Barbara McGillivray

Methods in Latin Computational Linguistics

Historical languages, corpora, and computational methods -- Computational resources and tools for Latin -- Verbs in corpora, lexicon ex machina -- The agonies of choice: automatic selectional preferences -- A closer look at automatic selectional preferences for Latin -- A corpus-based foray into Latin preverbs -- Statistical background to the investigation on preverbs -- Latin computational linguistics.

Semantic Structure from Correspondence Analysis

A common problem for clustering techniques is that clusters overlap, which makes graphing the statistical structure in the data difficult. A related problem is that we often want to see the distribution of factors (variables) as well as classes (objects). Correspondence Analysis (CA) offers a solution to both these problems. The structure that CA discovers may be an important step in representing similarity. We have performed an analysis for Italian verbs and nouns, and confirmed that similar structures are found for English.

Computational valency lexica for Latin and Greek in use: a case study of syntactic ambiguity

Abstract We have built a corpus-driven valency lexicon for Greek verbs by following an approach devised for Latin data. We have then used the lexicon to detect a specific type of potentially ambiguous syntactic patterns in Latin and Greek hexametric poetry, which can consistently be disambiguated by prosodic breaks. Such disambiguating breaks were then mapped onto the metrical structure of the lines containing the ambiguous patterns, in order to assess their correspondence to metrical boundaries and to gather independent evidence on the phonetic nature of the boundaries themselves in view of further investigation. From a methodological point of view, the lexica have enabled us to draw on a relatively large set of texts to study a rare phenomenon and to establish a semi-automatic procedure that can be replicated on larger and compatible corpora.

Latin Computational Linguistics

Computational linguistics and quantitative corpus linguistics offer valuable resources and techniques for doing what Latin linguists have always done. The crucial frequency information contained in the valency lexicon allows for further corpus-based synchronic and diachronic investigations on verbal argument structure which account for language usage. This chapter contributes to advances in Latin Linguistics and also aims at inspiring computational linguists. The approach relies on the relations of synonymy and hyperonymy as defined in Latin WordNet, while the latter measures the similarity between two words in terms of the syntactic and lexical contexts. The chapter exploits existing collections of Latin texts, aiming at showing how new resources can be built from these collections and how they can be quantitatively analysed. The methodological contribution in the chapter is intrinsically cross-disciplinary, because it defines Latin computational linguistics from the combination of computational methods and Latin data.Keywords: corpus-based synchronic and diachronic investigations; Latin computational linguistics; Latin valencylexicons; Latin WordNet; quantitative corpus linguistics