Backward Anaphors.

Abstract

This paper stipulates Referring expressions (with the special reference of „Backwardness‟ of such identity relations) in Punjabi. In natural languages, such expressions are in anaphoric nature and linguistically also specified with all possible categories (e.g. nominal, pronominal, verbal etc.). In general, anaphors are basically lexical categories that are unmarked and explicit and implicit with various manners in the natural language. In GB framework, it is mostly discussed with principles of (A) as dependent in the same local domain of the sentence, in the form of lexical anaphors. But here, we check its counter part as cataphora (he=John) in such relation and will discuss with some instances in ILCI Punjabi corpus. In fact, the study of anaphors is computationally a new research area in Punjabi. From the grammar point of view, it is basically encountered as pronominals in most of the cases and somehow it can be managed with Government and Binding framework (1981, 82). But still, there is no specific algorithm, which can attract machine-learning process in Punjabi. In this paper, the main concern is to prepare provisional algorithm platform for Punjabi through the discussion of cataphora or may be backward anaphora in 1,000 sentences of ILCI Punjabi corpus. We manually search around 1,000 sentences and find some instances of backward anaphors. We analysis them according to Lakoff (1968) findings as backwardness for English and collect Punjabi instances on the basis of such inquiry. As mentioned in the previous sections, it is not the fully-grown area of research in Punjabi. It is the initial endeavor where we are trying to use the instances of cataphora for computational purposes. By following previous proposals of cataphora, we discuss POS tagged Punjabi files and then separate cataphoric instances. We propose some steps as an instruction in the form of cyclic manner of searching first some possible referring expressions as PR_PRP and then move on to another kinds of DM_DMD categories. It may assume that the cyclic conditions may bring positive results with probably some changes in future. It also accepts that the cases like doubling; large group of referent choice may also be discussed surely when we put such instructions in the system form.