Anastasios Asimakopoulos

    Dear @rmwebb @judith-gorham @vera-t @ziyinini @dbtekid @hilary @ana93 and @babruwan Thank you so much for your contributions! It was a real pleasure reading everyone’s ideas.

    Below is a summary of what you mentioned with regard to the purpose/benefits of the activity/tool as well as some details I have added to clarify things. Have a look and do let me know if you need further clarification – happy to answer your questions.


    • Summarising the most typical collocates in a visual way (the four circles were generated by using the visualisation tool from Word Sketch and they summarise collocates of ‘develop’ in four different subcorpora)
    • Identifying associations between collocates of a word e.g. infection, disease, cancer, etc. → same semantic field (almost like a ‘magnet’); vocabulary does not exist in a vacuum. Guesses are due to lexical priming.
    • Collocates give information on the aboutness of the subcorpora (disciplinary categories) e.g. infection, disease, cancer, vaccine, resistance, etc. → Life Sciences.
    • The idea of ‘zooming out’ to see the big picture first before ‘zooming in’ specific examples = typical pattern in corpus linguistics research that combines quantitative (collocates determined statistically) and qualitative (collocates examine in context in concordance lines)
    • Identifying collocational similarities and differences of a term between different subcorpora (AH, SS, PS, LS) e.g. theory, model, method, strategy were found in various disciplinary groups, but can still differ in the way they are used in sentences.
    • Fun and engaging problem-solving activity – interactive

    Your ideas

    • Free writing practice e.g. sentences or paragraphs using the collocates.
    • Students choose a subcorpus that represents their discipline. Further study of context using Concordance i.e. how does a shared collocate differ? For example, develop + theory in LS vs develop + theory in AH with a focus on subjects i.e. Who develops theories in SS? Who develops theories in AH?
    • Adapting to include fewer collocates – that’s possible by controlling the number of collocates on the visual but that does not necessarily remove the general ones and only include discipline-specific collocates. It is very important to consider how many words and which ones to remove as this affects the difficulty of the task. For example, I didn’t think it’s worth removing theory and asking people to drag it to the circles three times – that would have resulted in doing the task mechanically as opposed to trying to make connections.
    • Adapting to examine discipline-specific collocates e.g. Chemistry as opposed to those from the disciplinary group e.g. Physical Sciences. Unfortunately, this is only possible with a full membership/ institutional login. However, there is a way around it though. You can use Concordance and Text Types to find a word in a specific discipline, then use the Collocations button from the concordance lines (top right). This approach, however, requires more decisions e.g. selecting the window span you are interested in (L5-R5, L3-R3, R1-R5, etc.) as well as the association measure. T-score is similar to frequency so it will give you a lot of grammatical words for collocates. MI is for exclusive but rare combinations so it favours really rare words in the corpus. LogDice, on the other hand, focusses on exclusive but frequent collocates, and it’s the association measure that Word Sketch uses. It is also the only standardised measure (with a theoretical maximum value of 14), so this gives you a frame of reference, unlike MI and t-score that require cut off points.

    If you are arriving a little late, feel free to add more ideas/comments or respond to the ones I and everyone else has shared.