A Comparison of the Microstructures of Three English-Persian Dictionaries Based on Fillmore's Frame Semantics

Moving toward more scientific lexicography in Iran requires lexicographers and researchers to pay closer attention to and develop deeper understandings of the interaction between theory and practice in lexicography. If lexicographers ignore theories in their analysis while creating an entry's microstructure, the result will be inconsistency in the microstructures as well as a negligence of some aspects of the words' behaviors (Atkins, 2001, pp. 1-2). In addition, since the microstructure of bilingual dictionaries has directly to do with the quality of translators' finding the right equivalents, such inconsistency and weakness may be problematic for translators. The analysis of bilingual dictionaries is one of the areas of lexicography that seems to require more scholarly work. One way to do so is by evaluating the microstructures of dictionaries. It goes without saying that such studies can result in optimizing the process of lexicography in general and more efficient use of bilingual dictionaries in particular, especially by translators as specialized users of such dictionaries. Therefore, drawing on Fillmore's semantic frames, the present descriptive-analytical research paper, that analyzed corpus data, aimed to compare three bilingual English-Persian dictionaries in order to identify some of their microstructural differences and postulate some ways for the improvement of such dictionaries. The author argues that there is a reciprocal relationship between the development of lexicography and by extension dictionaries on the one hand and the improvement of language and translation on the other hand. One of the approaches to lexicography is the cognitive one in which frame semantics is a prominent theory. This theory can, among others, illuminate the relationships between the senses of lexical entries and lexical units. Moreover, this can cast light on how senses can be best structured within a single entry. Despite its rather long history, frame semantics has recently been used in semantic tagging and in compiling corpus-based dictionaries (Fontenelle, 2009, p. 38). It has also been applied in lemmatization and parsing. This theory is the central part of the FrameNet project started in 1997 in the University of California, Berkeley (Boas, 2009, p. 16; Fontenelle, 2009, p. 38). Frame semantics is different from similar approaches to lexical meaning in that the meanings of words are realized though comparing them with the common bodies of knowledge, i.e. frames. In other words, words are only related to each other through the frames they share (Fillmore & Atkins, 1992, pp. 76-77). In Fillmore's theory, the notion of frame replaces concepts such as scene, scenario, meme, schema, etc. (Fillmore & Baker, 2010; Fillmore, 1982). To explain his theory, as Rojo López (2002) points out, Fillmore has frequently given the example of two English synonymous words, those of coast and shore, which considering the conceptualization situations of words, evoke different semantic fields (water in case of shore and land in case of coast). However, this depends on the viewer's view from the sea/lake etc. or the land.    The corpus of this research consisted of, using their short popular titles, Aryanpour, Moaser-e Hezareh and Moaser-e Pouya English-Persian dictionaries from each of which 17 entries, related to the two frames of MOTION (6 entries) and KINSHIP (11 entries), were selected based on the information available in FrameNet online database. The corpus dictionaries were compared and contrasted with Oxford Advanced Learner's Dictionary (OALD) (2002) as the comparand dictionary. The categories selected for comparison comprised collocations, examples, inflections, parts of speech, synonyms/antonyms, valency, and the number of senses and equivalents. These categories in corresponding entries of the corpus dictionaries were compared with each other and with OALD. The data collected contained both quantitative (as in the number of examples, senses and equivalents) and qualitative (for collocation, valency) information. The findings indicated weak hierarchical structuring and differences in differentiating senses, dissimilarity in the inclusion of different parts of speech in and across entries, a lack of correspondence between some proposed equivalents and the selected frames, and more importantly defects in considering important parameters such as collocations, examples, and valency. Specifically, the entries analyzed in the corpus dictionaries did not include specific parts devoted to collocations, nor did enough examples for each sense, while OALD provides users with a large number of examples that at least indirectly help learners to notice collocations of the given words. This shows that to enrich bilingual English-Persian dictionaries, compilers should be advised to start using Persian corpora. Another defect found was that the corpus dictionaries do not generally seem to care for valency. Due to the important characteristic of Frame Semantics in defining relations, valency can play a substantial role in the microstructures of dictionaries and in helping translators to find the right equivalents. Yet, the dictionary of Moaser-e Hezareh has marked transitive and intransitive verbs, which has to do with the description of valency. What is more, in the entries studied, even though the dictionary of Moaser-e Pouya provides a higher number of senses, this might not guarantee a more efficient use of the dictionary mainly because of examples shortage, valency description, and insufficient number of equivalents.