Barkema, H. (1996)

PDF hosted at the Radboud Repository of the Radboud University
Nijmegen
The following full text is a publisher's version.
For additional information about this publication click this link.

http://hdl.handle.net/2066/28061
Please be advised that this information was generated on 2024-04-14 and may be subject to
change.
IDIOMATICITY AND TERMINOLOGY: A
MULTI-DIMENSIONAL DESCRIPTIVE
MODEL
Henk Barkema
Abstract. In this article we discuss the traditional terminology for idioms and other
lexicalised expressions. We claim that it is inadequate for accurate, explicit and
detailed descriptions of the various linguistic characteristics of expressions. As an
alternative, we introduce a well-defined model with which three different types of
linguistic characteristics can be captured with a small set of labels: composition-
ality, collocability and flexibility. What is new in our approach is that a set of labels
is given to one expression instead of only one label Each of these either indicates
how the expression scores on a given scale or indicates additional characteristics of
the expression (e.g. idiosyncratic syntax). In an appendix we provide an inventory
of terms discussed in this article, with their definitions.
1. Introduction1
The introduction of new terminology for a topic that has been explored
quite thoroughly already is a risky enterprise. By adding a new set of terms
to one which has become firmly established, one inevitably runs the risk of
causing confusion where one wanted to bring about clarity. However, the
various traditional classifications of idioms, and other lexicalised
expressions in articles and handbooks of grammar are generally used to
categorise these expressions under single labels. In this way the impression
is given that expressions are homogeneous in their characteristics .2 In our
opinion, however, the traditional one-dimensional classifications can only
be used to indicate which expressions have more or less the same
characteristics, because generally the types of characteristics of lexicalised
expressions are only loosely related. It is true that practically all such
expressions are more or less limited in their morpho-syntactic freedom,
that generally substitution of lexical items in these expressions is limited to
some extent and that there are many with idiosyncratic semantic charac
teristics. But if one tried to create classes with exactly the same character
istics, the result would be a number of classes that is far greater than the
traditional set of three, which distinguishes between idioms, restricted
collocations and free constructions. To illustrate the weakness of the
present classifications, we discuss two examples .3
1 l would like to thank Wietske Vonk, Peter-Arno Coppen, Martin Everaert and the
anonymous referee of Studia Linguistica for their valuable comments on earlier versions of
this paper, I am very grateful to Jan Aarts; many parts of this article are the results of
discussions we had together.
* See also e.g. Fernando & Flavell (1981:17 ff.).
The definitions of words and expressions in this article are from Longman Dictionary o f
English Language and Culture (Summers 1992, henceforth L D O E L C ), unless other
references are given.
Linguistica 50(2) 1996, pp. 125-160, © Tlic Editorial Board of Studia Linguistica 1996.
Published by Blackwell Publishers, 108 Cowley Road, Oxford 0 X 4 1JF, UK, and
238 Main Street, Cambridge, MA 02142, US A
126 Henk Barkema
(1) the bottom/first rung of the ladder (‘the lowest level in the organiza
tion’)
Expressions like this one are traditionally classified as ‘idioms’ because of
their non-compositionality. Alternative terms are ‘figurative idiom’
(Cowie 1983) and ‘non-compositional metaphor’ (Wood 1986). Cowie
uses the term “figurative idiom’, because people will generally be able to
imagine a plausible relation between the idiomatic and the literal meaning
of the expression. Although the expression is non-compositional, Wood
(1986) does not use the term ‘idiom’ because lexical substitution is
possible: bottom can be replaced with synonyms, antonyms, near
synonyms, etc. (see also section 2 . 1). Such substitutions are not possible in
an idiomatic expression as hook, line and sinker (‘with expressions of
belief: completely’). The present terminology cannot adequately label
these different degrees of lexical freedom. Finally, the expression can take
various different syntactic forms, for example ladder can be preceded and
followed by modifiers. This is not possible in hook, line and sinker, which
can take one form only. This difference in morphosyntactic flexibility
cannot be expressed with the traditional terminology either.
(2) bed and breakfast (‘(a private house or small hotel that provides) a
place to sleep for the night and breakfast the next morning for a fixed
price’)
This expression is different from example 1 because the meanings of its

lexical items form part of the meaning of the expression. It is traditionally
classified as 'binomial5(see Quirk et al. 1985) or ‘collocation* (see Wood
1986, and Cowie et al. 1983). The term ‘binomial’ highlights the presence
of two nouns in the expression, while ‘collocation" focuses on its limited
lexical freedom. These terms do not show that the expression has an
institutionalised meaning which is to some extent idiosyncratic, because it
is more than the sum of the meanings of the lexical items in (and the
syntactic structure of) the expression (see the definition of example 2).
This is not the case with Just Married for example, which has a special
pragmatic function, but no idiosyncratic meaning. No labels are available
to express the difference in meaning between this expression and example
2; nor are there labels to indicate that bed and breakfast is more
morphosyntactically frozen than, for instance, developing country, which
can be pluralised. Finally, although breakfast in example 2 can be replaced
by board, no further lexical variation is possible, as in day after day, in
which day can be replaced with near-synonyms as night, second, month,
year, etc., while it has more freedom than e.g. wishful thinking, which is
lexically invariable. Again: no terms are available to indicate this dif
ference.
On the basis of these two examples it is safe to conclude that the variety
of characteristics among lexicalised expressions is so great that if one used
© T h e Editorial Board of Studia Linguistica 1996
Idiomaticity and terminology 127
the traditional classifications, one would be bound to sweep a lot of

