The relationship between comprehension language training and production language training was examined by teaching 12 adults and 12 children a miniature language. The relationship between the two types of training was determined by examining their effects on the generalization of names. The results indicate that production training generalizes more extensively than comprehension training. The advantages and disadvantages of both types of training are discussed.

This study explores the metalinguistic abilities of prelingually deaf children aged 4–7, who are users of Signed English, with regard to their explicit segmentation of Signed English sentences into words. Subjects exhibited varying metalinguistic abilities that generally increased with age and that were similar to the developmental pattern found in hearing populations. Based upon performance with respect to four factors (i.e., explicit segmentation, omissions of function words, content words, and inflectional morphemes), subjects were grouped into four classes. In Class 1, sentences or groups of words were not segmented. in Class 2, major constituents of the sentences were segmented. In Class 3, major constituents and some function words were segmented. In Class 4, the entire sentence was segmented. Patterns of omissions found throughout the classes for function words and inflectional morphemes are discussed. Also addressed are implicit segmentation skills found to develop prior to the development of explicit abilities.

Younger readers, mean age 6;11, and older readers, mean age 8;7, matched on IQ and SES, read 18 consonant phonemes embedded in nonsense CVCs. Results indicated that (a) within groups, younger readers made significantly more errors on digraphs than graphs; (b) younger readers made significantly more errors on graphs in the final position; (c) there was no position effect for either group in reading digraphs. Analysis of errors based on the number of distinctive features shared within stimulus-response pairs revealed that (a) on graphs, both groups made significantly more errors that shared two distinctive features; (b) on digraphs, the majority of errors for younger readers shared one distinctive feature; (c) the predominant strategy on digraphs for both groups was to read one letter. These findings demonstrate that most oral reading errors produced by beginning readers are not random, but rather are systematically related to the stimulus.

Categorical perception of a synthetic /r/-/l/ continuum was investigated with Japanese bilinguals at two levels of English language experience. The inexperienced Japanese group, referred to as Not-experienced, had had little or no previous training in English conversation. The Experienced Japanese group had had intensive training in English conversation by native American-English speakers. The tasks used were absolute identification, AXB discrimination, and oddity discrimination. Results showed classic Categorical perception by an American-English control group. The Not-experienced Japanese showed near-chance performance on all tasks, with performance no better for stimuli that straddled the /r/-/l/ boundary than for stimuli that fell in either category. The Experienced Japanese group, however, perceived /r/ and /l/ categorically. Their identification performance did not differ from the American-English controls, but their overall performance levels on the discrimination tests were somewhat lower than for the Americans. We conclude that native Japanese adults learning English as a second language are capable of Categorical perception of /r/ and /l/. Implications for perceptual training of phonemic contrasts are discussed.