A Corpus-based Investigation of the Near-synonymy of ‘‘ziyâd and farâvân’’ Size Adjectives according to Behavioral Profile Approach

Investigations show that in Persian no significant research has been done on lexical relationships of size adjectives. In this paper, near-synonymy relationship of ‘‘ziyâd and farâvân’’ size adjectives and their inflectional forms has been studied based on Behavioral Profile Analysis approach of Divjack and Gries (2006) and the data of Hamshahri 2 corpus (Alahmad and et.al, 2009). Accordingly in this study, first, by using AntConc software (2011) the data related to ‘‘ziyâd, ziyâdi, ziyâdtar, ziyâdtarin, ziyâdtari, farâvân, farâvâni, farâvântar, farâvântarin, farâvântari’’ size adjectives were extracted from the corpus and then were annotated on the basis of the morphological, syntactic and semantic features. The result of behavioral profile approach which is acquired according to hierarchical cluster analysis indicated that the highest level of near-synonymy belongs to ‘‘ziyâdi, farâvâni’’, ‘‘ziyâdtari, farâvântari’’, ‘‘ziyâdtar, farâvântar’’, ‘‘ziyâd, farâvân’’ and ‘‘ziyâdtarin, farâvântarin’’ groups and the lowest level of near-synonymy is between ‘‘ziyâdi, farâvâni, ziyâdtari, farâvântari’’ and ‘‘ziyâd, farâvân, ziyâdtarin, farâvântarin, ziyâdtar, farâvântar’’ groups. Finally, annotation analysis based on Z-scores has specified all meaningful and meaningless ID-tags of size adjectives.

1. Introduction

Persian is one of many languages throughout the world which have a class of words whose grammar is clearly different from the others, and can be identified as adjectives. Adjectives, also, can perform some semantic tasks. For example, one major task of adjective is to show the property of things. For example, Ali's house is big, this sentence could referred to Ali's house, that it has a property of being big.

Although traditional grammarians stressed on the formal features, today, linguists typically classify adjectives according to semantic features. However, one lexical field that has received a lot of attention both in general and in corpus is adjective size, presumably because it includes two pairs of canonical antonyms and many studies have already focused on size adjectives. For example, Charles and Miller (1989) showed that big and little as well as large and small tend to occur in one and the same sentence in the Brown corpus with a probability that is much larger than chance would predict and concluded (Gries and Otani, 2010). In Persian, however, no attention has been paid to the size adjectives. Therefore, this is a motivation for us, to investigate near-synonymy of ziyâd and farâvân size adjectives and their inflectional forms according to behavioral profile analysis approach.

The body of this paper, generally, structured as follow: In section 2, we examine briefly background study of adjectives, in particular, size adjectives, synonymy and the application of behavioral profile. In section 3, we investigate behavioral profile approach. Section 4 deals with methodology of research according to corpus-based data, process of annotation and random data selection by using Kerjcie & Morgan table (1970). In section 5, we analyze statistical information acquired from the process of annotation by using SPSS software. And finally in section 6, we present the results of data.



2. Methodology

Divjak and Gries (2006) identified the following four steps for the behavioral profile approach: 1. The retrieval of (a representative random sample of) all instances of the lemmas from a corpus. 2. Manual analysis and annotation of many properties of each match in the concordance of lemmas. 3. The conversion of data into a co-occurrence table. 4. The evaluation of the table by means of exploratory and other statistical techniques. In this study to do the above steps, we first used Antconc software to retrieve all matches of the lemmas ziyâd and farâvân plus their inflectional forms from 150 million-word corpus of Hamshahri 2 AleAhmad et al (2009). Then we annotated morphological, syntactic and semantic features to lexical forms in order to identify distributional features. And finally, we utilized SPSS software to evaluate the data and acquire Hierarchical agglomerative cluster charts.



3. Discussion and Results

No considerable study has been done on lexical relationships of size adjectives in Persian and only some grammarians and linguists such as Anvari and Givi (2008), Assi (2006) and etc simply have categorized size adjectives. Consequently, it is defficult to compare the achievements of this study with previous studies on near-synonymy.

In this article, generally, determining near-synonymy is performed according to the observation and analyzing similarities and differences of distributional behaviors of size adjectives and therefore is different from previous researches such as Abdolkarimi (2014, 2016) and etc. Our findings ,also, support Gries and Otani (2010) assumption on theoretical implication of behavioral profile analysis and most corpus-based works that believe formal differences reflect functional differences. For example, formal tag levels such as active, passive, plural, singular and etc have significant role in near-synonymy of “ziyâd and farâvân’’ inflectional forms. The findings of this study, moreover, clear that size adjectives (base form) or along with “tar” inflectional suffix are always dependent of a noun or a causative or predicative verb but size adjectives plus “tarin” superlative suffix can only be dependent of a noun not a verb, because no occurrence is observed for “tarin” as dependent of verbs.



4. Conclusion

In this study, we explored the semantic field of size by analyzing the distributional behavior of lexical forms "ziyâd, ziyâdi, ziyâdtar, ziyâdtarin, ziyâdtari, farâvân, farâvâni, farâvântar, farâvântarin, farâvântari" according to behavioral profile analysis and also three variable with 61 ID tags level were annotated to find near-synonymy relationship between size adjectives inflectional forms. The analysis of hierarchical agglomerative cluster charts all ID tags signified that mentioned lexical forms in terms of tag level made sub-groups which have near-synonymy between their own members sub-group. These sub-groups include "ziyâdi, farâvâni", "ziyâdtari, farâvântari", "ziyâdtar, farâvântar", "ziyâdi, farâvân", "ziyâdtarin, farâvântarin", "ziyâdi, farâvâni, ziyâdtari, farâvântari", "ziyâd, farâvân, ziyâdtarin, farâvântarin" and "ziyâdtar, farâvântar, ziyâd, farâvân, ziyâdtarin, farâvântarin".