A Ten-step Introduction to Concordancing through the Collins Cobuild Corpus Concordance Sampler

1. Introduction

If you are a student of English, a teacher, a translator or if you are writing in English, analysing English, or have any questions about how English works, concordancing can be of great benefit to you.

This mini-course is designed to introduce concordancing to students and teachers, native and non-native speakers of English and people with a general interest in language and learner autonomy. It is therefore casting its net over a wide range of people. It does not assume a great deal of linguistic knowledge: all required terminology, both in computer use and linguistics, is explained as it is introduced. The main purpose of these sessions is to introduce the techniques involved in searching for answers to questions. Just what can be asked will be revealed at every step, as we see how searches can be formed and refined.

This first Session is a short introduction. In Session 2 we will dive directly into concordancing using the Collins Cobuild Corpus Concordance Sampler. This has been chosen because it is readily accessible through the internet and because of its rich variety of functions that demonstrate many features of full concordancers. Throughout these sessions it will be shortened to CCS.

If you are a non-native speaker of English, it is likely that you will want your English to be as close as possible to the norms of English. You might even think that English is spoken and written with so much variation that the norms are too unstable to be grasped. There is, of course, a core language which represents the vast majority of English, without which it could not be called one language. It is not always easy to gain access to those norms, i.e. the most likely way a native speaker would express something, and grammar books are not always able to answer questions, especially about word use: modern dictionaries, based on corpus research, provide more reliable information. Corpus-based grammar books now exist also, but so far, they are of more use to the language professional than the general user. One of these is the Longman Grammar of Written and Spoken English. Read a review of it in TESL-EJ.

What is Cobuild?

It is an acronym: Collins Birmingham University International Language Database. A great deal of pioneering work in corpus linguistics has been done at Birmingham University. Chris Allen gives an overview of the project from an insider’s point of view:

What is a corpus?

A collection or body of texts in electronic form. The plural is corpora. The Cobuild corpus is referred to as The Bank of English. This link outlines their project and concludes with a link to the Sampler, which is what the current site is about.

What is a concordancer?

Software for looking into a corpus.

What is a concordance?

The lines of text illustrating the search word, the node.

A caveat for a corpus

A corpus is assembled by collecting texts in electronic form. The texts are usually chosen to represent such things as:

genre	contract, letter of appointment, theatre program
domain	the family, at work
register	conversation, fiction, newspaper language, academic prose
mode	writing, speaking, gesturing

Textové pole: Jan Svartvik wrote: Every corpus that I've had a chance to examine, however small, has taught me facts that I couldn't imagine finding out about in any other way. Importantly, texts are not corrected according to any grammar or spelling rules, taboo words are not “cleaned up”, and general abuse of the language sits happily alongside general use of the language. Slips of the tongue, pen and keyboard remain intact. It is therefore a descriptive sample of the language, not a prescriptive one: this makes it rich, but it also means that you should usually look for significant patterns, not oddities. More on this later.

A caveat for a sampler

In this Sampler, a database of millions of words is searched and up to forty concordance lines are shown. In a full concordancer, for example Microconcord (by Tim Johns and Mike Scott) which has its own corpus of only 2 million words, and the British National Corpus of 100 million words, you find:

	Microconcord	BNC
hand	459	33484
grant	81	7594
unemployment	120	6409
university	225	16316

Somewhat more than forty! Further machine intervention is required when you have large numbers of finds. From an introductory point of view, the forty lines limitation is a manageable number to deal with. And there are techniques for refining your search to get forty sharply focused lines, as we shall soon see.

When forty lines are shown, however, it is not clear how many they were selected from. For example, if you are comparing the use of “at least” and “at the least”, knowing their relative frequencies would be a useful starting point since frequency is an indicator of typicality.

	CCS	Microconcord
at least	40	662
at the least	34	1

Ultimately, a sampler can answer the question: is there any evidence for…?, rather than a more decisively-framed question.

Non-native students learning English

Data Driven Learning refers to studying English by isolating patterns that occur in real language. The student answers his or her language questions by analysing the data the concordance produces. The remainder of these sessions shows just how that is done. These procedures were pioneered largely by Tim Johns who named this pedagogical application Data Driven Learning. Links to some others involved in this form of language study appear throughout these sessions.

One last caveat

Since you have got this far, let me close the introductory session with another caveat, this time for this website. You are reading its first version before it has even been tried out on anyone. Therefore, all comments and suggestions will be gratefully received, read and taken on board. My contact: thomas@fi.muni.cz.

