Grammars as User Models

This paper describes how grammars of native languages can be used is tutoring systems to facilitate instruction in a second language. The grammar functions as a user model, which enables the system to customise its responses by addressing problems that may be due to interference from the native language. Its correction strategy is based upoa comparison of the native language model wi th a model of the target language. The problems and solutions presented in this paper are related to the more general question of how modelling previous knowledge facilitates instruction in a new skill. 1 . I N T R O D U C T I O N People learning a second laaguage sometimes kave problems that caa be attributed to interference of their native language. This interference shows up in systematic errors in both speech and written text. Our concern is with syntax and syntactic interference. We waat to capture a student's knowledge of his/her native grammar that leads to this interference and use it to assist the student in learning that of the second laaguage. Much research in AI has been devoted to the development of question answering systems, expert systems, and tutorial systems. Part of this research has involved enabling these systems to produce more "natural ," "cooperative" responses, to recognize misconceptions and to provide explanations. One approach to improved interaction has been to incorporate a user model, e.g. | 1 , 5, 7). The claim we make in this paper is that grammars can serve as user models. We address the question of how correspondences between the grammars of two languages can provide an account of syntactic errors made by native speakers of one language attempting to learn a second one. This account can then be used in correcting the student. As aa example of this, we describe a computer tutoring system, VF* which focusses on the acquisition by non-English speakers of English constructions formed from a verb plus particle or verb plus prepositional phrase. It has often been claimed that people rely heavily on their previous knowledge when learning a new skill [6, 8). While this previous knowledge caa sometimes help their learning, it caa oftea hinder it as well: Halast and Moraa (2), for example, describe a case ia which people reasoa incorrectly by analogy from a previous skill. Ia learning a second language, one's native language may be a source of both correct and incorrect analogies. Thus VF* includes a grammar of the native language in order to detect and correct errors due to incorrect analogy. This grammar also enables Vp to explain the differences and similarities ia the verbal constructs of the two languages in focus. 2. OUR APPROACH TO SECOND LANGUAGE The fundamental claim of one theory of second laaguage acquisition, the contrastive analysis theory (CA) |3j, is that when people are learning a secoad language L2, the patterns of the language to be learned are matched with those of the native language L1. Those that match are fairly easy to learn while those that do not, become more difficult and result in errors. The work presented here has been developed along the lines of CA theory. Since interference errors reflect differences between the grammar of the student's native language and that of the target laaguage, they are oftea predictable from a contrastive analysis of the two grammars. This work focusses on one problem that appears on the syntactic level of L2 acquisition, that is, the use of the complex construction verb plus preposition/particle, ia the English of non-native speakers.** Consider the following interaction between a tutorial system for English and a student: Tutor : TRANSLATE THE FOLLOWING SENTENCE: E1 penso' en v i a j a r a Europa ahora. Student: He thought in t r a v e l l i n g to Europe nov. Tutor : I nco r rec t . The cor rec t answer ia 'He thought of t r a v e l l i n g to Europe now.' or 'He thought about t r a v e l l i n g to Europe now'. In Eng l ish , you can use the verb aa in the f o l l ow ing cases: 'There i s l i t t l e oppor tun i ty t o what the l ong te rn so lu t i on nay be ' 'With the beginning of the new day we had to th ings aga in ' 'She haa to a catchy nana f o r t h i s system' The system here provides the correct answer but it fails to recognise that the user has translated directly from Spanish.*** It leaves students to determine the relationship between the use of certain structures or patterns ia the new domain (or laaguage) by themselves. Here, the tutorial system could have provided a more adequate explanation by informing the user why the error occurred. *This work is partially supported by NSF grant IST81-12439 and by ARO grant DAAG29-84-K-0061. **Note that we are dealing only wi th second laaguage acquisitioa, aot third or fourth. Here the learner generalises only from his/her native language. ***By 'direct translation' we mean the most common translation used for a certain word, that is, the first one found when looking up that word ia a dictionary. This is different from the 'corresponding translation' which is the actual translatios of the words and which varies in the context of the sentence. For example, the direct translation of < e n > is < i n > while the corresponding translation of < e n > when used with the verb , < t o t h i n k > , i s < o f > o r < a b o u t > .