valuable information under the carpet. It is true that the various features
can be described in terms of degree of compositionality, and of colloca
tional and morphosyntactic freedom, as we have just done. However, such
descriptions lack conciseness and precision.
In this paper we introduce a multi-dimensional model which can be
used for much more explicit and precise classifications of lexicalised
expressions than have so far been possible with the traditional ter
minology. Another advantage of the model is uniformity in description,
because the characteristics of an expression are pinpointed with a fixed set
of parameters. Although traditional classifications can only be used for the
description of types, our model can also be used for tokens. Not only can
one describe whether a token is a base or variant form, but also, if it is a
variant form, which types of variation it has.
2. Discussion of the literature

2.1. Idiom
For more than a century the majority of linguists have used the term
‘idiom' to refer to lexicalised expressions with idiosyncratic meanings.4 An
example is Henry Sweet’s (1889:139) definition:
The meaning of each idiom is an isolated fact which cannot be inferred from
the meaning of the words of which the idiom is made up.
This can be regarded as the standard definition of the term and, generally
speaking, later definitions boil down to the same two things: (a) idioms are
expressions which contain at least two lexical items and (b) the meaning of
an idiom is not the combinatorial result of the meanings of the lexical items
in the expression. A more recent example of such a definition is:
... I shall regard an idiom as a constituent or series of constituents for which
the semantic interpretation is not a compositional function of the formatives
of which it is composed (Fraser 1976:103).
Although practically all definitions of idioms require them to be ‘lexically
complex’, an exception to this rule is Charles Hockett’s definition
(1958:172), which implies—as Hockett admits—that words and mor
phemes can be idioms too and that all forms which are not idioms
themselves, contain them:
Let us momentarily use the term for any grammatical form the meaning of
which is not deducible from its structure. Any Y, in any occurrence in which it
is not a constituent of a larger Y, is an idiom.
In this sense idioms are the basic semantic units of a language, no matter
whether they are morphemes, words, phrases, clauses or sentences. As the
4
See also Wood (1986).
©The Editorial Board nf Studia Linguistica 1996.

128 HenkBarkema
definition is not restricted to expressions with at least two lexical items,

however, it has been rejected as impracticable by the majority of linguists.5
To be able to distinguish between expressions which contain both
idiomatic and non-idiomatic parts (such as rain cats and dogs) and ones
that are wholly idiomatic, additional terms have been introduced. For
example, van der Linden (1992:8) distinguishes between ‘idiomatic
expressions’ and ‘idioms’:
Idiomaticity is a property of aspects of the meaning of complex (multi-
lexemic) expressions. Idiomaticity implies that these aspects are exclusively a
part of the meaning of the expression as a whole.
An idiomatic expression is an expression some aspect(s) of the meaning of
which is (are) subject to idiomaticity. Idioms are expressions all aspects of the
meaning of which are subject to idiomaticity.
In our opinion, there is nothing wrong with well-defined classifications
that are focused on one linguistic characteristic of lexicalised expressions
(in this case compositionality). However, occasionally linguists have
included other types of characteristics in their idiom definitions. For
example, in the following quotation the phrase ‘wholly non-productive in
form’ means that lexical substitution (by synonyms, near-synonyms or
antonyms) is impossible:
An idiom is a complex expression which is wholly non-compositional in
meaning and wholly non-productive in form (Wood 1986:2).
Although Wood’s (1986) description of compositionality is thorough and
at the same time allows for different degrees, we regret the inclusion of
productivity in her definition of ‘idiom’ because the two different types of
characteristics of lexicalised expressions—non-compositionality and non-
pro ductivity—are not mutually inclusive. As a consequence of her
definition an idiomatic expression as throw in the towel (‘to admit defeat’)
cannot be regarded as such, because one of its lexical items {towel) can be
replaced by another [sponge); instead it is classified by her as a ‘non-
compositional metaphor’. In this way the set of idioms is restricted to a
subset of expressions that are generally regarded as idioms, not because of
their non-compositionality, but as a result of another linguistic charac
teristic. In this case we would opt for a classification with two subsets of
idioms, namely one containing ‘wholly nonproductive’ idioms and the
other containing idioms which are only productive to some extent.
Some linguists have included another type of characteristic, namely the
relation speakers can perceive between the literal and the non-
compositional meanings of idiomatic expressions, sometimes called
‘warranty’ or ‘motivation’.6 If such a link cannot be found, an expression is
3See for similar approaches e.g. Fernando & Flavell (1981:23), or Williams (unpublished
manuscript) who uses the term ‘semantic idiom’.
6 See Carter (1987) and Fernando & Flavell (1981:32). Geeraerts (1992) and Ullmann
(1973) have adopted this term from de Saussure (191 6 ) ~ see Fernando & Flavell (1981:28),
© T he Editorinl Board of Studia Lingüistica 1996.