Go to top

2. Getting started: find a word

Without further ado, let’s jump in head first and see. We will now perform some searches and see what a concordance is, and what concordancers look like.

Note: if you click on this link, Cobuild, you will open the CCS in this window, leaving this one.

There are two alternatives:

(a) using the right mouse button, choose to open the link in a new window, or

(b) Ctrl N will open another window and you can click on the link.

At the bottom of the screen you have the navigation buttons that take you from one window to another.

The Cobuild link is to this address: http://titania.cobuild.collins.co.uk/form.html

Search One

Is whose used with animate and inanimate antecedents.

Type the word whose into the box and click Show concs. The search word is referred to as the node. And the format in which the search results are displayed is referred to as KWIC, which stands for Key Word In Context.

So, do we say, I saw a car whose owner … ?

Search Two

Type the word step into the box and click Show concs.

What do you notice about the order of the first words to the right of the node?

Correct! They are in alphabetical order, with numbers, punctuation and codes (e.g. [f]) at the top. In this concordancer sampler, this is the only possible order to view the concordances in. In other programs, sorting by the first word to the left is one of a number of possible sort options. Each sort reveals different information.

Observation: does the node appear with its declensions or conjugations? See Morphology below.

What do the forty lines of step tell you about the word? What properties of this word can you observe?

· a verb (finite or infinite or both?)

· in a phrasal verb construction

· in a delexical verb construction

· as a noun – with different meanings (polysemy)

· as part of a compound noun

· common words following it

· commons words preceding it

· in fixed phrases

· in a metaphorical sense

All of this can be demonstrated with this simplest of searches with only forty lines. By this stage you might be wondering what a word is. Click this link for an introduction to the quandary.

Search Three

We might expect that X proves Y in a legal or scientific sense. What evidence do you find for that assumption? Type in prove or proves and click Show concs. Read the contexts observing the sorts of things than can be X (the subject) and Y (the object).

Search Four

Can you pretend something, i.e. a noun or a noun phrase? Can pretend be followed by “-ing” forms, infinitives with or without “to”, “wh-” clauses, “that” clauses, or anything else?

To find out which of these complementation patterns pretend has, type pretend into the box and click Show concs.

What do you notice about this verb when displayed in the KWIC format?

And what mostly precedes it? Modals, auxiliaries and “to”. This is because the word form pretend is uninflected. The full lemma is pretend, pretends, pretended, pretending.

Consider why this information is useful?

Search Five

Question: what parts of speech (POS) can fast function as?

Type the word fast into the box.

Can you say which lines exemplify which POS and how many of each?

FAST	line numbers	totals
Noun
Verb
Adjective
Adverb

This is indicative of the random sampling.

Search Six

Question: Many present participles (-ing forms) are used adjectivally. Is this true of pretending? If so, attributively and predicatively? These are important questions when studying vocabulary.

So far, we have been generating concordances for base (uninflected) forms. But if you want to search for pretending only, type in precisely that. It appears in verb groups (e.g. present continuous) and as an adjective. In Session 4, we will see how to focus such a search.

Search Seven

Comparing how we use singular and plural forms of nouns can be instructive. Try a search on hand and another on hands. What do you notice about the different usage?

Search Eight

If you haven’t already encountered problems using capital letters, try searching for the BBC. Is it possible that there is no mention of the BBC in a corpus of 56 million words? No. It only works if you use lower case letters, i.e. bbc. Similarly, People, is not a valid search, whereas people is, and the results include People.

Search Nine

Try typing in a number and searching. No luck? You have to use a backslash before the number, e.g. \1970 or \12.

Closing comment: it cannot be denied that the language which appears when corpora are searched is noticeably richer than composed, edited or carefully selected examples, for inclusion in textbooks and dictionaries. This richness is an important factor in learner input.

Go to top

3. Searching for lemmas and word families

The traditional hierarchy of morphology

Lemma

A lemma is a group of words created through inflectional processes – the yellow half.

How many word forms does a regular verb consist of?

worry, worries, worrying, worried.

How many word forms does an irregular verb consist of?

cut cuts cutting

bring brings bringing brought

sink sinks sinking sank sunk

How many word forms does a regular adjective consist of?

steep steeper steepest

Fortunately, you don’t have to enter the full lemma to find it. Simply type @ after your word.

