Massimo Moneglia

C-ORAL-ROM : integrated reference corpora for spoken Romance languages

1. Acknowledgements 2. Preface (by Blanche-Benveniste, Claire) 3. 1. The C-ORAL-ROM resource (by Moneglia, Massimo) 4. 2. The Italian corpus (by Cresti, Emanuela) 5. 3. The French corpus (by Campione, Estelle) 6. 4. The Spanish corpus (by Moreno-Sandoval, Antonio) 7. 5. The Portuguese corpus (by Bacelar do Nascimento, Maria Fernanda) 8. 6. Notes on lexical strategy, structural strategies and surface clause indexes in the C-ORAL-ROM spoken corpora (by Cresti, Emanuela) 9. 7. Appendix (by Moneglia, Massimo) 10. Bibliography

Plagiarism Detection through Multilevel Text Comparison

The paper presents the implementation of a tool for plagiarism detection developed within the AXMEDIS project. The algorithm leverages the plagiarist behaviour, which is modeled as a combination of 3 basical actions: insertion, deletion, substitution. We recognize that this behaviour may occur at various level of the document structure: the plagiarist may insert, delete or substitute a word, period or a paragraph. The procedure consists in two main steps: document structure extraction and plagiarism function calculation. We propose a recursive plagiarism evaluation function to be evaluated at each level of the document structure which is based on the Levenshtein edit distance. We also propose a method that will eliminate unnecessary chunks comparison, avoiding similarity calculation of chunks which do not share enough 4-grams. We describe the similarity algorithm and discuss some implementation issues and future work

Spoken corpora and pragmatics

The goal of this paper is to present arguments in favour of two points related to the study of oral corpora and pragmatics: a) at the level of annotation, corpora must ensure the parsing of the speech flow into utterances on the basis of prosodic cues and provide an easy access to the acoustic source; b) at the level of sampling, corpora must ensure the maximum representation of context variation, rather than speaker variation. We will present the reasons which support the very basic prosodic annotation of speech (prosodic boundaries) as a means to obtain relevant data from the speech flow. Starting from our present knowledge about the distribution of speech acts types in spoken corpora, we will present the reasons why building corpora in accordance to a context variation strategy should expand our knowledge of pragmatics. Additionally, we will claim that prosody is the necessary interface between locutive and illocutive acts and we will show that a deeper prosodic analysis is necessary to grasp unknown speech act types from language usage. Finally, we will briefly sketch the main assumptions of the Language into Act Theory (CRESTI, 2000) which is dedicated to the link between prosody and pragmatics and helps make explicit core aspects of pragmatic knowledge.

Keyword extraction in open-domain multilingual textual resources

The paper presents three methods for keyword extraction in open-domain multilingual textual resources, which will be implemented within the AXMEDIS framework. The procedures exploit, in an integrated architecture, statistic analysis and linguistic resources/tools, allowing document indexing and retrieval through high-level descriptors.

Natural Language Ontology of Action: A Gap with Huge Consequences for Natural Language Understanding and Machine Translation

Action verbs are the less predictable linguistic type for bilingual dictionaries and they cause major problems for MT technologies that are immediately evident to the user. This is not only because of language specific phraseology, but is rather a consequence of the peculiar way each natural language categorizes events i.e. it is a consequence of semantic factors. In ordinary languages the most frequent Action verbs are “general”, since they extend productively to actions belonging to different ontological types. Moreover, each language categorizes action in its own way and therefore the cross-linguistic reference to everyday activities is puzzling. But the actual variation of verbs across action types is largely unknown. This paper sketches the problem constituted by the Ontology of Action when disambiguation and cross-linguistic reference to action is concerned and presents the IMAGACT Ontology Infrastructure, which aims at filling this gap by exploiting multilingual spoken corpora.

Multi-Term Keywords for Indexing Multilingual Textual Repositories: Developing Language Resources and Algorithms.

The tool for keyword extraction developed within the AXMEDIS project have been designed for working in a multilingual environment and new algorithms have been developed to generate keywords with higher representativeness for content search and identification. The paper specifies the linguistic criteria followed for building language resources for French, Italian, and German, which are comparable to those available for English and the algorithms used for enhancing keyword performance. The new algorithm extracts keyword collocations from the document and generates multiword keywords. The use of multi-term descriptors is a good means to better identify the content. The precision obtained with respect to mono-term keyword increase the performance of 100% relative factor

Le unità di riferimento per l'espressione delle emozioni : dati dal corpus C-ORAL-ROM

SOMMARIO Il lavoro propone un sondaggio sulla manifestazione acustiche delle emozioni nel corpus di parlato spontaneo C-ORAL-ROM (Cresti&Moneglia, 2005) e verifica la pertinenza alla loro annotazione della nozione di enunciato (Cresti,2000). Nel lavoro viene mostrato che espressione emotiva negli enunciati ha forte definizione semantica, ma che la presenza delle emozioni propriamente dette (Damasio, 2000) e sporadica (frequente solo nelle sessioni in cui la stimolazione contestuale e particolarmente forte). Sulla base della descrizione di esempi tipici si evidenzia che le proprieta acustiche con cui l’emozione e espressa sono qualita olistiche degli enunciati e non influiscono sugli indici prosodici che esplicano il valore illocutivo. La ricerca mostra anche che le manifestazioni espressive (enfatiche) che si riscontrano in modo pervasivo nel parlato spontaneo non esprimono propriamente emozioni e parallelamente pertengono, non l’enunciato (unita minima interpretabile pragmaticamente), ma le unita di informazione in cui questo si articola (Cresti, 2000).