Large Newspaper Corpus Research Articles

Prescribing French: A corpus-linguistic approach to official terminology in French newspapers

ABSTRACTIn France, English is often perceived as a negative influence on the language in the eyes of purist institutions like the French Academy. Terminological commissions have been established to replace foreign expressions with French terminology that is regularly published in the Journal officiel de la République française. Although the Toubon Law of 1994 prescribes the use of this terminology in government publications, speakers are merely encouraged to do so. This article investigates the variation between English lexical borrowings and their prescribed equivalents in a large newspaper corpus containing articles from 2000 to 2017 in order to see whether formal written language complies with the purist recommendations. Time is treated with a new dynamic approach: the probability of using a prescribed term is estimated three years before and three years after official prescription. Fifty-four target terms are selected from the lexical fields of computer science, entertainment industry and telecommunication, including emblematic prescribed words such as courriel and mot-dièse. The analysis reveals that prescription is only effective when it follows already attested use. Furthermore, conservative newspapers show higher proportions of recommended terminology, especially as compared to newspapers specializing in technology.

A social-event based approach to sentiment analysis of identities and behaviors in text

ABSTRACTWe describe a new methodology to infer sentiments held toward identities and behaviors from social events that we extract from a large corpus of newspaper text. Our approach draws on affect control theory, a mathematical model of how sentiment is encoded in social events and culturally shared views toward identities and behaviors. While most sentiment analysis approaches evaluate concepts on a single, evaluative dimension, our work extracts a three-dimensional sentiment “profile” for each concept. We can also infer when multiple sentiment profiles for a concept are likely to exist. We provide a case study of a large newspaper corpus on the Arab Spring, which helps to validate our approach.

Construction and evaluation of event graphs

AbstractEvents play an important role in natural language processing and information retrieval due to numerous event-oriented texts and information needs. Many natural language processing and information retrieval applications could benefit from a structured event-oriented document representation. In this paper, we proposeevent graphsas a novel way of structuring event-based information from text. Nodes in event graphs represent the individual mentions of events, whereas edges represent the temporal and coreference relations between mentions. Contrary to previous natural language processing research, which has mainly focused on individual event extraction tasks, we describe a complete end-to-end system for event graph extraction from text. Our system is a three-stage pipeline that performs anchor extraction, argument extraction, and relation extraction (temporal relation extraction and event coreference resolution), each at a performance level comparable with the state of the art. We presentEvExtra, a large newspaper corpus annotated with event mentions and event graphs, on which we train and evaluate our models. To measure the overall quality of the constructed event graphs, we propose two metrics based on the tensor product between automatically and manually constructed graphs. Finally, we evaluate the overall quality of event graphs with the proposed evaluation metrics and perform a headroom analysis of the system.

A word clustering approach to domain adaptation: Robust parsing of source and target domains

We present a technique to improve out-of-domain statistical parsing by reducing lexical data sparseness in a PCFG-LA architecture. We replace terminal symbols with unsupervised word clusters acquired from a large newspaper corpus augmented with target domain data. We also investigate the impact of guiding out-of-domain parsing with predicted part-of-speech tags. We provide an evaluation for French, and obtain improvements in performance for both non-technical and technical target domains. Though the improvements over a strong baseline are slight, an interesting result is that the proposed techniques also improve parsing performance on the source domain, contrary to techniques such as self-training, thus leading to a more robust parser overall. We also describe new target domain evaluation treebanks, freely available, that comprise a total of about 3,000 annotated sentences from the medical domain, regional newspaper articles, French Europarl and French Wikipedia.

Degrees of grammaticalization in three Romance languages: A comparative analysis of existential constructions

Grammaticalization is generally assumed to be a gradual process and languages are thought to be grammaticalized to different degrees. For Romance languages, it has often been claimed that French is far more grammaticalized than Spanish, and that Italian occupies an intermediate position between these two. The aim of this study is to investigate to what extent this claim applies to existential constructions. In present-day French, Italian and Spanish, the most frequent existentials are il y a, c’e and hay respectively, all three being highly grammaticalized expressions. My analysis of these constructions will be based on the six parameters of grammaticalization as defined by Lehmann (1985, 1995), looking at instances drawn from a large newspaper corpus. The discussion will suggest that, as expected, French existential constructions display several signs of a highly advanced grammaticalization process which their Spanish and Italian counterparts lack. Moreover, it will be shown that the existential construction is least grammaticalized in Italian.

Deverbal nouns, lexicalization and syntactic change

Deverbal nouns are often referred to as hybrids, containing both verbal and nominal features. In many languages deverbal nouns tend to develop into what Vendler (1967) called perfect nouns. Various stages of this development in Norwegian are proposed, drawing on Grammaticalization Theory and Lexicalization Theory. Frequency data are provided from a large newspaper corpus. The deverbal nouns are analyzed as going through a process of lexicalization where reduction in compositionality, reduced token frequency and increased idiosyncrasy are central elements. This process is triggered by relevance to the root, language use, isomorphism and the maximal difference principle. The article shows that Norwegian deverbal nouns are in a state of flux and that even different members of the same morphological type may behave quite differently syntactically and semantically.

Toward Automatic Determinationof the Semantics of Connectivesin Large Newspaper Corpora

We explored the possibility of using automatic techniques to analyze the use of backward causal connectives in large Dutch newspaper corpora. With the help of 2 techniques, Latent Semantic Analysis and Thematic Text Analysis, the contexts of more than 14,000 connectives were studied. The method of analysis is described. We found that differences that have been suggested in the literature via hand-based analyses between these types of connectives (e.g., on dimensions such as subjectivity, change in perspective, and factuality of the connected segments) also appear in our corpus of 16.5 million words.

Lexical Frequencies in a 300 Million Word Corpus of Australian Newspapers. Analysis and Interpretation

Corpus linguistics, descriptive, sociolinguistics, and psycholinguistics use corpora and generalise their findings beyond the samples contained in them. That raises the problem of the representativity of the data base and of the application of methods for the presentation of findings. Although this paper originated in the context of the pluricentricity of English in the lexis of mainstream Australian English (mAusE), it was inspired by the current debates about corpus methodology (Kretzschmar et al. 1987). It is based on a large newspaper corpus that extends over a period of six years. It studies the distribution patterns of a small set of lexical items that are derived from Aboriginal languages or relate to Aboriginal concerns. While there appears to be a fairly consistent stable core, these items manifest significant differences in occurrence over the six-year period and in the media outlets and that raises the questions of what a replicate study of these items (or of others) would find and whether a corpus can claim to be representative in the first place.