Try searches on precede@, contribute@, fast@, knife@, corpus@, simple@

Word Family

A word family is a group of words created through word formation processes– the blue half above. It is less precisely determined than the lemma. Use the wildcard symbol (the asterisk) to find any string of letters that follows what you enter. [JET1]

preten*
prohib*
argu*

Since the wildcard finds anything starting with your entry, a search such as mid* or mon* will return quite various results. Try it.

For observing prefixes at work, you could search mega*, under*, mis*, etc. Unfortunately, it is not possible in the CCS to search with wildcard first, e.g. *ness, *wise, *fold, *ish.

Go to top

4. Parts of Speech searches

Every word in the CCS has its part of speech (POS) marked, or tagged. You can imagine that tagging every word in a multi-million word corpus is a daunting job. In fact, it is done by computers with an estimated accuracy rate of 95%. Being able to search for a word by POS is often essential. For example, the word form ROSE can be a flower or the past tense of rise. Mixed search results will not be helpful.

Some POS issues have been deliberately avoided in CCS, such as the use of some participles as adjectives: winning is not marked as an adjective in “winning smile” and failed is not marked as an adjective in “a failed bank”, but homing is in “homing device”. Also above and fast cannot be located using the adverb tag. Nevertheless, it still provides a view on language that I couldn't imagine finding out about in any other way (apologies to Jan Svartvik).

Textové pole: TIP: remember that using Ctrl N, you can open several windows and perform several searches at the same time. As we noted in Session 2, fast can be a noun, verb, adjective and adverb. Being able to specify a search word’s POS allows you to find a word in more specific contexts and in a more specific sense. If you were wondering whether to use the adjective fast or quick in a particular situation, you would obtain more useful data by specifically searching for the words as adjectives. Search for fast/JJ and then search for quick/JJ.

The query syntax is: the word, a slash and the tag in CAPITAL LETTERS.

Here is list CCS POS tags. Other corpora have different tags, and other concordancing programs have different ways of forming a query.

Table of POS tags

This is an expansion of the information provided on the Cobuild site below the concordance entry box.

NOUN	a macro tag: stands for any noun tag	walk/NOUN
VERB	a macro tag: stands for any verb tag	dog@/VERB
NN	common noun	peer/NN
NNS	noun plural	needs/NNS will not show the word as a 3^rd person singular verb.
JJ	adjective	sound/JJ not as a verb or noun.
DT	definite and indefinite article	This is used in word strings, as we shall see in Session 6. It gives a, an and the.
IN	preposition	This is used in word strings, when you want a word plus preposition.
RB	adverb	Is there an adverb derived from prohibit? *prohibit/RB Or from ration/RB*?
VB	base-form verb	trigger/VB or impact/VB
VBN	past participle verb	read/VBN – useful if studying passive or perfect aspect. And you can separate out adjectival uses.
VBG	-ing form verb	read/VBG – useful if studying continuous aspect. And you can separate out adjectival uses.
VBD	past tense verb	set can be present and past. set/VBD only shows concordances where it is a past tense verb.
CC	coordinating conjunction	e.g. and, but
CS	subordinating conjunction	e.g. while, because
PPS	personal pronoun subject case	e.g. she, I
PPO	personal pronoun object case	e.g. her, me
PPP	possessive pronoun	e.g. hers, mine
DTG	determiner-pronoun	e.g. many, all, both, some

Note: these POS tags become much more powerful when used in combination as we see in Session 5.

Refining your searches

You can now refine the searches you tried in the previous session.

Lemma:

Try a search on peer, which is a proper noun, and has two meanings as a common noun. It is also a verb. Try peer@. Try peer@/NOUN and peer@/VERB

Word Family:

Try preten*/JJ. and see adjectives starting with “preten”.

Try prohib*/RB and you will see the derived adverb.

Try contra*/NNS and you will see quite a few nouns in the plural that start this way.

What nouns are in the contract family? Try contract*/NOUN

What adjectives derive from oil? Search oil*/JJ.

What adjectives derive from club? club*/JJ

Go to top

5. More than one word

If seeing specific word groups in context would answer your question, type in each word with a + between each word, and no spaces. For example knife+and+fork.

