St. Clair, Michelle Christina (2007) Language structure and language acquisition: grammatical categorization using phonological and distributional information. PhD thesis, University of York.
Abstract
This thesis addresses the question of how words are grouped according to their
grammatical categories during language acquisition. Over the past 20. years a general
consensus has developed that distributional and phonological cues are important cues that
language learners utilize in the grammatical categorization process (e.g., Kelly, 1992;
Redington, Chater, & Finch, 1998). The combination ofthese cues was investigated with
artificial language learning experiments, which combined two categories of.
phonologically coherent words with co-occurring distributional cues, and corpus analysis
techniques. Experiments I to 4 indicated that both phonological and distribution cues are
necessary for the categorization of high and low frequency words. Additionally, these
experiments indicated that distributional information alone was sufficient to categorize
high frequency words, but that phonological cues were necessary for low frequency
words. It was also found that succeeding bigram distribution cues induced more
grammatical categorization than the preceding bigram cues. This is explained by the
Rescorla-'\yagner (1972) model of associative learning; associations were stronger
between the category words and succeeding cues as a single succeeding cue followed all
category words. Associations were weaker with preceding cues as numerous category
words followed the preceding cues. Experiments 5, 6 and 7 also found that the
effectiveD~ss of the distributional cues was influenced by prior linguistic experience,
resulting in higher learning with distributional cues which were phonologically consistent
with distributional cues found in the participants' native language (English). This thesis
also investigated the debate as to what type of distributional cue is most useful in the
categorization process, with some researchers advocating trigram cues (Mintz, 2002)
while others advocate bigram cues (Monaghan & Christiansen, 2004; Valian & Coulson,
1988). The results of a corpus analysis and two experiments provided evidence that
trigram cues (aXb) are very effective at categorization, but preliminary evidence suggests
that this categorization may simply be due to the combined influence of the beginning
and ending bigrams (aX and Xb). Overall, this thesis indicates that phonological and
distributional cues are key to grammatical categorization, which occurs through
associative learning principles; grammatical categorization progresses .faster with
succeeding cues; and bigram distribution cues may be the initial source of distributional
information in the grammatical categorization process.
Metadata
Awarding institution: | University of York |
---|---|
Academic Units: | The University of York > Psychology (York) |
Identification Number/EthosID: | uk.bl.ethos.485141 |
Depositing User: | EThOS Import (York) |
Date Deposited: | 28 Sep 2015 08:43 |
Last Modified: | 28 Sep 2015 08:43 |
Open Archives Initiative ID (OAI ID): | oai:etheses.whiterose.ac.uk:9943 |
You do not need to contact us to get a copy of this thesis. Please use the 'Download' link(s) above to get a copy.
You can contact us about this thesis. If you need to make a general enquiry, please see the Contact us page.