Talking Business


Most action in a business context is carried out through language and as such communication can be said to constitute the foundation of organisations. Language is not exclusively a medium through which we transmit information in both spoken and written form but it also contributes to our construction of identity and interpersonal relations. Consequently, language use is a form of social action and when we speak we therefore act. Today we increasingly do so on social media.

Over the last few years, I’ve been researching such speech acts in blogs and microblogs to show how they can be more easily identified in large data collections and how language has adapted to the online environment. While this is of interest regarding language use in general, my most recent research project is specifically aimed at the language used for business communication purposes on social media. Especially in the field of customer communication, it is essential to know which linguistic behaviour triggers positive or negative reactions and how to separate positive from negative comments in a reliable manner.

This is where corpus linguistic studies can help. Corpus linguistics facilitates, on the one hand, the quantitative analysis of large samples of data, while, on the other hand, allowing for the analysis to consider the context in which linguistic constructions appear. Thus, contrary to other methodological approaches, texts are not broken up into single words but studied in their entire, original form. This is important as function words, which are often not regarded as carrying significant meaning, may determine whether a comment is positive or negative (e.g. Your service is shit vs Your new app is the shit, see Lutzky and Kehoe 2016).

In our previous studies, we started out from the concept of collocation, which is the calculation of the probability with which words appear in each other’s vicinity, in order to distinguish between often very different uses of words. We then introduced the new calculation of shared and unique collocates, i.e. the comparison of collocates that a word shares with others and those that distinguish them from each other. This innovative approach to the study of collocation allows for different uses of the same word to be separated in large corpora in order to narrow down the sentiments expressed by bloggers and microbloggers. For more information, see Lutzky and Kehoe (2016, 2017).


The Virtual Theorist – a resource for studying literary theory

Over the summer I contributed to a project called The Virtual Theorist which aims to discuss and illustrate different approaches to the interpretation and analysis of literary texts. It introduces a variety of literary theories and shows how these theories can be applied when studying a specific literary text, in this case the poem ‘Goblin Market’ by Christina Rossetti.

My own contribution discusses the linguistic subdiscipline of Literary Linguistics or Stylistics and in the analysis of the poem I use a corpus stylistic methodology. While corpus stylistics is a methodology that has been applied, for example, in the study of novels and short stories, poems are a type of text that is usually not studied in this way. By using the tools available through WebCorp, I show how corpus linguistic means allow us to view a poem from a different perspective and add to our understanding of its themes, its use of word clusters or characterisation. You can find the introduction to Literary Linguistics, the analysis of the poem and a sample bibliography here:

(Note that this webpage was designed primarily for undergraduates.)

eMargin as an annotation tool for narrative analysis

This week I’m at the International Conference on Narrative at Manchester Metropolitan University. I will be talking about Early Modern English witness depositions and their narrative structure. Basically, I am studying the depositions given by witnesses which were recorded by scribes in written form, often before the actual proceedings were held in court. Now, while they were originally spoken texts, the scribes noted them down in written form and usually transformed them to some extent. For example, they would render them as third person narratives instead of first person ones and include different types of legal formulae and phrases in order to disambiguate certain references (see e.g. Kytö and Walker 2006, Kytö, Grund and Walker 2011).

For my study, I made use of the (collaborative) text annotation tool eMargin, which was developed by the Research and Development Unit for English Studies (RDUES) at Birmingham City University. eMargin is an online tool that is available for free and allows users to upload texts and then annotate them, for example, for qualitative linguistic analyses. It allows them to highlight parts of text in different colours, attribute functions to these colours and add comments to their annotations. In addition to colours, tags can be created and visualised in form of a tag cloud.

Now, while I am using eMargin for this project individually, it can also be used collaboratively through the function of groups. That is to say that several eMargin users can be added to the same group and then work on the annotation of a text collaboratively. While this is an option for specific research projects, it also lends itself very well for classroom activities. Thus, students could as part of reading and analysing a text, whether that is from a linguistic, literary or other point of view, annotate it together and thereby share their ideas on the use of particular constructions, themes, or passages.

Coming back to Early Modern English witness depositions, I used eMargin to annotate them specifically for their narrative structure. I started out from the Labovian (1972) narrative model which he derived from his recordings of spoken, first person narratives. While this model is based on contemporary data, I applied it to Early Modern English texts to see if the different sections of abstract, orientation, complicating action, resolution and coda can be identified in these primarly third person narratives too (see also Grund and Walker 2011). My main interest was, however, to find out what types of internal and external evaluation are attested in these texts and to draw some conclusions as to their purpose and use.


eMargin, developed by Andrew Kehoe and Matt Gee, Research and Development Unit for English Studies, Birmingham City University. Available online at

Grund, Peter J. and Walker, Terry. 2011. “Genre characteristics.” In Testifying to Language and Life in Early Modern England, Merja Kytö, Peter J. Grund and Terry Walker (eds), 15-56. Amsterdam: Benjamins.

Kytö, Merja and Walker, Terry. 2006. Guide to A Corpus of English Dialogues 1560-1760. Uppsala: Uppsala University.

Labov, William. 1972. Language in the Inner City: Studies in the Black English Vernacular. Philadelphia: University of Pennsylvania Press.