L e x e m e	phrasal verb	take off, step down, enter into
	modals	be about to, had better, be bound to
	compound noun	coffee table, step son, word family
	compound preposition	away from, regardless of, in comparison with
	fixed phrase	in the light of, open to suggestion, up up and away
	collocation	vivid imagination, irresistible temptation, little imp
	idiom	storm in a teacup, bull in a china shop
	quotation	much ado about nothing, couldn’t give a damn
	discourse markers	be that as it may, comparatively speaking, in other words

POS tags can also be part of a search. Try impact+IN to find what prepositions follow impact. More specifically, you could try impact/VERB+IN.

Discontinuous groups

Many phrasal verbs are used discontinuously, i.e. other words appear between the verb and the particle. Add the maximum number of intervening words you want in your search. For example take+3for, or take@+3for.

Delexical verbs are almost always discontinuous. Search give+3smile, make+3speech, take+3photograph, have+3bath.

Does whether or not appear as a fixed unit, or can it be separated? Search whether+or+not and whether+5or+not.

Honing in

It will also happen that after searching a single word, you will want to find more examples of one of the groups in which it occurs. When you search for teach, the second last line contains teach you how to.

Search for the whole chunk.

teach+1how	many lines of teaching someone how to do something
teach+PPO+how	gives examples where the object is a pronoun

Practice: search the left word and then one of the chunks in which it turns up.

mouse	è	cat and mouse
table	è	table talk
pause	è	pregnant pause.

Alternative search items

If you want to search for two items at the same time, use the vertical bar | (Shift + the key beside the arrow at the top right of the English keyboard).

Ø criticize|criticise

Ø dove|dived

Ø though|although

Ø precision|preciseness

Ø open+2up|out

Training in combining queries

Do we say on a/the train or in a/the train? Since a/the is not at issue, search in+1train and then on+1train. Unsurprisingly, we say both, and equally unsurprisingly, they mean different things. You could limit your search to finding only articles between the preposition and the noun. on+DT+train. Or if you want them on the same screen in|on+DT+train. With a maximum number of forty finds, this can sometimes present too few results.

Is train also a verb? train@/VERB Does it ever have the railway sense when used as a verb? What is the verb that occurs frequently with in train? What compound nouns does train occur in?

Alternatives to POS tags

We noted above that participles were not always marked as adjectives. We know that adjectives typically precede nouns so searching for welcoming+NOUN is likely to find welcoming as an adjective, but it also finds it in the continuous sense e.g are welcoming. Since noun phrases are often launched by determiners, DT+welcoming+NOUN is more likely to prevent any continuous uses appearing.

To search for any –ing form before a specific noun, e.g. ceremony, you would search DT+VBG+ceremony. Similarly, DT+VBN+house will find past participles as adjectives. This is a useful strategy for finding specific types of collocations.

Apostrophes

The plus sign is also used when searching for words with apostrophes, e.g. can+t, michael+s. The query to find examples of "Bob's your uncle" is bob+s+your+uncle

Go to top

6. Searching for patterns

As we have seen, POS queries are not limited to specifying a word’s POS. This allows you to search for a word or lemma in conjunction with a POS. The following sections are designed to illustrate how you can obtain results that illustrate a word’s patterns using different queries

Take AIM as an example.

Follow these searches through and check your finds against these observations.

aim+NOUN	aim as common noun, proper noun and verb
aim/VERB	as a verb, aim most frequently appears with prepositions.
aim/VERB+NOUN	the base form aim is not frequently used
aim@/VERB+NOUN	this search shows more examples of aiming at something
aim@/VERB+DT+NOUN	since noun phrases typically start with a determiner, this search yields the most results of aim’s objects
aim@/VERB+1NOUN	this shows the random selection of things that can appear between the lemma aim and a noun.
aim@/VERB+1DT+NOUN	a combination of the two searches above.
aim@/VERB+IN+DT+NOUN	this shows prepositional phrases that follow aim
	we saw in the above step that at and for were the most frequent. Towards, in and of occurred only once each. Are they so insignificant, or is that a result of the random selection or tagging errors?
aim@/VERB+of aim@/NOUN+of	A tagging error. When aim is followed by of, it is a noun
aim@/NOUN+of	What are the structures here? 1. One structure is of with an –ing form. 2. The aim of X + to be (without to) + infinitive (with to).
aim@/NOUN+of+VBG	For more examples of 1.
aim@/NOUN+of+2NOUN+be@+to.	For more examples of 2.
NOUN+of+2NOUN+be@+to	Try this search and see if this structure is unique to aim.
aim@/VERB+to	When do we say aim to? When to is part of the infinitive. So now we have observed that aim is also followed by the infinitive with to.
aim@/VERB+to+DT+NOUN	Does to launch prepositional phrases?
aim@/VERB+in	In the examples found here, in is not bound to aim, rather it launches a prepositional phrase.
aim@/VERB+towards	Four finds is not many. Is there another way of expressing that concept?
aim@/VERB+for	Do aim for and aim towards express the same thing? Have a look at the for list and consider where towards could be substituted without changing the basic meaning. Unlikely!
take@+aim	this delexical form seems to be restricted to the target sense of aim.

