|
|
|
|
|
|
|
|
Kara McBride
Ph.D. Candidate
Second Language Acquisition and Teaching
CALICO: Computer Assisted Language Instruction Consortium
University of Hawaii, Manoa, Oahu, Hawaii
May 16 -20, 2006
|
“The Maze Task as Learning Activity and Testing Method”
ABSTRACT
While pretty much everyone learns his or her first language fully, it is rare for a second language learner to ever reach the same level of proficiency. This is one of the primary mysteries of second language acquisition theory: why does learning a second language (L2) require so much effort? Are additional languages learned by different cognitive mechanisms than the first language?
What are these mechanisms?
Because language is unique to humans, to be able to answer these questions is to learn more about the human mind. It also helps us perfect second language teaching methods and design learner materials, on computer or otherwise.
The maze task is a new technique that we (in the lab in which I work) are exploring to do both these things: both to investigate the nature of L2 acquisition, as well as stimulate it. The task is performed on a computer. The participant goes from screen to screen and creates sentences by choosing, between each pair of choices, the one word that, when combined with the previous words of that item, continue a grammatically viable sentence. Because the participant must choose words, the activity is similar to a production task. Unlike most production tasks, the creator can force the participant to create specific structures and/or use certain vocabulary. The computer records the participant’s choices and reaction times, which allows the investigator to know where sentence parsing breaks down and which junctions are most difficult (as shown by slower reaction times).
This presentation reports on two large studies. In the first, the maze was used as a learning activity. Six different sections of second semester Spanish classes at UA were given maze activities twice weekly for eight weeks. The activity was reported to be entertaining and motivating by the majority of the participants. Comparing class grades of participants, and between the control group and those who received the training, we find evidence that the task is effective as a practice method but, unsurprisingly, not as a way of teaching entirely new concepts.
In the other study, the maze task was used as a testing method and helped reveal differences in kinds of language development. This second study was focused mainly on listening comprehension. The maze task involves written language, but it shares in common with spoken language a time constraint and the fact that the perceiver is forced to process words in a linear fashion (unlike regular reading of printed text, where one can go back to previous sections of the text when necessary). The maze task was included in this study as a way of getting at the relative importance of distinct language processing abilities, such as the ability to keep up with the pace of input, the ability to use contextual cues, and accurate grammatical parsing, to name a few.
This second study was done online in Chile. 140 English as a foreign language students took two pretests online: one was a listening comprehension test, and the other was a maze task. Ten lessons (about five weeks) later, the participants were given posttests of the same formats. The listening comprehension tests had four dialogues: two delivered in natural, fast speech (that is, “fast” in the ear of the language learner); the other two were delivered in a slower, clearer “teacher talk” style. The ten lessons that the participants were trained on were the same in content for all participants, but participants were divided into four training groups, differing on whether they heard training dialogues spoken in a fast/natural style, or in the teacher talk style, or if they could choose between speeds.
There were main effects for speed, group, and test (pretest vs. posttest), and interesting interactions between factors. The differences, across groups, in gain scores on the maze task helped to clarify what had happened. Distinct language skills were improved in the different training groups, helping to elucidate the multiple components that make up language ability and the way that they interact. A model for listening comprehension is proposed that is able to explain the findings in both studies.
|
|
|
