Applied Corpus Linguistics最新文献

In this paper, we discuss a variety of misunderstandings that have arisen – and still linger – in the field of Law and Corpus Linguistics (LCL). Many have to do with the interdisciplinary nature of legal scholarship and practice on the one hand and corpus linguistics (CL) on the other. Our goals are to address these misunderstandings to explicate them, illuminate the assumptions that co-motivated them in the first place, and provide advice as to how to discuss, maybe refute, and avoid them moving forward, especially given the progress made to-date. In order to illustrate our discussion, we have separated the critiques into two major stages in the collaborative process – (i) a legal stage and (ii) a corpus linguistics stage. In stage (i), we address issues such as the desire to involve a corpus linguist, the question of whether the use of CL outsources a judicial task, and the role CL plays in legal theories of interpretation. In stage (ii), we discuss common critiques of CL applications to legal interpretation such as the claim that the method is inherently subjective, the potential arbitrariness of corpus compilation and selection, and the variable role that context plays in such applications. The final section provides our set of recommendations connecting the two stages to allow for the iterative fine-tuning process we think is required for successful collaboration in academic and applied legal settings; we conclude with our view on who should do corpus linguistics in legal contexts, hopefully facilitating further talk across the interdisciplinary aisle.

Writing proficiency is associated with linguistic complexity. We used measures of linguistic complexity to investigate the development of children's narrative writing using a large corpus of short stories (N>100,000) written by children aged 5–13 in the UK. Linguistic complexity was assessed using both lexical (N = 30) and syntactic (N = 14) measures. Most measures were associated with age, with writing by older children showing greater lexical density, sophistication, and diversity than writing by younger children. Older children also used longer sentences, and longer T-units and clauses, and the density of smaller syntactic units inside larger units was also higher. Principal Component Analysis identified a number of dimensions associated with complexity, with the first two dimensions capturing nearly 50 % of variance. Lexical diversity was mainly represented on the first dimension and syntactic complexity on the second. Across the age range, there was wider variation in syntactic complexity than in lexical diversity, suggesting that syntactic development is subject to more individual differences than the ability to use a diverse set of lexical items. Our findings quantify the nature and content of children's writing through mid-childhood, and we discuss the utility of analysing children's writing using a computational, data-driven approach.

Collocational competence is essential for all learners, particularly for multilingual learners. This corpus-driven study assembled a 474-collocation list from a digital science corpus compiled from several thousand middle school science resources. Using a corpus of more than 2.7 million tokens and more than 400 node words, the collocation list was extracted by combining two approaches: frequency-based and expert-judged. The Digital Science Collocations List (DSCL) provides middle school learners and teachers with an unprecedented resource covering Life Science, Physical Science, and Earth and Space Science. This list may be especially useful to multilingual learners as most of the collocations in this list are composed of patterns that they struggle with (e.g., adjective + noun and verb + noun).

This article offers an overview of developments in a newly emerging interdisciplinary research field: legal corpus linguistics. The field brings together corpus research and legal theory by applying corpus-analytic techniques and linguistic concepts to facilitate the interpretation of legal texts. Despite the field's short history, it has already contributed important new insights into the meaning of statutory texts and parts of the U.S. Constitution, insights that may have significant practical implications for the American legal system. Our article provides an overview of relevant developments in legal corpus linguistics, from early success stories to recent and ongoing collaborative work between corpus linguists and legal scholars. It aims to highlight the benefits and illustrate the potential of this type of interdisciplinary work by summarizing three recent case studies, each of which deals with an important topic in American constitutional law. The case studies focus in turn on the following parts of the U.S. Constitution: (1) Article III and the meaning of “cases,” (2) the Appointments Provision in Article II, Section 2 and the meaning of “such inferior officers,” and (3) the Impeachment Provision in Article II, Section 4 and the meaning of “misdemeanors.” All three case studies use corpus analysis to explore phraseological patterns in large collections of Founding Era texts to provide insights into the meanings of the selected words and phrases in context during the time the Constitution was drafted and ratified. The article discusses the practical relevance of results from these case studies and potential implications of this and related work in legal corpus linguistics for contemporary and future litigation.

When the meaning of an ambiguous word, phrase or grammatical structure in a statutory provision is disputed, courts are tasked with identifying the best meaning of the contested language. A common method of resolving linguistic ambiguities is to investigate the meaning of the contested word or structure in statutory provisions with similar subject matter. While the subject matter of a text has a demonstrated effect on language use, register variation research shows that the function of a text is also highly influential in predicting linguistic variation. Thus far, the function of a statutory provision (e.g., obligation to act, authorization to act) has not been considered in legal interpretative research. In the present study, I investigate the extent to which function influences the lexico-grammatical characteristics of statutory texts. 2,573 statutory provisions from the Arizona State Code are individually assigned to one of seven categories representing their function: Duties, Permissions, Impersonal Rules, Operational Definitions, Prohibitions, Procedural Guidelines, and Criminal Offenses. Key feature analysis is used to identify and describe patterns of lexico-grammatical variation between the seven functional types. Results reveal a great deal of lexico-grammatical variation associated with function in the register of statutory law. Furthermore, some functional types of statutory provisions are more linguistically distinct than others. These findings suggest that it may be beneficial to consider communicative function when investigating legal interpretative questions.

