Formulaic Language Project
This is a project to explore the factors involved in the measurement of repeated word sequences in language sampled from a range of corpora. Some of the factors investigated are the effects of nativeness and language proficiency (e.g. length of time as a writer of academic English) on the number and range of formulaic sequences in learner language.
We are working on a project examining the measurement of recurrent sequences of words across various corpora. The aim is to better understand the factors that affect the use of formulaic language across spoken and written language of different text types, by native and nonnative speakers and children.
The growth of research into phraseology, native-like formulaic language use, and the idiom principle in corpus linguistics, cognitive linguistics, psycholinguistics and applied linguistics (Ellis, 2008) calls for clarity of conceptualization across these fields (Gries, 2008) and, more so, for basic investigations into the metrics for operationalizing formulaicity in texts.
One of the primary aims of this project is to investigate the effect of a broad range of independent variables, including text- and corpus size, number of authors, register (spoken/written) and genre differences, vocabulary diversity (such as token-token and entropy) speaker age, nativeness and proficiency, upon different measures for identifying formulaic (recurrent) sequences. We are interested in both contiguous sequences (n-grams), limited span frameworks (phrase frames and skip-grams) and collocational clusters (item-sets and concgrams). The measures evaluated include frequency above a certain threshold, Mutual Information, t-score, gravity counts, various reference lists of formulas and various dispersion measures.
We want to determine how the different measures reflect the independent variables and how each measure contributes to the convergent and differential validity of operationalization. These exercises in corpus metrics usefully further the standardization of the measurement of formulaicity in language texts and they prove the foundations for triangulation with psycholinguistic definitions of formulaicity as well as for studies of first and second language acquisition, instruction, and evaluation.
References
Ellis, Nick C., Matthew Brook O’Donnell, Ute Römer, Stefan T. Gries & Stefanie Wulff. Measuring the formulaicity of language. Paper Presented at the American Association of Applied Linguistics Annual Conference 2009, Denver, CO, 21-24 March.
O’Donnell, Matthew Brook & Ute Römer. Proficiency development and the phraseology of learner language. Paper Presented at the 30th ICAME Conference 2009, Lancaster, UK, 27-31 May.
O’Donnell, M B, Römer, U., Ellis N. C. Examining formulaic sequences in corpora of second language writing. Presentation at SLRF 2009, Michigan State University, East Lansing, MI, USA, 29 October – 1 November 2009.
Coming soon
If you are an instructor at the University of Michigan and would like to learn more about corpus analysis and how to use corpora with your students, you may be interested in joining our Corpus Analysis Group.
The MCL team frequently provide introductions to corpus analysis and training in the use of corpus tools for scholars visiting the ELI and also in writing classes offered in the English Language Institute.
This is a project to explore the factors involved in the measurement of repeated word sequences in language sampled from a range of corpora.
These projects illustrate the range of research being carried out by members of the MCL team.
This page lists a number of useful corpus tools and corpora you can search online to retrieve examples of real spoken and written English.
On these pages you will find information about conferences members of the MCL team presented at or helped organize.