V

ocabulary has been traditionally thought of as

Norbert Schmitt individual words. Of course, this layman's
Ronald Carter view is inadequate because vocabulary in-
cludes many units which are larger than individual
University of Nottingham orthographic words. For example, a single meaning is
attached to each of the following: give up, fish shop,
freeze-dry, and bum the midnight oil. Research using
large corpora has shown that these multi-word lexical
units (MWU) are ubiquitous in language use, at least
in English (Moon, 1997). This is especially true of
spoken discourse (McCarthy and Carter, 1997). More-
over, this same corpus research is now beginning to
indicate that there is lexical patteming that exists
beyond even the MWU level. Some longer strings of
language recur frequently and are often connected
with the functional usage of language. These longer
strings have been called various names, including
lexical phrases (Nattinger and DeCarrico, 1992), lexical
chunks (Lewis, 1993), lexicalized sentence stems (Pawley
and Syder, 1983), and ready made

Lexical (complex) units (Cowie, 1992). Fol-

lowing Schmitt (2000), we will
refer to these strings as lexical
phrases. This paper will argue that

Phrases in lexical phrases are a key element of

fluent language production and
that they also play a part in vo-
cabulary learning.

Language The Psychological

Lexical Phrases
Reality of

Most of the evidence for lexical

Learning phrases comes from research into

large databases of written and spo-
ken language called corpora. Al-
though these corpora contain the
linguistic production of countless
~~<:-li, §~!,*,'ln:.l3lt~lexical phraseaO native-speakers, the data contained is only the mani-
!t~t1f:-:JVl-C~ t ~ lexicalphr ases/i, 'liil~
0 festation of linguistic mental processing, and does
~ h, ~Jlil)'t!f:*1'f ~ h -cv' ~.ljj.~li'il);!~W<:- not give us direct insight into the processes them-
;l;,~ lexicalphrasesli, ~.liil) <Pf: 114 il);O't: i
0 selves. Of course, the fact that lexical phrases are so
~tl-c1'f~l-C.l3~, ~h~ .~~Rm~ commonplace in the corpora surely suggests that
m a-l!7Jlt-Cvl~ tv' -J ~ t 1;t/IA!jl§~li',*,<:-ME~JJ they are in some way important in the mental pro-
~h-CV'~o ~Vf.li, iflv'~~iia-T~p~, cessing of language. But do lexical phrases exist in the
lexicalphrasesa--t-il)~~il)i$k;[;'" t L-Cf<l'! -J ~ mind as single units, Le. do they have
t ;O{-C'l' J., -:Ji ~,if Lv'~~i5il)-il1l71{!1:tf:
0 psycholinguistic reality?
~.Iiil)<p<:-<:-~..t.;OL:>-Cv'J., t vl-J il)!i t -C{,~ To answer this question, it might be useful to first
~~~t<:-;l;,J.,o i~, ~1'~~~l~il)S. look at individual words. Aitchison (1987a) suggests
a-.m~t~M.l-C •• a-.MTJ.,il)<:-~~ words can be stored either as whales or composed
<, lexicalphrasesa- f~ -:>-C •• a-}l![ML -Cv ' on-line by combining stems and affixes. She reviews
J., lexicalphrasesld , I>iJ ~ '¥-f: t -:>-C b *l~<:-
0 psycholinguistic experimental results and concludes
;l;,J., it.:, lexicalphrases~§~!,*,'liiil)..t. <:-{,
0 that inflections are added on-line because they are
1t~<:-;l;,J.,o Ali, lexicalphrasesiJ{P;O'nli, rule-based and relatively consistent, and therefore
T <·f:-t-ha--t-il)t~JllGmf:::jJ-M i.. -t-h~il)~! can be added with little cognitive effort. On the
a- ~1~L -Cv' J., lexicalphrasesld, ~~IHifma-
0 other hand, prefixes and derivative suffixes are less
@.~~J.,1t~~~.<:-;l;,J.,il)<:-,m.~.~ transparent and are evidently retrieved from
J.h~hJ.,....z1' l:';l;,J.,o iiJ.7Ji:l;;f:-:Jv'-C!i, memory as polymorphemic whales. This is partly
t!~;O{~h!ia/Jl~d,:(ii.p~ <:-;l;,J.,o because the affixation is somewhat arbitrary and

6 The Language Teacher 24:8

__________________ -LFeature: Schmitt & Carder

needs to be memorized on a case-by-case basis. For makes lexical phrases so practical is that they are
example, there is little reason why the noun deriva- frequently used to express some functional use of
tion of adjust could not be adjustim1, but it just hap- language, thus warranting the memory space neces-
pens to be adjustment. (Of course, there are sary to store them.
phonological bases for some affix combinations;
they are simply easier to pronounce.) One example Production and Reception-Lexical Phrases
of the psycholinguistic reasoning which leads to the Enable Fluency
conclusion that derivations are stored as wholes If we accept that lexical phrases exist in the mind,
involves the 'bathtub effect' (Aitchison, 1987b: then what insights does this give us into the underly-
Chapter 11). This refers to the fact that the begin- ing psycholinguistic processes of language produc-
ning of words tend to have the highest saliency, the tion? One place to start is by looking into the
ends of words the next highest, and the middle of strengths and limitations of the mind itself. One re-
words the least. If stems were stored separately from source the mind seems to have plenty of is long-term
