A review is offered of various bilingual memory studies that have been carried out during the past quarter century. The vast majority of these studies have employed lexical items as the investigative tool, tested outside of a grammatical or semantic context. Some of the major limitations of such lexical studies are discussed. A number of arguments are provided to support the trend in recent studies to examine bilingual memory within the framework of contextualized language units. In contrast to the earlier lexical studies, which largely focused on storage issues, the more recent studies tend to examine issues concerned with lexical access and processing of the bilingual's two linguistic codes. It is argued that studies employing contextualized linguistic units are more reflective of natural language processing.

An earlier experiment by Byrne (1981) found that young, poor readers tend to act out sentences containing adjectives with object control, like easy, as though they were adjectives with subject control, like eager. Byrne interpreted this result as evidence that poor readers lag in the acquisition of syntactic knowledge underlying this distinction. However, the possibility that a processing limitation could have contributed to the poor readers' difficulties with objectcontrol adjectives had not been fully explored. In an effort to tease apart these alternatives, we tested comprehension of object-control adjectives in second grade good and poor readers, using both an act-out task and a sentence-picture matching task. Contrary to Byrne's (1981) results, we did not find significant group differences in interpreting object-control adjectives with either task. Reasons for the discrepancy are suggested, and remedies for the pitfalls in designing experiments to assess syntactic knowledge in young children are proposed.

The role of four cues in the assignment of actor role in Arabic was studied. These cues included animacy, case marking, gender agreement, and word order. The cues were set-in-a competing/converging fashion following the framework of the Competition Model of Bates and MacWhinney (1981, 1982, 1987). Fifty-four sentences representing all possible permutations of the cues were presented to 100 native speakers of Arabic, who were asked to identify the actor in each sentence. Subjects relied mainly on gender agreement and case marking in the assignment of the actor role. The results suggested that cue validity is very highly correlated with cue strength. It was also found that cues do not exhibit an additive nature, and that cross-linguistic formal similarities do not warrant similar processing strategies. Order of nouns tended to represent a “latent strategy” to be utilized only when other relevant cues were not available.

This article reports a longitudinal study of developing communication in two profoundly deaf preschool boys growing up in oral deaf families who use oral English as their primary language. The children were videotaped in play interactions with their profoundly deaf mothers. The nature of the gestural communication used by the dyads is the focus of interest in this article. In contrast to hearing mothers of deaf children, the two mothers used extensive gestures to accompany their speech, including rich and varied gesture sequences. The children also developed a repertoire of gestures that, in most respects, mirrored the properties of the input gesture they received. The results provide a window on the natural ontogenesis of a compensatory gestural system.

Bialystok and Ryan's (1985a) model of metalinguistic awareness, which consists of two processing components (the control of linguistic processing and the analysis of linguistic knowledge), was investigated in children between the ages of 5 and 7 years. Each of the two components was assessed by four metalinguistic tasks. The results were consistent with Bialystok and Ryan's model in that confirmatory factor analysis, using LISREL, supported both the control of linguistic processing and the analysis of linguistic knowledge. Support for the first component was stronger than for the second; however, this may have been due to the problems encountered with two of the tasks. The results further suggested that the two components of metalinguistic awareness are interrelated, as their correlation was moderately high.

Practice on cognitive tasks, in general, and word recognition tasks, in particular, will usually lead to faster and more stable responding. We present an analysis of the relationship between observed reductions in performance latency and latency variability with respect to whether processing has merely become faster across the board or whether a qualitative change, such as automatization, has taken place. The coefficient of variability (CV) - the standard deviation of response time divided by the mean latency - is shown to be useful for this purpose. A cognitive interpretation of the CV is given that relates it to issues of skill development.Data from second language learners' word recognition performance and from a simple detection task are presented which confirm predictions drawn from this interpretation of the cognitive significance of the CV. Initial improvement in a second language word recognition task was interpreted as involving more efficient controlled processing, which later gave way to automatization. The implications of this index of skill are discussed in relation to second language development and the general issue of automaticity of processing components in cognitive skills.

This investigation focused on early school-aged children's ability to negotiate shared reference in a task-oriented communication game. Conversational partners worked together on initiating and refining mutually acceptable referential labels for a series of abstract figures. Each child participated as both a sender and a matcher. In Experiment 1, developmental trends in the use of conversational strategies such as utterance-contingent queries, acknowledgments, and negations were observed. However, communicative efficiency was disrupted when the children switched dyadic roles. They reintroduced figures with novel labels when switching to a speaking role from a previous matching role, and vice versa. In Experiment 2, communicative efficiency was again disrupted at the point of role switch, even when the children were given feedback designed to minimize cross-role inconsistency. Interestingly, in both experiments, communicative inefficiency did not reduce a dyad's overall communicative success. These results suggest that, while children are in the process of learning how best to coordinate the use of conversational strategies and procedural rules, they still manage effective communication, albeit with a little extra effort.