Set phrases around globalization : an experiment in corpus-based computational phraseology

We refer here to phraseology as the study of set phrases in the broadest sense, including partly fixed phrases (routines and formulae, collocations), and also very fixed phrases (idioms and proverbs). Contrary to other studies centred around collocations, computational phraseology (Heid 2007, Corpas Pastor 2013) investigates the close relationship between all categories of set phrases or phraseologisms, presents them as a continuum ranging from low to high idiomaticiy and/or fixedness, and considers set phrases as one of the foundation stones of linguistic theory. Corpus-based computational phraseology has recourse to huge corpora, as it has been demonstrated that, while set phrases as a whole represent a high percentage of word combinations in any text (Sinclair 1991), the frequency of a given set phrase in a corpus is usually low (Moon 1998). Computational phraseology largely relies on the statistics of word cooccurrence (Evert 2004), but also explores other techniques derived from information retrieval, such as clustering (Baeza-Yates & Ribeiro-Neto 1999). The aim of this study is to investigate whether the automatic extraction of set phrases around the term globalization yields interesting semantic information and may therefore contribute to a better understanding of how the notion is perceived by society in different languages. For the purpose of this experiment, a reasonable compromise had to be reached between the size of the corpora (as large as possible for phraseology extraction) and the time needed for computational processing (the larger the corpus, the slower the algorithm). We therefore used corpora of 200 million words (tokens) : a portion of the WaCky corpora (Baroni et al. 2009) for English, Italian, French and German. We then created similar web corpora of the same size for Spanish and Dutch, by using the WebBootCat tool provided by the Sketch Engine (www.sketchengine.co.uk). Apart from those 200 million word corpora in 6 languages, reflecting general language use, we also compiled 6 specialized corpora (one for each language), of 5 million words each, by using globalization and related terms as seed words with WebBootCat. The algorithm used here for the automatic extraction of phraseology from the corpora is non-parametric and does not directly derive from the theory of statistics. It is an adaptation of a well-known technique used in information retrieval (Baeza-Yates & Ribeiro-Neto 1999) : metric clusters. The results of this experiment show that, in the 6 languages, the notion of globalization has already pervaded several layers of phraseology in the broad sense, especially clichés, routines and formulae, as illustrated by : in the era of globalization (English), à l'heure de la mondialisation (French), nell'epoca della globalizzazione (Italian), la era de la globalización (Spanish). In all 6 languages, set phrases around globalization contain references to a few common notions such as time, context and (neo-)liberal. The most striking finding, however, is that corpora of the same size, analyzed by the same algorithm within the framework of computational phraseology, reveal crucial differences between the 6 language communities : set phrases highlighting some negative aspects of globalization, for instance, are much clearer in French, Italian, and Spanish than in English, whereas German and Dutch use very few negative set phrases with globalization. The experiment also confirms that computational phraseology provides useful additional tools for deriving semantic and sociolinguistic information from large linguistic corpora. References Baeza-Yates, R. & B. Ribeiro-Neto (1999). Modern Information Retrieval. New York: ACM Press, Addison Wesley. Baroni, M., Bernardini, S., Ferraresi, A. & E. Zanchetta. (2009). The WaCky Wide Web: A collection of very large linguistically processed Web-crawled corpora. In: Journal of Language Resources and Evaluation 43, p. 209-226. Corpas Pastor, G. (2013). Detección, descripción y contraste de las unidades fraseológicas mediante tecnologías lingüísticas. In Olza, I. & R. Elvira Manero (eds.) (2013), Fraseopragmática. Berlin, Frank & Timme, p. 335-373. Evert, S. (2004). The statistics of word cooccurrences -word pairs and collocations. Stuttgart, Universität Stuttgart. Heid, U. (2007). “Computational linguistc aspects of phraseology II”. In: Burger et al. (eds.), Phraseologie / Phraseology. An International Handbook of Contemporary Research. Berlin, New York, p. 1036-1044. Moon, R. (1998). Fixed Expressions and Idioms in English. Oxford, Oxford University Press. Sinclair, J. (1991). Corpus, Concordance, Collocation. Oxford, Oxford University Press