the affixes, one would expect the middle segment of memory capacity. However, resources in relatively
derivations to be the most salient, not the prefix. short supply are working memory and processing
Therefore, the evidence seems to show that many speed. This limits the amount of information the
words are stored as wholes. If this is true, the same mind can process at anyone time. The result is that
should be possible for multi-word strings. In an early the mind can store vast amounts of knowledge in
influential paper on lexical phrases, Pawley and Syder long-term memory, but it is able to process only
(1983) argue this is exactly the case. They do so by small amounts of it in real time, such as when one is
highlighting the 'puzzle of native-speaker fluency'. speaking (Crick, 1979). Pawley and Syder (1983) ar-
The puzzle is that native-speakers have cognitive limi- gue that the mind makes use of a relatively abundant
tations in how quickly they can process language but resource (long-term memory) to compensate for a
are also able to produce language seemingly beyond relative lack in another (processing capacity). It does
these limitations. Pawley and Syder look at the this with language by storing a number of frequently
psycholinguisitc literature and conclude that native needed word strings (lexical phrases) as individual
speakers have been shown to be unable to process a whole units which can be easily called up and used
clause of more than 8-10 words at a time. When without the need to compose them on-line through
speaking, they will speed up and become fluent dur- word selection and grammatical sequencing. In this
ing these clauses, but will then slow down or even way there is less demand on cognitive capacity be-
pause at the end of these clauses, presumably to for- cause the lexical phrases are already 'ready to go' and
mulate the next clause. Speakers seldom pause in the require little or no additional processing.
middle of a clause. Together, this evidence suggests By stringing lexical phrases together, native speak-
that speakers are unable to compose more than about ers are able to produce stretches of fluent language.
8-10 words at a time. Because the use of lexical phrases lessens the cogni-
Native-speakers can, however, fluently say multi- tive load, they are able to concentrate on the content
clause utterances. Consider the following examples: and organization of what they want to say. In con-
1) You don't want to believe everything you hear. trast to this, second language learners may often have
to pause between every 2 or 3 words because of the
2) It just goes to show, you can't be too careful.
need to process language as well as thoughts.
3) You can lead a horse to water, but you can't Lexical phrases do not always have to be used in
make him drink. production as invariable wholes. They can also be
They have increasingly more words, and Example 3 used as a ready-made scaffold upon which to build
is clearly beyond the limit of 8-10 words. Yet native language. An example of this is the lexical phrase
speakers can say them all without hesitation. Pawley
__ (person) thinks nothing of __ ing (verb)
and Syder suggest that these examples can be flu-
leading to
ently produced because they are actually lexical
phrases-prefabricated phrases which are stored as 1. Diane thinks nothing of hiking 20 miles.
single wholes and are, as such, instantly available 2. He thinks nothing of teaching six classes a day.
for use without the cognitive load of having to as-
3. Mike thinks absolutely nothing of flying ten
semble them on-line as one speaks. For this to work,
hours to go to a conference.
only sequences of words which are frequent and
familiar occurrences in language can be recognized With the basic pattern already available from a
as somehow being 'cohesive' and stored as wholes. lexical phrase, it would seem relatively easy to add
Less common combinations or strings will continue on modifiers and clauses to customize the pattern to
to generated via syntactic rules. If this were not true, the situation at hand. Once again, we can see that
our minds would become cluttered with countless the use of lexical phrases aid fluency, even when
word strings seldom, if ever, used. The thing that the language being produced is a creative adapta-

August 2000
7
Feature: Schmitt & Carter _

tion of an existing fixed pattern. the words sled and walked

Thus far, we have discussed the effect of lexical
2) System learning: learning the system or 'rules',
phrases on the productive side of spoken discourse.
such as sl=s+l and walkedewalk-ed
However, the use of lexical phrases also eases the
processing load of the listener. Because lexical Lexical phrases clearly fall into the category of
phrases can be recognized as individual wholes, this item learning, because their key feature is that they
spares the listener some of the processing effort of are wholes. Grammarian the other hand, falls into
