Key n-grams, key lemmas

[alex boulton]

Dear Laurence

You gave me a nice workaround to create key n-grams lists with AntConc 3, but I don't see how to this could function with AntConc 4. Any tips would be gratefully received!

I've also been looking at key lemmas as opposed to keywords, I was wondering if anyone had an opinion on this.

Best wishes

alex

PS I'm still finding v4 wonderful, thanks so much for all your work

[elco...@gmail.com]

Hi,

I think you can simply create frequency lists of clusters or lemmas, save them, and upload them in the corpus manager as target or reference corpora, and then use the Keyword analysis option.

Hope this works.

[Laurence Anthony]

HI Alex and all,

Sorry for the very slow response. In AntConc 4.2, it was quite difficult to get the word list formatting correct to load word lists as corpora, which also made it difficult to do the same thing for n-grams and clusters. So, today, I released AntConc 4.2.1, which addresses this issue. 

Now, it should be straightforward to load a corpus comprised of a word list, n-gram list, cluster list or even collocate list. So, to generate key n-grams, key clusters, or key collocates, just follow the steps below:

1. Generate your target and reference lists (clusters, n-grams, collocates, words) using the standard AntConc tools. 

2. Making sure that "All results" are showing, export the results via the File Menu (or simply copy the results table to a text file directly)

3. In the Corpus Manager, choose the word list option to create a corpus, and then load in the respective tables for your target and reference corpora to create new word list corpora for each of them. I recommend you name your new corpora appropriately, e.g. BE06_2-grams.

4. Select your newly created word list corpora as the target and reference corpora, and in the main display, generate keywords, just as you would for a regular corpus.

Let me know how it goes. I've tried it here and everything went smoothly. Note that I plan to add a feature in an upcoming release, where you can choose the 'source' of the keyword tool as the Cluster, N-Gram, Collocate, or Word tool. This would then simplify the procedure as you wouldn't need to export the results for the target corpus as AntConc would be able to use the results directly. It would work in a similar way to how the Word Cloud tool works, where you can set the source for the lists as being any one of the tool outputs.

I hope that helps!

Laurence.