The goal of this study is to assess the degree of resemblance between texts generated by artificial intelligence (GPT) and (written and spoken) texts produced by human individuals in real-world settings. A comparative analysis was conducted along the five main dimensions of variation that Biber (1988) identified. The findings revealed significant disparities between AI-generated and human-authored texts, with the AI-generated texts generally failing to exhibit resemblance to their human counterparts. Furthermore, a linear discriminant analysis, performed to measure the predictive potential of dimension scores for identifying the authorship of texts, demonstrated that AI-generated texts could be identified with relative ease based on their multidimensional profile. Collectively, the results underscore the current limitations of AI text generation in emulating natural human communication. This finding counters popular fears that AI will replace humans in textual communication. Rather, our findings suggest that, at present, AI's ability to capture the intricate patterns of natural language remains limited.

When faced with a word or phrase that is not defined in a statute, judges generally interpret the language of the law as it is likely to be understood by an ordinary user of the language. However, there is little agreement about what ordinary meaning is and how it can be determined. Proponents of corpus-based legal interpretation argue that corpora provide scientific rigor and increased validity and transparency, but there is currently no consensus on best practices for legal corpus linguistics. Our objective in this paper is to propose some refinements to the theory of ordinary meaning and corpus-based methods of analyzing it. We argue that the scope of legal language is established by conceptual (intensional) meaning, and not limited to attested referents. Yet, most current corpus-based approaches are purely referential (extensional). Therefore, we introduce a new methodology—prototype by component (PBC) analysis—in which we bring together aspects of the componential approach and prototype theory by assuming that categories are gradient entities that are characterized by gradient semantic components. We introduce the analytical steps in PBC analysis and apply them to Nix v. Hedden (1893) to determine whether tomato is a member of the category vegetable. We conclude that conceptual categories have a prototypical reality and a componential reality. As a result, attested referents in a corpus can provide insights into the conceptual meaning of terms and the degree to which concepts are members of categories.

Scholars and practitioners interested in legal interpretation have become increasingly interested in corpus-linguistic methodology. Lee and Mouritsen (2018) developed and helped popularize the use of concordancing and collocate displays (of mostly COCA and COHA) to operationalize a central notion in legal interpretation, the ordinary meaning of expressions. This approach provides a good first approximation but is ultimately limited. Here, we outline an approach to ordinary meaning that is intensionalist (i.e., 'feature-based'), top-down, and informed by the notion of cue validity in prototype theory. The key advantages of this approach are that (i) it avoids the which-value-on-a-dimension problem of extensionalist approaches, (ii) it provides quantifiable prototypicality values for things whose membership status in a category is in question, and (iii) it can be extended even to cases for which no textual data are yet available. We exemplify the approach with two case studies that offer the option of utilizing survey data and/or word embeddings trained on corpora by deriving cue validities from word similarities. We exemplify this latter approach with the word vehicle on the basis of (i) an embedding model trained on 840 billion words crawled from the web, but now also with the more realistic application (in terms of corpus size and time frame) of (ii) an embedding model trained on the 1950s time slice of COHA to address the question to what degree Segways, which didn't exist in the 1950s, qualify as vehicles in this intensional approach.

This paper explores the potential of generative artificial intelligence technology, specifically ChatGPT, for advancing corpus approaches to discourse studies. The contribution of artificial intelligence technologies to linguistics research has been transformational, both in the contexts of corpus linguistics and discourse analysis. However, shortcomings in the efficacy of such technologies for conducting automated qualitative analysis have limited their utility for corpus approaches to discourse studies. Acknowledging that new technologies in data analysis can replace and supplement existing approaches, and in view of the potential affordances of ChatGPT for automated qualitative analysis, this paper presents three replication case studies designed to investigate the applicability of ChatGPT for supporting automated qualitative analysis within studies using corpus approaches to discourse analysis.

The findings indicate that, generally, ChatGPT performs reasonably well when semantically categorising keywords; however, as the categorisation is based on decontextualised keywords, the categories can appear quite generic, limiting the value of such an approach for analysing corpora representing specialised genres and/or contexts. For concordance analysis, ChatGPT performs poorly, as the results include false inferences about the concordance lines and, at times, modifications of the input data. Finally, for function-to-form analysis, ChatGPT also performs poorly, as it fails to identify and analyse direct and indirect questions. Overall, the results raise questions about the affordances of ChatGPT for supporting automated qualitative analysis within corpus approaches to discourse studies, signalling issues of repeatability and replicability, ethical challenges surrounding data integrity, and the challenges associated with using non-deterministic technology for empirical linguistic research.