An incremental model of syntactic bootstrapping

Syntactic bootstrapping is the hypothesis that learners can use the preliminary syntactic structure of a sentence to identify and characterise the meanings of novel verbs. Previous work has shown that syntactic bootstrapping can begin using only a few seed nouns (Connor et al., 2010; Connor et al., 2012). Here, we relax their key assumption: rather than training the model over the entire corpus at once (batch mode), we train the model incrementally, thus more realistically simulating a human learner. We also improve on the verb prediction method by incorporating the assumption that verb assignments are stable over time. We show that, given a high enough number of seed nouns (around 30), an incremental model achieves similar performance to the batch model. We also find that the number of seed nouns shown to be sufficient in the previous work is not sufficient under the more realistic incremental model. The results demonstrate that adopting more realistic assumptions about the early stages of language acquisition can provide new insights without undermining performance.