing, but ADIOS suggests some testable possibilities. An interesting aspect of ADIOS for frequency effects in multiword utterances is that a sequence can have low global usage frequency, yet stand out in a circumscribed set of contexts and become entrenched and unitized (Waterfall, Sandbank, Onnis, and Edelman 2010). In the current study we thus looked for frequency effects using low frequency collocations.…”

Proceedings of the 7th Workshop on Cognitive Aspects of Computational Language Learning

“…Using an automated procedure of this kind, Brodsky et al obtain a proportion of 21.5% of the words in Waterfall's (2006) corpus occurring in variation sets, and 18.3% of the words in the English CHILDES collection . Similar studies have been performed by Onnis et al (2008) and Waterfall et al (2010). For example, when Onnis et al used an automated procedure based on Waterfall's (2006) criteria on the Lara corpus from CHILDES (involving one child between 1;9 and 3;3 years), they obtained a proportion of 27,9% of the utterances being inside variation sets.…”