In lieu of an abstract, here is a brief excerpt of the content:

18 Use and abuse of autonomy in computerassisted language learning: some evidence from student interaction with SuperCloze Vance Stevens Introduction Although the situation is steadily being corrected, it has often been noted that CALL (computer-assisted language learning) has so far developed well ahead of its research base (e.g. Dunkel 1991). The result is that developers of CALL often work on intuition alone and have little real idea what students actually d o wit h thei r program s (Chapell e 1990) . To compound thi s situation, what research there is on CALL effectiveness is often done using procedures wher e th e researche r intrude s o n th e learner , thu s possibl y contaminating the autonomous aspects of the process under study. Feldmann and Stemmer (1987) discuss the various cognitive limits that may interfere with concentration on the task under study when student s are aske d t o "thin k aloud " abou t wha t the y ar e doing . I t follow s tha t intrusive protocols could influence results in studies such as that of Windeatt (1986), who videoe d screen s a s his subject s though t alou d whil e doin g computer-based cloz e exercise s an d foun d tha t ther e wa s littl e us e o f program help features. Stevens (1991a, 1991b, 1991c), on the other hand , finds through non-intrusively tracking students working under self-acces s conditions that they sometimes overuse, even abuse, help features rathe r than rel y o n thei r competenc e i n the language t o solve problems. Thu s degree of intrusion may be a factor in the outcome of such studies. Research into what students do with CALL in self-access should ideally be carried out non-intrusively, yet due to the intrusive nature of most studies of the medium, rarel y is CALL studied i n its pure self-acces s state . On e reason for this is the difficulty i n controlling variables in a process which 282 Vanc e Stevens the experimenter essentially observes without interference. Also, for ethical reasons, researchers who identif y individua l subject s mus t infor m the m prior to including them in a study, in effect saying: "You are subjects in an experiment but please carry on as if you weren't!" As this could render it impossible t o study self-acces s with tha t set of subjects, one solution, a s with the present experiment, is to use subjects anonymously; that is, record their key presses on computers but take no record of who the individuals were who made them. Although many data are accordingly lost, such as relative Englis h proficiencie s o f subject s exhibitin g certai n performanc e behaviours, the process unde r stud y ca n at least be assume d t o be i n a virtually uncontaminated state . Another issue in CALL is the degree to which giving students control in self-access affects thei r learning. As Chapelle and Mizuno pointed out , as of 1989, the issue of optimal degree of learner control over CALL "has not yet been investigated". However, Pederson (1986) compared two groups of students , one of which was allowed t o refer a t any time to a readin g passage durin g th e cours e o f answerin g question s o n that passage, an d found tha t th e passage-unavailabl e treatmen t resulte d i n significantl y higher level s o f comprehensio n becaus e thos e student s wer e force d t o process the text when they had their one chance to read it. One purpose of the presen t stud y i s to gain furthe r insight s int o how contro l ove r hel p features affects the degree of engagement with the target language for the students in the study. Although CAL L is typically referred t o as a generic entity, in fact it s manifestations ar e many : wor d processing , simulation , concordancing , database exploration , an d almos t anythin g els e wher e computer s manipulate a human language or use one as an interface. Thus, as a study of 'CALL' would rank in scope with a study of 'the world', that scope must be narrowed down . Suggestions such as Kleinmann's (1987) that CALL...

Share