One of the earliest studies of learner strategies was made in 1976 by Wong-Fillmore. Her aim was to investigate how children increased in communicative competence in English. By investigating children learning English as a second language (L2) in American playgrounds, she identified three social strategies and five cognitive strategies that were used by the most successful learners. The social strategies were deemed to be the most important and were as follows:

• joining a group and acting as if you understand what is being said;
• giving the impression, with a few well chosen words, that you speak the language;
• counting on friends for help.

The five cognitive strategies were:

• assuming what people were saying was relevant to the situation at hand;
• getting some expressions and starting to talk;
• looking for recurring parts in formulae that have been learned;
• making the most of whatever language they had;
• working on the big things first and saving the details for later.

Wong-Fillmore proposed that there were links between the social strategies and some of the cognitive strategies. Although this study was fairly unsystematic and unempirical, it prepared the ground for a large amount of empirical research in the area.

Naiman et al (1978) were the first researchers to attempt a systematic, empirical observation of good language learners or 'GLLs'. Their aim was to isolate the common characteristics, personality traits, cognitive styles and strategies of the GLL. The research was divided into two parts. In the first part, Naiman et al interviewed a group of successful language learners about the strategies they used, and in the second part they observed four classes of 10 to 14-year-olds studying French in Canada. The combination of these two approaches enabled them to achieve much greater levels of detail than those achieved by the Wong-Fillmore survey. They identified the following five 'good language learner strategies':

• the 'active task approach' in which the learners actively involved themselves in the language learning task;
• the 'realization of the language as a system' in which they saw the organizing framework behind the language;
• the 'realization of language as a means of communication and interaction';
• the 'management of affective demands' which meant that the students were not embarrassed to speak the language or put off by their mistakes;
• the 'monitoring of L2 performance' in which learners tested their guesses and made adjustments accordingly.

Despite identifying these attributes that were common among the good language learners, Naiman et al also observed a wide range of individual differences among the school children and the GLLs, leading them to stress the importance of the complexity and individuality of each learning situation and career. They went on to conclude that 'the study as a whole suggests that the successful or good language learner, with predetermined overall characteristics, does not exist' (Naiman et al, 1978: 102) and that a student should adapt his/her language learning to suit his/her personality. They did, however, feel able to conclude that GLLs are capable of honest self-assessment and flexibility.

A further attempt at a systematic observation of the strategies used by GLLs was made by Rubin in 1981. After having employed a variety of techniques, ranging from classroom observation to directed self-report, she concludes that six strategies are frequently employed by GLLs. They are:

• clarification;
• monitoring;
• memorization;
• guessing;
• deductive reasoning;
• practice.

Rubin places emphasis on the personal and metacognitive strategies: 'By choosing and prioritising, students set their own learning goals.' (1981: 25)

Skehan (1989: 79-80) is critical of these early surveys into learner strategies. Firstly, he points out that the Naiman et al survey places a great deal of emphasis on self-report data and therefore should be treated with caution. The interviewee may consciously or unconsciously distort the facts. Furthermore, it is questionable whether subconscious strategies can be assessed through conscious means. Secondly, Skehan draws our attention to the possibility that the findings may reflect the nature of the survey itself, ie a communicative survey such as Wong-Fillmore's predicts social strategies, whereas an academic survey such as that of Naiman et al predicts cognitive strategies. Thirdly, he attacks their unrepresentative subject selection. Wong-Fillmore only looked at six subjects, Naiman et al only looked at good language learners (who may have identical strategies to the bad ones), and Rubin only looked at students of the University of Hawaii. Indeed, Naiman et al themselves observe that their study contained too few subjects and too many variables, and that a longitudinal study would probably have been more productive.

These criticisms are probably less applicable to the later learner strategy research. For example, Huang and Van Naerssen (1985) found that learner success correlated significantly with strategies such as speaking the L2 with other students, thinking in the foreign language and participating in group oral communicative activities. They named these 'functional practice strategies'. The first longitudinal study to appear was carried out by Chesterfield and Chesterfield (1985). After having studied fourteen Mexican children learning English in the USA, they found that the first strategies to be used were receptive and self-contained but that the students subsequently employed strategies that permitted interaction (for example, requests for clarification and verbal attention-getting). Both these surveys support Ellis's (1985) proposition that good language learners pass from formulaic speech (for example, pattern memorization) to creative speech (that is, speech which is creative in the Chomskian sense, where the student is able to produce entirely novel sentences) via a process of hypothesis testing.