Lancaster University Department of Linguistics and Modern English Language
Corpus Linguistics Home
Page index
Intro to BNCweb
More on BNCweb
Refining BNCweb Queries
DIY Corpora
Page One
Page Two
Page Three
Page Four


"Delicious" and "tasty":
how do they behave with other words?


We will now look look at collocation function of BNCweb. For this, we will consider the behaviour of the near-synonymous words, "delicious" and "tasty".

Do you think the word delicious is used in different contexts from the word tasty? Can you think of particular words which co-occur with either of them? What is a main semantic difference between the two?

Your task is to search the two words delicious and tasty, and find collocates to identify semantic differences between the two. In order to do this, we will use the Collocation function of BNCweb.

If you are not already logged into BNCweb, you will need to log in now.

  1. First, search for the word delicious.
    How many times does it occur?
    How many texts does it appear in?
    What is its normalised frequency?
  2. Sort the concordance lines by R1 to see what words follow delicious. Then look at the left context by sorting the lines by L1.
  3. Now select Collocations and press [Go!]
  4. Press [Submit]. It will produce a collocation database.
  5. You will get a screen looking something like this screenshot. Try out the functions as described below.

BNCweb's collocation functions

Mutual information tends to pick up characteristic collocations regardless of the absolute frequencies of the collocates.

Log-likelihood looks quite similar to Rank by Frequencies, but the scores are more reliable in that they take into account the absolute frequency of each collocate.

Log-log is somewhat in the middle of MI and Log-likelihood.

Observed/Expected is another measure which is used as the basis of Chi-square and Log-likelihood. It also take into account the absolute frequency of each collocate.

  1. To make things easy, letís just look at what nouns the word delicious modifies. Change the settings as follows and then obtain the collocation lists:
    1. Limit the Window size to "+1" and "+1". This will pick up the collocates which appear in the R1 (first word after delicious) only.
    2. Limit the Tag to any noun.
    3. Choose Log-likelihood for statistics.
  2. Follow steps 1 to 6 for the word tasty and analyse your results. Can you find any difference in the way these two words are used in context?

Look at the list of words below. Which collocate with delicious? Which collocate with tasty? Do any collocate with both words?

meal food(s) smell
recipes dishes seafood
monster* sensation lunch
cakes morsel(s) treat
titbits snack(s) ideas
bread treble prizes

*"Delicious Monster" seems to be the name of a band (?).

What observation can you make about the two sets of collocates?


Extra Exercises
  • What are the collocates of the words oppressive and repressive? From looking at the collocates, do the words have similar or different uses?
  • Examine the collocates of distance. Then carry out searches of distance as a noun (distance=NN1) and distance as a verb (distance=VVI). How do the collocates differ in each case?