parsing the language stream into individual words, the category of system learning. But these two types
recognizing each of those individual words, and of learning are not mutually exclusive; rather they
extracting meaning from the composite of the indi- feed into one another. Thus once a lexical phrase is
vidual word meanings and the related grammatical known, it can be analyzed and segmented into its
structuring. So language which includes recogniz- constituent words. In this way, unanalyzed phrases
able phrases is easier to understand. This can be re- can be analyzed to provide additional vocabulary.
lated to the cooperative principles described by Hakuta (1974) was the first to suggest that phrases
Grice (1975) which state that the speaker needs to could be analyzed into words plus grammar. Wong-
use language which is relevant, clear, and appropri- Filmore (1976: 640) also believes that L2 children
ately concise. The use of lexical phrases in language use many prefabricated phrases which "evolve di-
helps to realize these cooperative principles. Cer- rectly into creative language". Peters (1983) presents
tainly, lexical phrases are so commonplace in lan- the argument in its most considered form, propos-
guage that they are expected in any discourse. In ing that learning vocabulary from lexical phrases is
addition, the use of lexical phrases tends to make a 3-part process. First, lexical phrases are learned
language clearer. This is because most lexical which are frozen wholes with no variation possible.
phrases are the linguistic expressions of functional At this point they are unanalyzed and are single
language use. An example of this is Have you heard lexemes. Common examples are idioms e.g. kick the
the one about ..., which is used almost solely as the bucket; burn the midnight oil, and proverbs e.g. An
typical introduction to telling a joke or humorous apple a day keeps the doctor away; A stitch in time saves
story. This use is ritualized and can be anticipated nine. Also included are some expressions which are
whenever a speaker is preparing to tell a joke. Since tightly related to a functional use e.g. Ladies and
it can be anticipated, its meaning is instantly clear. Gentlemen ... which is a typical opening address in a
Also, since it is a ritualized, nothing else needs to be formal situation.
said in introducing the joke. Thus lexical phrases are Second, a language learner may realize that some
typically the most economical way of performing variation is possible in certain lexical phrases, and
functional language (sometimes called speech acts), that they contain open slots. For example, after hav-
which is a major reason why the formulaic use of ing heard the phrase How are you today? several times,
lexical phrases is so tightly related to functional lan- it may be acquired as a lexical phrase with the mean-
guage use. This also means that lexical phrases ing of a greeting'. However the learner may later no-
I

which realize functional language use are maximally tice the phrases How are you this evening? or How are
relevant to the functional situations in which they you this fine morning? At that point, the learner may
are used. Taken together, the use of preformulated realize that the underlying structure is actually How
sequences seems to have real advantages for both are you ?, where the slot can be filled with most
the speaker and listener. Thus it is not at all surpris- time references. The learner is then able to realize
ing that McCarthy and Carter (1997) found wide- that what fits in the slot is a separate lexical unit from
spread evidence of these sequences in their analysis the rest of the phrase, which opens the door to learn-
of the CANCODE2 spoken corpus. Common ex- ing that lexical unit. Thus phrases can be
amples are the thing is _ (meaning 'the problem/ segmentalized into smaller lexical units, oftentimes
point is _), and I see (I understand). individual words. Lexical phrases at this stage are
partly fixed and partly creative.
Learning Vocabulary by the Segmentation of Third, this segmentation process can continue until
Lexical Phrases all of the component words are recognized as indi-
We have seen how lexical phrases aid in language vidual units by use of syntactic analysis. When this
use, but it has also been argued that they play a part happens, every word in the lexical phrase is poten-
in vocabulary acquisition. To illustrate how this tially available for learning. This does not mean that
might work, we need to first introduce the idea of the segmentation process has to continue to this
item learning versus system learning. In language acqui- point; in fact it can stop at any stage. There are some
sition, learning seems to take place in two ways in lexical phrases which the learner may never start to
phonology, morphology, and other linguistic aspects: analyze, and which may be retained only as
