We present a model and an experimental platform of a bootstrapping approach to statistical induct... more We present a model and an experimental platform of a bootstrapping approach to statistical induction of natural language properties that is constraint based with voting components. The system is incremental and unsupervised. In the following discussion we focus on the components for morphological induction. We show that the much harder problem of incremental unsupervised morphological induction can outperform comparable all-at-once algorithms with respect to precision. We discuss how we use such systems to identify cues for induction in a cross-level architecture.
This dissertation is a comprehensive examination of L1 Japanese particle acquisition using two se... more This dissertation is a comprehensive examination of L1 Japanese particle acquisition using two sets of corpora of naturalistic speech, JCHAT, longitudinal speech data of three Japanese children and the mothers (Miyata 2004a, b, and c, and MacWhinney 2000), and CHJ, adult-to-adult telephone conversation. The analysis reveals that despite differences in their language environment, all three children complete particle acquisition around MLU (m) 3.00 regardless of differences in the threshold of particle use. Further, none of the three children mimicked their mother's frequency of particle use, even as they all conform to a particular sequence of particle acquisition. The first set of particles that appear in children's utterances are speech acts related particles, basic Sentence Final Particles (such as-ne and-yo) and the nominative marker-ga and topic marker-wa. Under Fiengo and McClure's (2002) adaptation of Austin's (1953) Speech Act Theory,-ga and-wa are viewed as pragmatic "direction of fit" indicators. The data reveals that although both-ga and-wa can mark the subject of a sentence, children initially hypothesize an association between-ga andwa and certain kinds of speech acts. Specifically,-ga is linked to statements and-wa to questions. The remaining particles (such as case markers,-o and-ni, and other particles,-ni,-de,-kara, and-made) appear after-ga and-wa in accordance with the v development of cognitive ability, memory, lexicon, and syntax. Two trends were found in the general acquisition sequence of these particles: (i) monosyllabic particles are learned earlier (more specifically, particles such as-to and-mo that form complex NPs precede case markers and postpositions, which are more complex because they convey relational functions within the sentence) than di-syllabic particles and (ii) for ambiguous particles such as-ni and-kara, concrete usages are acquired before conceptually abstract ones. vi Acknowledgements I have always considered myself a pessimist. Yet recently, I've begun to think that maybe I have been just blind. Maybe I am getting wiser at last! I had rarely realized it at the time, but whenever I felt lost, I was always fortunate enough to meet some superb individual who would lead me to back to the right path. These people not only supported me in my intellectual growth, but also on a personal level. I owe my sincerest gratitude to each and every one of them. Here are just a few examples. I vividly remember Helen Cairns' psycholinguistics class my sophomore year of college, which led to such a sense of exhilaration. When I was pondering the idea of going to a graduate school, Mike Hall said, "Mari, I'll write your recommendation. Tell me all about yourself!" with his big smile. (I had this extraordinary, long letter of recommendation a couple weeks later). Chuck Cairns once sat me down in his office and told me when I had a serious academic setback, that all I have to do is to get rid of the "big ugly albatross" around my neck. If he didn't say in that way that I very well may have quit school at that point. Janet Fodor and Di Bradley's psycholinguistics seminar planted more than just a seed of a dissertation topic in my head. Martin Chodorow, oh so patiently, went over the data analysis for my dissertation. And I've been extremely lucky to have the opportunity to teach Japanese under the supervision of William McClure at Queens College. Eventually with Bill's intellectual guidance and support, I was able finally come to the point of finishing my dissertation. Bill, I am so grateful. My family in the States, the Sakas', have provided great moral support from the get go. They welcomed me, and accepted me into the family. I am really lucky to have you all. And my mother, thank you so much for believing in me and giving me the many many words of encouragement. And my father, I hope I made you proud. And I hope vii you know that I am grateful for all you have done for me. I am also grateful for the best wishes my brother, aunts, and uncles always have given me. And last but not least, to my cheerleader, best friend, and husband, William, there are no words that could describe my appreciation. I am truly thankful that you let me be who I am, and supported me no matter what-even when I was stressed out and cranky. And to Bailee and Momoji, thank you, my faithful poochies-your loud snoring from under my desk always brought a smile to my face even when I was frustrated with my dissertation writing in the middle of the night. Also, Cookie and Ally, my doggy angels, I won't forget the smiles you've given me over the many years we had together. viii Table Contents ABSTRACT .
Uploads
Papers by Mari Fujimoto