No abstract

In this note we discuss pMLU, a whole-word measure for phonological development that was proposed by Ingram (2002). Ingram's rules for calculating pMLU are analysed and we point at the crucial role of the level of transcription for making pMLU measurements comparable over different corpora. The main aim of the paper is an assessment of the reliability and the validity of pMLU. The assessment is accomplished using a computational tool for measuring pMLU on two large Dutch CHILDES corpora. We propose minimal sample sizes for reliable measurements relative to the stage of phonological development.

This paper investigates to what extent the grammatical class(es) of a word can be predicted on the basis of phonological and prosodic information only. We report on several experiments with an artificial learning system which has to assign English word forms to their appropriate grammatical class, using various types of phonological and prosodic information. First of all, we examine several phonological cues which were claimed by Kelly (1996) to be particularly good for distinguishing English nouns from verbs. Our results indicate that these cues are indeed partially predictive for the problem at hand and reveal that a combination of cues yields significantly better results than those obtained for each cue individually. We then show experimentally that ‘raw’ segmental information, augmented with word stress, allows the learning system to improve considerably upon those results . Secondly, we investigate several generalizations of the approach: basic segmental information also proves to be more predictive when the task is extended to encompass all open class words in English, and these findings can be replicated for a different (though related) language such as Dutch.

A longitudinal analysis is presented of the fillers of a Dutch-speaking child between 1.10 and 2.7. Our analysis corroborates familiar regularities reported in the literature: most fillers resemble articles in shape and distribution, and are affected by rhythmic and positional constraints. A novel finding is the impact of the lexical environment: particular function words act as 'anchor' words that attract occurrences of schwa fillers after them. The child inserts significantly more schwa fillers in these contexts. The anchor words are among the most frequent words preceding articles in the input, indicating a sharp sensitivity to such distributional regularities. Nasal fillers too are affected by distributional learning, but at the phonological level: the child first uses nasals before [h]-initial nouns, and then generalizes this usage to all [h]-initial words. These observations are related to the growing body of evidence for the impact of distributional learning on early language production.

No abstract