Grammar patterns

In English, perfect and continuous aspects, the passive, causative and conditionals are formed by auxiliaries in contrast to many languages where they are formed with suffixes. Auxiliaries are words, and this is what concordancers work with best.

What structures will the following queries exemplify?

have\|has+been+VBG
had+been+VBG
have@+be@+VBD
have@+PPO+VBD
get@+VBN
have@+PPS+ever+VBN
if+1had+VBD+6would
if+PPS+4will

Active voice: how active?

When is get used to form the passive? get@+VBN or get@+1VBN will provide some examples that could support your hypothetical answer to this question. It is necessary to separate the constructions in which get is a full verb meaning obtain or become etc from those where it is an auxiliary. How? Human intervention, i.e. do it yourself.

See The Get Passive for an English lesson on this issue.

Curious and Curiouser

Did you know that stative verbs cannot be used in continuous structures? Go to this grammar link for a statement of the rule. For a longer list of examples, have a look at netgrammar. What if you came across one of these verbs used continuously? Would you doubt the rule? How can we find evidence to support or refute it?

Try be+VBG or be@+VBG.

Search for some of the examples given at those sites. e.g. be@+hating, or since continuous forms are often discontinuous (sic) be@+2owning.

Colligation

The above section referred to grammar patterns in the abstract, as the foundation of clauses. The more familiar concept of collocation, as we shall see in detail below (Session 8), refers to frequent co-occurrences of words, e.g. logical conclusion, end result, to answer a call. Colligation, however, refers to a word’s syntactic patterns, which is an important part of knowing a word. In Barnbrook’s words, colligation refers to collocation patterns that are based on syntactic groups rather than individual words (1996).

Very interestingly, corpus analysis has shown that words with the same complementation can be grouped into semantic classes. For example, when bleed, care, cry, fear, feel, grieve, mourn and weep (Levin 1993:192) are followed by for someone, they express sympathy. When these words have different complementations, they do not form this group.

For a solid introduction to this notion, go to the Forum section (p.3) of this link by Susan Hunston.

Which verbs can you find that have the following structures? Can you see any semantic similarities within each pattern? Try the following:

VERB+NOUN+as+JJ
VERB+on+to+NOUN
it+VERB+to+VB
VERB+from+NOUN+to+NOUN
draw+2NOUN+from

Articles – using DT in queries

Is it true that polar adjectives tend to be preceded by the definite article?

Search DT+first|next|last and compare your findings with a+first|next|last.

Right and Wrong

*This is the best way how to learn English. (The asterisk is a convention used in grammar books to indicate that a sentence is unacceptable). Unacceptable? Try this procedure:

way (or way@)	gives an overview of what typically follows the word. There are two ways of complementing way with a verb: of + ing form and infinitive with to.
way+how (or way@+how)	of all the concordances, only one has this use of way how.
way+of+VBG	convincing results
way+to+VB	convincing results

Decisions decisions: when to use the infinitive and when to use the gerund?

Correct?

Which is correct: it is me or it is I? Search be@+me and be@+i. Don’t forget that we do not use capital letters in queries except as POS tags. Alternatively you could search: it+be@+PPS and it+be@PPO.

What does “correct” mean? What does hypercorrect mean?

Go to top

7. Language varieties

Whole texts are tagged according to variety and other specifications. Corpora designed for research into stylistics or pragmatics, for example, are likely to be tagged in great detail and include the age, gender and nationality of the speakers, the date of publication, etc.

In the CCS, you have three choices, namely:

q British books, ephemera, radio, newspapers, magazines (26m words)

q American books, ephemera and radio (9m words)

q British transcribed speech (10m words)

If restricting the text type according to these criteria could be helpful, select the appropriate check boxes before hitting Show Concs.

Spoken and Written Language

You might like to read this introductory article by Mario Rinvolucri, Distinctions and Dichotomies on spoken and written language.

