The attitudes and learning strategies of 100 language learners who are also users of Self-Access Centers (SACs) in a theoretical context of Self-Access Language Learning (SALL). Though questionnaire survey, the present study discovered several significant correlations between learner’s attitudes towards SALL and their language learning strategies. It also established a profile of SALL center users in their attitudes and learning strategies from an comparative view based on their previous SALL experiences. The results showed that leaners who are more ready for Self-Access tend to use affective strategies. Memory strategies are practiced more often among leaners who prefer learn grammar without teachers and have more speaking activities in class, also who are relatively more self-aware of their own learning objectives. Cognitive strategies are widely used by leaners who prefer leaning grammar by themselves, also those who are more confident to solve problems by themselves. There is a stronger intention of control over content selection for leaners who had SALL experiences, especially in utilizing non-staffing SALL resources at SACs. / published_or_final_version / Education / Master / Master of Education

This study examined children's performance on a decontextualization task requiring the ability to deduce the meaning of unknown words from verbal context and their ontological knowledge structure as indicated by their judgments about both anomalous and sensible statements containing the unknown words. A comparison was made between performance on the decontextualization task and verbal and nonverbal ability and between the subjects' ontological knowledge structures and verbal and nonverbal ability. It was hypothesized that performance on the decontextualization tasks would be positively correlated with both ability measures, but ontological knowledge structure would remain constant across ability levels. First, third and fifth grade subjects' participated in the study. Performance on the decontextualization task correlated positively with verbal ability for all three grade level and with nonverbal ability for grade one. Presence of the M-constraint (Keil, 1979) was evident across ability levels as well as grade levels. Greater differentiation in ontological knowledge was indicated across grade levels but not across ability levels within a grade level. Results supported previous research of Keil (1979, 1983a, 1983b).

No description available.

No description available.

No description available.

No description available.

No description available.

No description available.

This thesis investigates the role of implicit knowledge in second language acquisition, presenting five experiments and related simulations based on artificial grammar learning. It examines whether second language learners can acquire implicit knowledge of noun–verb agreement. In addition it tests whether their ability to do so is influenced by the number of words that intervene between the relevant noun and the finite verb in the input sentences, as this affects performance in artificial grammar learning, the serial response time task and the statistical learning paradigm. Experimental participants were exposed to a modified version of Persian or Basque while performing a memory task. Next, two grammaticality judgement tests (one timed and one untimed) and a sentence correction task assessed whether the participants had acquired either the target noun–verb agreement or pairs of adjacent words that appeared frequently in the learning items. Performance reflecting implicit and explicit knowledge was distinguished according to three criteria based on R. Ellis (2005) and according to the assumption that the former is not under conscious control. Participants’ performance suggested that they had implicit knowledge of the adjacent word pairs. Similarly, the results indicated that they had implicit knowledge of subject–verb agreement when a single word intervened between the subject and the verb, but not with two intervening words. Connectionist simulations supported the results and indicated that performance was unlikely to improve if more exposure were given. Although the influence of additional factors is discussed, the results generally supported the view that an increase in the number of intervening words reduces learning outcome. The intriguing similarity between the results of this thesis iv and previous research in artificial grammar learning, the serial response time task and statistical learning experiments suggests that future research should directly compare the paradigms to ascertain whether similar learning processes are engaged in each case.

Knowledge acquisition from databases is a research frontier for both data base technology and machine learning (ML) techniques,and has seen sustained research over recent years. It also acts as a link between the two fields,thus offering a dual benefit. Firstly, since database technology has already found wide application in many fields ML research obviously stands to gain from this greater exposure and established technological foundation. Secondly, ML techniques can augment the ability of existing database systems to represent acquire,and process a collection of expertise such as those which form part of the semantics of many advanced applications (e.gCAD/CAM).The major contribution of this thesis is the introduction of an effcient induction algorithm to facilitate the acquisition of such knowledge from databases. There are three typical families of inductive algorithms: the generalisation- specialisation based AQ11-like family, the decision tree based ID3-like family,and the extension matrix based family. A heuristic induction algorithm, HCV based on the newly-developed extension matrix approach is described in this thesis. By dividing the positive examples (PE) of a specific class in a given example set into intersect in groups and adopting a set of strategies to find a heuristic conjunctive rule in each group which covers all the group's positiv examples and none of the negativ examples(NE),HCV can find rules in the form of variable-valued logic for PE against NE in low-order polynomial time. The rules generated in HCV are shown empirically to be more compact than the rules produced by AQ1-like algorithms and the decision trees produced by the ID3-like algorithms. KEshell2, an intelligent learning database system, which makes use of the HCV algorithm and couples ML techniques with database and knowledgebase technology, is also described.