|
چکیده
|
Abstract Over the past decades, Corpus Linguistics (CL) has played an influential role in Applied Linguistics (AL). Software tools and corpus analytic techniques have allowed researchers to address previously intractable linguistic areas. However, despite flourishing nature, corpus-based studies have encountered serious debates over their methodological shortcomings in recent years. Accordingly, the present study was designed to appraise the methodological rigor and transparency in 2328 (16,896,000 words) corpus-based research articles published in 22 AL journals from 1986 to 2016. Based on the results of the pilot studies and previous literature, two coding schemes were developed, namely ‘domain-specific’ and ‘design-specific’. They covered 35 categories and 83 items in four main areas: (1) research article identification and research design, (2) characteristics of corpora, (3) corpus analytic techniques and approaches, and (4) reporting indicators of transparency. Analyses of the domain-based and design-based characteristics revealed that the corpus-based research articles, despite some improvements, such as increasing trends in reference corpora, diachronic analysis, and ethical issues, suffered from significant flaws. Lack of research questions, incomplete reporting practices regarding sampling strategies, lack of transparency regarding recall, precision, and reliability, low frequency of annotated corpora and software tools were among the severe weaknesses observed in such studies. The results also indicated that about 90% of corpus-based studies did not report their used analytics techniques. Compatible with the findings, the study puts forward several theoretical and practical implications and suggestions such as developing standards and guidelines for ethical issues and methodological rigor and integrity in quantitative, qualitative, and mixed-method corpus-based research articles. Keywords: Corpus-based study, methodological synthesis, corpus linguistics methodology, corpus-based research article, study quality
|