Ø Are moreover and whereas used in speech, or do they belong to the written language?

Ø would have thought – is this lexical bundle used in written English?

Ø You can find examples of question tags, can’t you?

Ø Are goodness me, for+all+i+care and for+heaven+s+sake actually used?

Ø This spoken search may surprise: like+VBD

UK and US English

Ø Who says: different from and different than.

Ø Dived or dove? Also, incidentally, Dove/VERB vs dove/NOUN

Ø Some say have, others take a bath or shower. Try this: have|take+DT+bath|shower

How are these words used differently on opposite sides of the Atlantic?

Ø momentarily

Ø smart

Ø fancy

Ø football

In which language variety does lanai appears as a common noun? And Hoosier? This link will take you a discussion of the word – where you will also see “unhandy” in its definition.

For more on these varieties, try "Or whose language is it anyway?" and

Potentially Confusing And Embarrassing Differences between American and British English.

Exploiting the varieties function

Do we write the 1970s with or without an apostrophe? Remember to use the backslash before numbers. Perform these two searches and note how many concordances there are of each: \1970s and \1970+s. If there are less than forty, we can assume that there are no more in the whole corpus, and that that number out of 56 million words is not very significant. Search \1970+s as US and again as UK. Does it seem that these are all the examples in the whole corpus? Search for \1970+s as transcribed speech only. What do transcribers know about writing decades with apostrophes?

Go to top

8. Collocation

The tendency for lexical words to occur together is called collocation, e.g. a vivid imagination, perform an analysis, deliver a speech, a problem child, rotten lover. But not all co-occurring items can be counted as collocations. The following are not collocations:

Ø Multi-word lexemes are not collocations, e.g. phrasal verbs, compound nouns

Ø Colligation e.g. rely on, wait for, crowd of, can’t help + -ing,

Ø Lexical bundles e.g. I don’t know, at the time of writing, it is interesting to note that, to be taken into a account

Ø Fixed phrases may be considered an extended collocation: e.g. rather you than me, if you’ve got the energy, not on your life, all’s well that ends well, under the weather, the nine o’clock news, not for love (n)or money, as far as I’m concerned,

Collocation is a major issue in current linguistic thinking and its applications to language learning and translation, in particular. This is partly because a sound knowledge of collocations brings language production closer to native speaker norms. Firth said in 1957 that you know a word by the company it keeps. For Cobuild’s purposes here, and on their Collocations CD, the company a word keeps is specified within four words to the left or right of the keyword (or node). In other concordancing programmes you can control this range.

First collocations search

Enter government into the Collocation Sampler. Click Show Collocates. It looks through the corpus and produces a list of the 100 most frequent words, and you will notice that many of the words are “government-type” words.

T-score and MI score

Enter experiment into the Collocation Sampler and choose T-score. Click Show Collocates. Open a new window and repeat this search choosing Mutual Information. You will notice that the results are quite different. T-score or MI score which are statistical statements of probability and significance of co-occurrence.

Here is a brief summary/extract of the information Cobuild provides at this site:

Raw freq often picks out the obvious collocates ("post office" "side effect") but you have no way of distinguishing these objectively from frequent non-collocations (like "the effect" "an effect" "effect is" "effect it" etc).

MI (Mutual Information) will highlight the technical terms, oddities, weirdos, totally fixed phrases, etc ("post mortem" "Laurens van der Post" "post-menopausal" "prepaid post"/"post prepaid" "post-grad").

T-score will get you significant collocates which have occurred frequently ("post office" "Washington Post" "post-war", "by post" "the post").

Note: If a collocate appears in the top of both MI and T-score lists it is clearly a humdinger of a collocate, rock-solid, typical, frequent, strongly associated with its node word, recurrent, reliable, etc etc etc.

For the full information that Cobuild supplies, click on the underlined column headings at the top of the collocations list at their site.

Further refining your collocations searches

You can refine your collocations in similar ways to your concordance queries. For example, all of the following will yield different results, and you need to consider what you are looking for. Try the following searches as collocations searches.

grant	You see that both refuse and refusal appear quite high in the list, which might also be considered from a real-world point-of-view.
grant@	the lemmas of the noun and the verb
grant/NOUN
grant@/NN	Since General Grant and Hugh Grant (hard to find links that don’t have pop-ups) occur frequently in the literature, there is some advantage to restricting your query to the common noun.

