Nov 29, 2014

Learners' use of collocations: insights from the research 2

"Perform surgery" or "carry out surgery"?
Photo by Austin Samaritans via Flickr [CC BY-SA 2.O]
What kind of collocations are most mistake-prone:
strong (e.g. honk the horn, shrug shoulders), medium-strong (e.g. wage a war, fail a test), medium-weak (e.g. perform an experiment, reach a compromise) or weak (e.g. see a film, read the newspaper)?

In the previous post in the series (see HERE), I looked at some studies focusing on collocational errors across different levels of proficiency and what causes them. One of the studies I cited was by Nadja Nesselhauf (I think it was her doctoral thesis). The study is also interesting in that she tried to pinpoint which collocations are more problematic for learners.

But first of all,

What counts as a collocation? 

Learners often have problems with collocations, but what kind of collocations cause more difficulty? On the one hand, they are idioms, such as foam at the mouth, which are technically collocations (verb + noun), and on the other hand, "looser" combinations of words such as pretty girl or watch TV, although seemingly obvious, are also considered collocations. Enclosed between these two extremes lies a vast array of collocations of varying fixedness.

Howarth (1998) proposed a continuum ranging from free collocations (e.g. pay the rent) to somewhat restricted collocations   and figurative idioms (e.g. sweeten the pill). Or using Howards’s example with the verb blow, the continuum looks as follows:

free collocations
figurative idioms
pure idioms
blow a trumpet
blow a fuse
blow your own trumpet
blow the gaff

Note: figurative idioms (e.g. take steps) are different from pure idioms in that their meanings are much easier to understand, i.e. meaning can be understood by looking at the components.

Howarth points out that the most problematic area for learners is the muddy middle ground. Indeed, his study which compared the production of collocations in native and advanced non-native academic writing, concluded that non-natives tend to make more mistakes with restricted collocations.

Let's now turn to Nesselhauf's study of collocation use in advanced students' writing (L1 German). She ordered the collocations on a scale similar to that of Howarth's.  However, the collocations that were found in the middle of the continuum were further divided into two groups. After Howarth, she uses the word "restricted", but for ease of explanation I’ll refer to collocations as weak or strong.

(a lot of restriction)
(little restriction)
(free collocations)
shrug your shoulders
fail a test
conduct research
see a film

Medium-weak (or little restriction) implies that, a verb collocates with many nouns, yet there are some limitations on its combinability. For example, exert can collocate with influence / pressure / control / power but not with the word rights or, similarly, perform an experiment / analysis / ceremony / task are all possible while *perform an interview is not.

Medium-strong collocations, on the other hand, are comprised of verbs that attract only a handful of possible nouns, such as fail an exam / test, i.e. their collocational range is narrower. Which brings us to the main point of this post:

Which type is most problematic for the learner?

Interestingly, the highest number of errors occurs in the collocations with less restriction i.e. combinations which lean towards the weaker end of the spectrum (medium-weak). At the same time, the lowest number of mistakes occurred in medium-stronger collocations, such as fail a test and pay attention.

How does the researcher explain this finding? She points out that students may not be aware of collocational restrictions. For example, they assume that reach, which collocates with agreement, compromise, conclusion, decision, goal can also go with aim (cf. achieve an aim or achieve a goal), which seems semantically possible. In other words, they may perceive certain combinations as possible although collocationally they are not.

It may also be the case that stronger collocations (e.g. pay attention) are more salient and easier to remember while weaker ones allow more variation and do not seem to show a logical pattern.

Her suggestion for language practitioners therefore is to adopt different strategies when presenting these two types of collocations. While dealing with stronger/more restricted collocations she suggests that it is enough to point out one or two collocates that the verb can take and stress that these are the only possible combinations. For example, run a risk but not *run the danger (of).

Conversely, when dealing with weaker/less restricted collocations, teachers should point out that they cannot be used freely and highlight semantically possible but collocationally impossible combinations using contrastive analysis. For example, when teaching German speakers (although the same suggestion may apply to any L1) it will be worth pointing out that reach can collocate with agreement, decision, conclusion, compromise or goal, but not *aim and contrasting it with the similar verb achieve.


Although the research mainly concerned verb+noun collocations, it would seem that stronger collocations with other parts of speech, e.g. noun + noun (sibling rivalry) or adjective + noun (terminal illness) pose less of a problem for the teacher or the learner because their idiomaticity or fixedness is more noticeable, that is provided learners have been trained to “notice". However, collocations which lean towards the weaker end of the strong-weak spectrum cause more difficulty because learners often map new English words onto their L1 equivalents without realising that they may differ from L1 in their collocational behaviour.

Some practical suggestions

Based on this and my previous Insights from the research post, here are some practical classroom suggestions.

When teaching collocations:

1. Provide negative evidence. 

For example, when teaching (or reviewing) earn give a few common collocates, e.g. money, a living, respect but not *earn experience:

earn money / a living / respect
earn experience -> gain experience

You can do this using collocation forks - see HERE.