unanalyzed wholes. Likewise, learners mayor may
1) Item learning: learning individual units, such as
not realize that certain lexical phrases contain vari-

8 The Language Teacher 24:8

__________________ ~Feature: Schmitt & Carter

ability and slots. When the variability is realized, it is enough input to distinguish their word-class.
possible that only the slots are analyzed; the rest of
the pattern may remain unanalyzed. Still, it seems Some Implications of Lexical Phrases
safe to assume many, if not most, of the lexical A main implication of lexical phrases is multiple
phrases a learner knows will eventually become fully storage in the mental lexicon. A large number of
analyzed, and Peters (1983) suggests that much of a lexical phrases are likely to be fully analyzed, even
learner's vocabulary is learned in this way. This is though they are retained in longer-term memory
especially true because learners are likely to eventu- because of their utility. Thus it is possible that the
ally know numerous lexical phrases, seeing how they production of a frequent sequence of words can
are easy to learn, efficient to use, and cover a wide stem from the retrieval of a lexical phrase, or from
variety of lexical content. the syntactic generation of the string from indi-
This segmentation process can lead to more than vidual words. (It would seem likely that the lexical
lexical knowledge however. Segmentation also re- phrase approach will be used when possible due to
quires grammatical information, which focuses at- the lower cognitive load.).
tention on syntax as well as lexis. Ellis (1997) argues This means that if a learner produces a sequence of
that grammar can be learned through the implicit words which contains an error, the source of the error
recognition of the patterns in strings of language, might be a weakness in lexical or grammatical con-
some of which are bound to include lexical phrases. struction, or it might be that a lexical phrase has been
In this line of reasoning, innate grammar would not acquired in a faulty manner. If the language error is
consist of an inborn understanding of grammatical due to a faulty lexical phrase, then any amount of
rules, but rather a facility for recognizing the sys- grammar-based correction would seem unlikely to
tematic patterns in language input. A perceptive remedy the error. What would be required is a re-
ability to recognize such patterning does seem to be learning of the correct form of the lexical phrase.
a sufficient condition for at least some types of One should not assume that this paper downplays
grammar acquisition; a model developed by Kiss the importance of grammar in language use or in
(1973) demonstrated that simply calculating which language teaching. The point is that language abil-
words occur sequentially eventually provides ity requires not only the ability to produce language

Figure 1. Suggestions for ways of teaching awkward and frequently occurring words such as just.

Keyword: just

Match these remarks and responses: Sometimes just is used to make a problem or
mistake seem less important or serious:
1. Would you like a cup of coffee?
2. Are you ready? It's time we were off. It just slipped my mind.
3. It looks as if the train is going to be late. I just couldn't get there any earlier.
4. Were you late last night? I'm just not going to get upset about it.
S. Everybody is worried about the situation.
Pres. Perfect: I've just passed my exam. (just = very
6. They've changed their mind again.
recently)
7. It's almost 9 o'clock. It's time we started.
8. Have you got Helen's phone number there? Pres. Cont: I'm just making some tea. (just = em-
phasizing exactly now)
a. That's just what we don't need.
was going to: I was just going to ring you. (just =
b. Oh, it's not just me, then.
very soon after now)
c. No, we got there just in time.
d. I think so. Just a moment-I'll have a look. Can I just ask/tell you/say that ... (If you know that
e. Not just now, thanks. an interruption will be quick)
f. Don't worry. I think everything is just about I was talking to her just now. (just now = a short
ready. time ago)
g. That's just what I expected.
I couldn't tell you just then. (just then = at that
h. Right, I'll just get my coat.
particular time)
Notice all the responses include the word just. It is very difficult to translate just, but it is used in a lot of fixed
expressions. Can you think of a similar word in your own language? Learn the responses so you can use them
yourself. Make sure you know the equivalents in your language.
Source: Lewis 1997. ©LanguageTeaching Publications

August 2000
9
Feature: Schmitt & Carter _

through syntactic generation (via grammatical com- Ellis, N.C. (1997). Vocabulary acquisition: Word structure,
petence), but also requires the ability to use lexical collocation, word-class, and meaning. In Schmitt, N.