The statistics can be very revealing, but the collocation list produced is often best considered a starting point. Select what is of interest there and return to the concordancer. For example, how do refuse and grant collocate? Try the following as concordance searches:

refus*+4grant	without searching for grant as a noun, you get refuse to grant someone something.
refus*+4grant/NN	the wildcard (*) allows the inclusion of the word family and the lemma.
grant@/NN+4refus*	CCS’s list does not distinguish left and right collocations, so both need to be checked.

Collocations and synonyms

What are the differences between great, grand, big and large? This is a “great, grand, big and large” question in which collocations play some role. Part of the answer to “what’s the difference between” can often be found in the word’s collocations.

Consider which of the following four adjectives collocate with the nouns in the left column. You might find yourself indicating some of them as strong (S) and weak (W) collocations rather than an absolute Yes or No. The nouns were chosen from the T-score collocations list for each of these adjectives, so you can check your responses yourself.

	great	grand	large	big
asset
brother
difficulty
effect
event
extent
fun
house
idea
impact
majority
opera
picture
pleasure
population
problem
quantity
question
scale
success
tour

As mentioned above, collocation plays some role in answering “what’s the difference”, but most of these nouns allow more than one of these adjectives. Sometimes they form a chunk, e.g., great grandfather, grand prix, Grand Canyon, (in) large measure, (to) great effect.

Big brother and big bang are chunks when preceded by “the”

e.g. … 4,000 million years since the Big Bang which created the earth's …,

and a collocation when preceded by “a”.

e.g. The next thing I heard was a big bang.

And there are metaphorical uses too,

e.g. … transformed by deregulation - the Big Bang . This brought in competition …

You might like to look up Big brother and big picture in the same way.

Try the same approach on another set of synonyms, for example, demonstrate, establish, prove, show. There is a most excellent online resource from Vancouver Wordnet that can be consulted in this connection.

Collocations and polysemy

We have already seen some examples of polysemy, that is, words with more than one meaning. Sometimes a word may collocate with one of a word’s meanings and not with another of its meanings. The noun race has two very different meanings. Search for its collocations and note which words collocate with which meaning:

contest
ethnic group

Another example is horse which has quite different collocations depending on its reference. Search of its collocations and note which words collocate with which reference.

agriculture
gambling
sport

These considerations are important when you generate collocation lists.

Collocations and connotation

Yet another factor in “what’s the difference between” lies in connotation, which you can find briefly described at Denotation and Connotation. When would you choose to describe someone as fat, obese, overweight, dumpy or corpulent? They are synonyms in as far as they carry more or less the same information, but they do not express the same attitude. As we saw in Collocations and Synonyms, some pairings are just a fact of life. As we now see, some pairings are a purposeful choice.

Search: What mongers do we have in English? Search monger. Search mongers. Search monger* and see how different the results are. Do mongers have positive, negative or neutral connotations?

Search galore or arch+NOUN for further examples of positive and negative charging.

9. Searching for Answers: you have to have a question

As you have seen, the searches we have been doing throughout these sessions have been motivated by a question. And many of these questions are typical questions that arise when one is producing language (i.e. speaking and writing). You might like to consider how else you can find answers to these questions without accessing a corpus.

Now that you know what the search results look like and have some idea of the information they reveal, you can start to consider how you will formulate your questions.

Interpreting results

As mentioned above, a search that returns fewer than 40 concordances suggests that what you are searching for is not very significant. While it is true that frequency is an indicator of typicality, …

But the purpose of the search has to be borne in mind. Are you looking for typicality or just some examples that it exists in English at all?

From another angle, you must also take into consideration that a single appearance of something among forty concordances could be significant. It is necessary to isolate it and search it directly. For example, VERB+2hypothesis shows

We might also turn this hypothesis on its head and argue …

Turning a hypothesis on its head only occurs once in this search, but turn@+3on+its+head reveals 25 concordances.

False friends

The internet will provide you with many pages of false friends: try an internet search on “false friends” English plus the name of another language into a search engine. Or click on Confusing Words to go to a list of words that are alike in various languages.

Practice makes perfect

So you think you know the parts of the body in English. Well, you probably do. But do you know what amazing things they can do or how they are described and referred to? Try this matching activity and then check your answers using CCS. You might like to add some other verbs beside each noun.

