Restrictive Glosses and CALL Comprehension: An Exploratory Study David J. Galloway

Computer-based L2 reading, as opposed to traditional printed readers, eliminate the arbitrariness of printed glosses. Such glosses are determined not by the individual student reading the text, but by the preparer of the text. There will always be words which are glossed which the student did not need, and words which are not glossed which the student did require. Creating a reading program in which all words have glosses eliminates this arbitrariness, but not without creating its own problems. When L2 students are confronted with a CALL reading program which allows unlimited use of clickable glosses, and which contains a gloss for every word, readers will overuse the glosses. Particularly weak students will sometimes check the glosses of up to 99% of the words in a text. Since glosses are a learning tool, and we desire the students to master a text without assistance as soon as possible, it is necessary to reduce their reliance on glosses while preserving their ability to check the definition of their choice of words and without reducing comprehension to a point where reader frustration inhibits the reading process. This project tested how restrictive glossing would affect both comprehension during a CALL reading session and student attitudes towards the session itself. The reading program employed contains a subroutine written which restricts total use of glosses while preserving the ability to access the definition their choice of words. The number of glosses used appears on the screen to keep the student aware, and is updated as additional words are clicked. Repetitions are counted as part of the total number of glosses, and once the total is reached, the program no longer shows definitions for words clicked. (Explanatory textual notes are not restricted and may be accessed throughout the reading process.)

The students were all enrolled in a first-year, second-semester Russian language course (n=11). At the time of the experiment, they had completed approximately twenty-four weeks of language instruction. They read two unmodified Russian skazki on Baba-Jaga from Afanas&soft;ev's collection, numbers 102 and 106 (1984 edition), designated texts A and B. These tales are similar in length (463 and 540 words, respectively) and theme. Glossing was restricted at either 15% or 30% in the tales the students encountered. The higher-than-average gloss restriction was devised since (1) multiple clicks of the same word were counted as part of the total allowed percentage, and (2) students were of such a low language competency that higher gloss allowance seemed advisable. The students were divided into four groups, each following one of the following orders of texts: (A at 30%, B at 15%); (A at 15%, B at 30%); (B at 30%, A at 15%); (B at 15%, A at 30%.) After reading each of the two tales, they completed an online comprehension test consisting of twenty English and twenty Russian multiple-choice questions.

The experiment results show that student comprehension of each text (A and B) was not a function of the glossing restriction. For each given text, the mean comprehension scores were practically identical. Individual comprehension scores went up when students read text B, as opposed to text A. This, however, is not a function of the 30% gloss that some students were able to employ when reading text B. All students' scores increased, showing that the change is a function of text difficulty rather than gloss restriction. Though the two texts are similar, text B is easier to read because of its repetition pattern, which is immediately discernible and follows the traditional triad structure. Each action or spoken line is repeated three times, which aids in comprehension.

Overall student reaction to gloss restriction was positive: most did not need more glosses than they were allowed. Many reported in their post-experiment questionnaires that the gloss restrictions of 15% versus 30% forced them to use contextual guessing strategies and other approaches to discerning meaning before resorting to the gloss. This type of response is particularly valuable in beginning language learners, who too often rely on word-for-word approaches to reading. It is hoped that the data from this experiment will suggest further refinements to CALL reading programs in a continuing effort to maximize student benefit from online reading.