Explorations in the distributional and semantic similarity of words

A straightforward computation of the list of the words (the `tail words' of the list) that are distributionally most similar to a given word (the `head word' of the list) leads to the question: How semantically similar to the head word are the tail words; that is: how similar are their meanings to its meaning? And can we do better? The experiment was done on nearly 18,000 most frequent nouns in a Finnish newsgroup corpus. These nouns are considered to be distributionally similar to the extent that they occur in the same direct dependency relations with the same nouns, adjectives and verbs. The extent of the similarity of their computational representations is quantified with the information radius. The semantic classification of head-tail pairs is intuitive; some tail words seem to be semantically similar to the head word, some do not. Each such pair is also associated with a number of further distributional variables. Individually, their overlap for the semantic classes is large, but the trained classification-tree models have some success in using combinations to predict the semantic class. The training data consists of a random sample of 400 head-tail pairs with the tail word ranked among the 20 distributionally most similar to the head word, excluding names. The models are then tested on a random sample of another 100 such pairs. The best success rates range from 70% to 92% of the test pairs, where a success means that the model predicted my intuitive semantic class of the pair. This seems somewhat promising when distributional similarity is used to capture semantically similar words. This analysis also includes a general discussion of several different similarity formulas, arranged in three groups: those that apply to sets with graded membership, those that apply to the members of a vector space, and those that apply to probability mass functions.

Beyond Words : The European Court of Justice and Legal Certainty in Multilingual EU Law

This study addresses the issue of multilingualism in EU law. More specifically, it explores the implications of multilingualism for conceptualising legal certainty, a central principle of law both in domestic and EU legal systems. The main question addressed is how multilingualism and legal certainty may be reconciled in the EU legal system. The study begins with a discussion on the role of translation in drafting EU legislation and its implications for interpreting EU law at the European Court of Justice (ECJ). Uncertainty regarding the meaning of multilingual EU law and the interrelationship between multilingualism and ECJ methods of interpretation are explored. This analysis leads to questioning the importance of linguistic-semantic methods of interpretation, especially the role of comparing language versions for clarifying meaning and the ordinary meaning thesis, and to placing emphasis on other, especially the teleological, purpose-oriented method of interpretation. As regards the principle of legal certainty, the starting-point is a two-dimensional concept consisting of both formal and substantive elements; of predictability and acceptability. Formal legal certainty implies that laws and adjudication, in particular, must be predictable. Substantive legal certainty is related to rational acceptability of judicial decision-making placing emphasis on its acceptability to the legal community in question. Contrary to predictability that one might intuitively relate to linguistic-semantic methods of interpretation, the study suggests a new conception of legal certainty where purpose, telos, and other dynamic methods of interpretation are of particular significance for meaning construction in multilingual EU law. Accordingly, the importance of purposive, teleological interpretation as the standard doctrine of interpretation in a multilingual legal system is highlighted. The focus on rational, substantive acceptability results in emphasising discourse among legal actors among the EU legal community and stressing the need to give reasons in favour of proposed meaning in accordance with dynamic methods of interpretation including considerations related to purposes, aims, objectives and consequences. In this context, the role of ideal discourse situations and communicative action taking the form of interaction among the EU legal community in an ongoing dialogue especially in the preliminary ruling procedure is brought into focus. In order for this dialogue to function, it requires that the ECJ gives persuasive, convincing and acceptable reasons in justifying its decisions. This necessitates transparency, sincerity, and dialogue with the relevant audience.

Words as Events : Cretan Mantinádes in Performance and Composition

Words as Events introduces the tradition of short, communicative rhyming couplets, the mantinádes, which are still sung and recited in a variety of performance situations on the island of Crete. The local focus on communicative economy and artistry is further examined in an in-depth analysis of the processes and ideals of composition. Short genres of oral poetry have been widely neglected in folklore research; however, their demand for structural and semantic coherence as well as their dialogic nature appeals to very different human needs from those of the longer poems. In contemporary Crete, poems also appear in written contexts, they are submitted to modern mass media, and people widely exchange them as text messages. By striving to understand the tradition during a period of change, this study analyzes the larger principles that ground the communication, self-expression and creativity in the genre. The aim of this research is thus twofold: to present this specific register of dialogic oral poetry as well as to create a theoretical approach sensitive to the creativity of such short registers. The study is based on long-term ethnographic fieldwork conducted in Crete between the years 1997 2009. An important aspect of the methodology was to create long-term ties with a number of key-informants who composed mantinádes. The interdisciplinary theoretical and methodological basis for this analysis is in the contemporary Finnish and international research on oral poetry and on anthropological research on communicative speech genres. These theoretical insights are extended by addressing questions of spontaneity and individual agency. Since mantinádes are at the same time a model for composition and a reserve of poems in fixed form, the aesthetics and objectives of performance and composition are also plural. In a traditional singing event, the basic motivation for the singers is to provide a meaningful contribution to a selected theme, which is the shared topic of the poetic dialogue. Consequently, similar topics giving rise to poetic associations can be encountered during various moments of everyday life: the dialogic nature is embodied in the tradition to such degree that it also arises in the performances and composition of single poems and outside of any institutionalized performance arenas. Therefore, as this study discusses in detail, even the apparently non-contextualized poems recited between locals or occurring in the contemporary mass media arenas, are understood and evaluated as utterances that provide an individual perspective within a certain dialogue.

Neurocognitive processing of inflected and derived words

The representation of morphologically complex words in the mental lexicon and their neurocognitive processing has been a vigorously debated topic in psycholinguistics and the cognitive neuroscience of language. This thesis investigates the effect of stimulus modality on morphological processing, the spatiotemporal dynamics of the neural processing of inflected (e.g., work+ed ) and derived (e.g., work+er ) words and their interaction, using the Finnish language. Overall, the results suggest that the constituent morphemes of isolated written and spoken inflected words are accessed separately, whereas spoken derived words activate both their full form and the constituent morphemes. The processing of both spoken and written inflected words elicited larger N400 responses than monomorphemic words (Study I), whereas the responses to spoken derived words did not differ from those to monomorphemic words (Study IV). Spoken inflected words elicited a larger left-lateralized negativity and greater source strengths in the left temporal cortices than derived words (Study IV). Thus, the results suggest different cortical processing for derived and inflected words. Moreover, the neural mechanisms underlying inflection and derivation seem to be not only different, but also independent as indexed by the linear summation of the responses to derived and inflected stimuli in a combined (derivation+inflection) condition (Study III). Furthermore, the processing of meaningless, spoken derived pseudowords was more difficult than for existing derived words, indexed by a larger N400-type effect for the pseudowords. However, no differences were observed between meaningful derived pseudowords and existing derived words (Study II). The results of Study II suggest that semantic compatibility between morphemes seems to have a crucial role in a successful morphological analysis. As a methodological note, time-locking the auditory event-related potentials/fields (ERP/ERF) to the suffix onset revealed the processes related to morphological analysis more precisely (Studies II and IV), which also enables comparison of the neural processes in different modalities (Study I).