The same things can be done with adjectives, for example:
mild weather / winter / flavour / surprise
mild light -> soft light

2. Use Contrastive analysis. 

If you know your students' L1 you can often anticipate their errors if you know that similar meaning is expressed in their L1 using a different collocation, i.e very often an equivalent noun will take a different verb. Contrast new items with students L1 to highlight possible issues. For example, ask a question, not *make a question for Spanish speakers (cf. hacer una pregunta) or make a speech and not *pronounce a speech for French speakers (cf. prononcer un discours)

Some English verbs that can be problematic are:
achieve, accomplish, comply (with), conduct, cope (with), overcome, perform, submit

3. Draw attention to form

When reading or listening, students often do not pay attention to form. This is especially the case with semantically transparent collocations, such as run a risk or hold a conference, whose meaning can be easily decoded when encountered in the input. When it comes to production, however, learners will often turn to L1 and combine words based on their L1 intuitions. Draw your students' attention to new combinations of already known words as well as seemingly "easy" or obvious collocations such as have an accident or give an example.

4. Explicitly teach collocations 

(as opposed to single words). Many words do not have one-to-one correspondence with L1. Even words that seem to have a direct translation equivalent will often behave differently in English, namely, have different collocations. Very often, these are so called cognates, i.e. words in different languages with the same etymology, such as execute = exécuter (Fr.) =  ejecutar (Sp.) It is very important to teach these with their lexical partners.

I know some of these echo my plea from one of my previous posts entitled Start teaching lexically but if you haven't read it I suggest checking it out. Click HERE


Howarth, P. (1998). The phraseology of learners’ academic writing. In A.P. Cowie (Ed.), Phraseology: Theory, analysis, and applications (pp 161–186). Oxford: Oxford University Press.

Nesselhauf, N. (2003). The use of collocations by advanced learners of English and some implications for teaching. Applied Linguistics, 24(2), 223–242.


  1. Thanks Leo, very thought-provoking. And it has got me thinking much more about what is grammar and what is lexis - something I hadn't considered as deeply before.

    1. Thank you for visiting my blog, Marjorie. And for the comment!

  2. Nice post, maybe I misunderstood, but are you really saying that "run the danger of" is ungrammatical? Have youchecked a corpus?

    1. Hi
      Well... that's the thing. "Run the danger" is what Nesselhauf gives as an example of an impossible collocation. She admits in the Notes though that:

      "Although run the danger is sometimes used, it is considerably less common than run the risk."

      Also, all four native speakers who were asked to "mark" her students' writing found it wrong or unacceptable. She also refers to BNC where there are 30 times more occurrences of run (the) risk than run the danger and when I consulted COCA I got the same picture.

    2. Thanks for the clarification. I agree about the respective frequencies, but I wouldn't characterize something used by John Locke as ungrammatical.

    3. anon you bring up a good point about considering frequency as well, Martinez has a nice conceptualisation alled the Freuquency Transparency Framework, I talk about that in this post


  3. Anon, the point is that some word partnerships are more likely and some are less likely (not the same as ungrammatical or right and wrong). Incidentally a lot of word partnerships used by John Locke are likely to be unlikely in modern English.

    1. Thank you for your comment, Paul.
      "Likely to be unlikely" ? :)

    2. Where I come from, if you mark an expression with *, you're saying that it's ungrammatical. I honestly don't know how someone can characterize something as "an impossible collocation" when it appears in published writing by native speakers. I understand the point about frequency, but the original point didn't say one expression was more frequent than another: it said it was ungrammatical. I don't mean to badger the point.

    3. Anon, I don't know the exact context Nesselhauf's subjects used run the danger of in. But the fact that it was corrected by more than one native speaker shows that it wasn't used appropriately. You can see the whole paper here: with many other examples of mis-collocation, such as *close lacks or *make homework.

      I used * (like Nesselhauf does) to indicate that a construction is not appropriate or infelicitous - in some way deviant, if you like. I don't know if the word "ungrammatical" applies her - we are talking about vocabulary and not grammar after all.

      Now, you will see that many examples she gives do not seem deviant (any more?), for example "Reach an aim". I would personally say and teach "achieve/reach a goal" but a simple google search yields a lot of instances of "reach an aim". I even found a couple of examples in COCA, although they are significantly outnumbered by "achieve/reach a goal". I think it all comes down to teaching your students the most frequent combinations which would be useful in a variety of contexts.

  4. Thanks Leo.
    Any ideas for a collocation-generating app?

    1. Hi Peter,
      Yes that would be a great idea.

  5. This is so impressive and nice title of blog. I am always searching such type of Blog that gave some useful knowledge and relax our mind. So Very thanks for design this type of content and what are you thing at time of write. Thanks for sharing views.

    old delhi tours & Guided Tours in Delhi

  6. Hi Leo!! I found your suggestions extremely helpful and useful. I have been trying to provide my students with as much lexical patters as possible to help them become better users of the language, but I have never thought of the "negative evidence". Thank you for the tips and for sharing your insights and research.

    1. Glad you found it helpful. Negative evidence is not my idea. I think I came across the idea in one of Michael Lewis's book. Possibly The Lexical Approach (1993). I think it's very important - it doesn't always stop students from mis-collocating but at least it raises their awareness, for example:

      take / pass / fail AN EXAM but not *make an exam

      (Sorry, I can't show it with strikethrough font because it's not available here in the comments)

      Thank you for visiting my blog.


Related Posts Plugin for WordPress, Blogger...