Nouns	Verbs
shoulders	pick
eye	wag
heart	lick
throat	grind
finger	blink
nose	gargle
buttocks	shrug
neck	pound
lips	crane
teeth	clench

Now find out which of the nouns can be used as verbs. For example, can one eye something, head something, mouth something or shoulder something? Search query: thumb@/VERB.

Can you match these? Can the concordancer help?

Adjectives	Nouns
accusing	shoulders
greasy	thumb
limp	lips
curvaceous	neck
hunched	finger
green	nose
stiff	buttocks
parted	wrist
pointy	palm

How are these words used metaphorically?

Nouns	Use
a shoulder of	burden
the eye of	music
to shoulder the	lamb
an ear for	stone
nerves of	a story
a nose for	steel
heart of	the storm

Metaphor is a fundamental aspect of communication. Many high frequency words are used in many different ways, which is partly why they are so frequent, and an aspect of their use that learners need to become familiar with. For more on this, you could read metaphors, an extract from Metaphors We Live By (Lakoff and Johnson, 1980)

Can you find parts of the body words used in idioms or other fixed phrases?

What about this: grin@+3grin@ or the same search on smile or look

Go to top

10. Further study

These sessions have been principally concerned with training you in an approach to accessing the GRAMMAR OF VOCABULARY you need when you need it.

A lot of current work in language and linguistics, including that relevant to learning foreign languages, gives preference to the structural elements of vocabulary over sentence grammar.

As far as concordancing is concerned, there are still many more things you can do with the CCS. And the CCS is not the only concordancer on the net. And the Collins Cobuild Corpus Sampler is not the only concordancer on the net. Firstly, there is the full Collins Wordbanks online, from which the Sampler is derived.

Some others are listed below and there are some available for purchase on CD, though not all of them have corpora included. As you read through the sites linked below, you will see other activities and approaches, and of course, other links!

It is possible to create your own corpus, though this is beyond the realm of this particular website. If you this is of interest to you, read Improvising corpora for ELT: quick-and-dirty ways of developing corpora for language teaching by Christopher Tribble and The Learner as Corpus Designer by Guy Aston. Obtaining copyright permission to use texts is a complex issue and corpora based on out-of-copyright data cannot represent a language in its current state.

The World Wide Web Access to Corpora Project (W3-Corpora) was run at the Department of Language and Linguistics at the University of Essex. Click their tutorial for the W3-Corpora Interface to access their corpus and concordancer. Although it works differently from the CCS, you will experience such practical things as:

· looking for the meaning of a word

· comparing two similar words/synonyms

· comparing how a word is used in different kinds of text

· seeing which preposition to use

· checking the spelling of a word

which you can apply and add to your repertoire of techniques.

Corpora have been constructed in many languages. For a list, look Mike Barlow’s list.

An important practical issue for the language learner: what do you do with the answers to your questions? Tom Cobb has written about this: Giving learners something to do with concordance output.

Some Resources

The home page of Tim Johns of Data Driven Learning fame contains a wealth of original material and links.

Mike Scott co-authored Microconcord (a DOS concordancer with 2 million corpus – still good and fast) with Tim Johns and then the more sophisticated Wordsmith Tools (for Windows).

Tom Cobb’s article, Breadth and depth of lexical acquisition with hands-on concordancing, describes an experiment using concordancing with students who needed to acquire a large vocabulary quickly.

Tom Cobb’s The Compleat Lexical Tutor includes a range of applications of corpus work to vocabulary teaching. An article by him about using concordance software to provide learners with a rich language learning experience can be found at Concordancing in the CEGEPs.

Spaceless: This concordancer takes the text of a web page and creates a list of sentences that contain the search term.

The VLC Web Concordancer (the Virtual Language Centre) provides basic concordance search and retrieval functions using corpus files which are located on the VLC server.

Corpora in the Teaching of Languages and Linguistics by Tony McEnery and Andrew Wilson contains the authors’ summary of their book.

Vance Stevens’ article: Concordancing with Language Learners: Why? When? What?

Gregory Handley describes his experiment introducing Data Driven Learning with his Japanese students in Sensing the Winds of Change.

If you would like to look into concordances and corpora more deeply, try this tutorial by Catherine N. Ball.

Click here for the University of Stirling’s Introducing the concordance site.

Click here for an overview of The Use of Corpora in Language Studies.

Links to glossaries

Ø Glossary of terms used in concordancing literature

Ø w3 project glossary

Ø Princeton WordNet glossary

Ø Mannheim WordNet glossary

Ø Using computers in Linguistics glossary