phrases. This is especially true if learners hope to And McCarthy, M. (eds.) Vocabulary: Description, Acquisi-
tion, and Pedagogy. Cambridge: Cambridge University
gain the pragmatic fluency which comes from
Press.
knowing the right lexical phrase for the right situa- Grice, H.P. (1975). Logic and conversation. In Cole, P. and
tion. Ultimately, language learners need both abili- Morgan, ].L. (eds.) SYl1tax and Semantics Vo/. 3: Speech
ties to use language well. Acts. New York: Academic Press.
The importance of lexical phrases would suggest Hakuta, K. (1974). Prefabricated patterns and the emer-
that we need to include instruction on them in our gence of structure in second language acquisition. Lan-
language teaching. As this is a new area, it is not guage Learning 24.
yet clear how we can best achieve this. We do not Kiss, G.R. (1973). Grammatical word classes: A learning
process and its simulation. In Bower, G.H (ed.) The Psy-
even have a list of the most frequent lexical
chology of Learning and Motivation: Advances ill Research
phrases in English as of yet, although this gap is and Theory. Vo/' 7. New York: Academic Press.
now being addressed by Dave and jane Willis. Lewis, M. (1993). TIle Lexical Approach. Hove, England:
Michael Lewis (1997) has done some preliminary LTP.
work in the area of teaching lexical phrases, advo- Lewis, M. (1997). Implementing the Lexical Approach. Hove,
cating a focus on inducing patterns from language England: LTP.
input and the return of a limited amount of pat- McCarthy, M. and Carter, R. (1997). Written and spoken
tern drilling. An example of this approach is illus- vocabulary. In Schmitt, N. And McCarthy, M. (eds.) Vo-
cabulary: Description, Acquisition, and Pedagogy. Cam-
trated in Figure 1. Lewis's proposals are generating
bridge: Cambridge University Press.
a great deal of interest and seem intuitively attrac- Moon, R. (1997). Vocabulary connections: Multi-word
tive, but at this point it must be said that there is items in English. In Schmitt, N. And McCarthy, M. (eds.)
little empirical evidence one way or the other as to Vocabulary: Description, Acquisition, and Pedagogy. Cam-
their actual effectiveness. bridge: Cambridge University Press.
Nattinger, ].R. and DeCarrico, ].S. (1992). Lexical Phrases
Conclusion and language Teaching. Oxford: Oxford University Press.
Corpus research is making it clear that the pattern- Pawley, A. and Syder, F.H. (1983). Two puzzles for linguis-
tic theory: Native like selection and native like fluency.
ing resulting from lexical phrases is a major com-
In Richards,]. and Schmidt, R. (eds.) Language and Com-
ponent of language. As SUCh,lexical phrases are munication. London: Longman.
likely to become an increasingly important topic in Peters, A. (1983). The Units of Language Acquisition. Cam-
Applied Linguistic circles, simply because lexical bridge: Cambridge University Press.
phrases are a key element in how language is used. Schmitt, N. (2000). Vocabulary in Language Teaching. Cam-
In light of their essential nature, we need to come bridge: Cambridge University Press.
to a better understanding of their behavior and Wong-Fillmore, L. (1976). The Second Time Around: Cogni-
develop innovative ways of incorporating lexical tive and Social Strategies in Second Language Acquisition.
Stanford University: Unpublished doctoral dissertation.
phrase instruction into the language syllabus.
Norbert Schmitt lectures at the
Note University of Nottingham, UK in
1 This is a revised version of a presentation given at the vocabulary studies, language test-
1998 PASE (Polish Association for Studies of English) ing, and TESOLmethodology. He
conference in Szczyrk, Poland. recently published Vocabulary in
2 The CANCODE (Cambridge and Nottingham Corpus of
Language Teaching with Cam-
Discourse English) is a 5 million word corpus of
bridge University Press. He likes
unscripted spoken discourse compiled at the University
of Nottingham in conjunction with Cambridge Univer- traditional Irish music and often
sity Press. dreams of renewing his private
pilot's license.
References [email protected]
Aitchison,]. (1987a). Reproductive furniture and extin- Ran Carter is professor of Mod-
guished professors. In Steele and Threadgold (eds.) Lan- ern English Language at the Uni-
guage Topics. Essays in Honour of Michael Halliday, Vol. 2. versity of Nottingham, UK. His
Amsterdam: John Benjamins. latest book on vocabulary is Vo-
Aitchison,]. (1987b). Words in the Mind. Oxford: Blackwell.
cabulary: Applied Linguistic Per-
Cowie, A.P. (1992). Multiword lexical units and communi-
cative language teaching. In Arnaud, P. and Bejoint, H.
spectives (2nd ed., Routledge). It
(eds.) Vocabulary and Applied Linguisitcs. London: would take a lot of money to get
Macmillan. him in a plane flown by Norbert
Crick, F.H.C. (1979). Thinking about the brain. Scientific Schmitt.
American No. 9. ronald.cartercsnottingham.ac.uk

10 The Language Teacher 24:8