‘opaque’ or ‘fully unmotivated’. The reason why they have added this
characteristic is that they regard both compositionality and motivation as
one and the same semantic characteristic; an example is Fernando &
Flavell (1981:29ff), who label a fully unmotivated expression as ‘pure
idiom’. In our opinion motivation and compositionality are two different
types of characteristics: compositionality concerns the relationship
between basic senses of the lexical items in an expression and the meaning
of the whole expression on the other and is, therefore, not restricted to
idiomatic expressions, but also concerns lexicalised literal ones (see
example 2; bed and breakfast), whereas motivation depends on the judge
ment of speakers about the relations between the literal and the
institutionalised metaphorical meanings of idiomatic expressions (see
§3.2). In relation to this distinction Cruse (1986:44) distinguishes between
the terms ‘opaque’ (a form of motivation) and ‘translucent* (a form of
compositionality),
Although linguists have often taken the view that constructions are
either fully compositional or fully non-compositional, there are a number
of lexicalised expressions that can be regarded as compositional to a
certain extent. With the exception of Cruse (1986:39-40), this fact has
received little attention in the literature* In section 1 we briefly discussed
bed and breakfast, whose meaning is more than the combination of the
meanings of bed and breakfast (see also §3.2).
22. Collocation
Other terms that have been used for a long time arc ‘restricted collocation’
(Aisenstadt 1979:71) and ‘frozen 00110031100’ (Backlund 1976). This type
of collocation differs from another type of construction—‘free combina
tion’ (Benson et al. 1986:xxiv), ‘free construction’ (Cowie 1981 and
Altenberg & Eeg-Olofsson 1990), ‘free word-combination’ (Aisenstadt
1979:71), ‘unrestricted collocation1 (Carter 1987:63) or ‘free combina
tion’ (Quirk et al '1985, § 16.3)—in that the substitutability of its lexical
items is limited:
Restricted collocability in English typically takes the form of word-

combinations the constituents of which are restricted in their com mu lability.
Such word-combinations we term ‘restricted collocations’ (RCs) (Aisenstadt
1979:71).
However, this characteristic is typical of idioms too. For example, Carter

(1987:58) describes idioms as cnon~substitutable’ or ‘fixed' collocations.
Generally the collocability of collocations is much greater than that of
idioms, but idioms like a drop in the bucket/ocean and the highest/top/
lowest/bottom rung of the ladder allow several alternatives, while the res
tricted collocations foot the bill, curry favour and catch one's breath allow
no substitution at all (see Cowie (1981:228), or Cruse (1986:41), who
0 The Editorial Board of Studia Linguistica 1996.
130 HenkBarkema
refer to such constructions with the term 'bound collocation’). On the

other hand there are expressions which are limited in their range of
collocability but are classified as free constructions.7 It is in fact not their
collocational restriction that sets restricted collocations apart from
idioms, but their special meaning characteristics. Aisenstadt (1979:71)
describes the difference between restricted collocations (RCs) and idioms
as follows:
RCs are not idiomatic in'meaning: they do not form one semantic unit; their
meaning is made up as the sum of the meanings of their constituents
(Aisenstadt 1979:71),
It is true that such expressions do not form one semantic unit, However,
such restricted collocations have special meaning characteristics. For
example, Nicolas (1995:234) claims that e.g. sharp turn and take a
decision are semi-compositional because in each expression one of the
lexical items has a 'specific, “non-free” sense’. In the same vein Cowie et al.
(1983:x) note that:
one word (i.e. in the case of two-word expressions) has a figurative sense not
found outside that limited context The other element appears in a familiar,
literal sense (cf. the verb and noun, respectively, in jog one's/sb's memory) and
the adjective and noun in a blind alley). Some members of this category allow
a degree of lexical variation (consider, for instance, a cardinal error, sin,
virtue, grace), and in this respect ‘restricted5 collocations resemble ‘open
ones\
Because of their semantic characteristics Cowie et al. (1983:x) decide to
use the term ‘semi-idiom’ for such expressions:
... one word may have a common, literal meaning, while the other has a
specialised sense which may be difficult to grasp. Examples of such ‘semi-
idioms’ are foot the bill and sink one's differences (where the first word in both
cases has a figurative meaning).
In our opinion the term ‘semi-idiom’ is more appropriate than ‘restricted
collocation5 because it is really the semantic characteristics of such
expressions that are essentially different from those of idioms, and not the
collocability characteristics. In this respect such semi-idioms are similar to
ones like rain cats and dogs that van der Linden (1992:8) classifies as
‘idiomatic expressions’ because .. some aspect(s) of the meaning of
which is (are) subject to idiomaticity’. The only difference between foot the
bill and rain cats and dogs is the number of (lexical) items in the idiomatic
part.
In line with the traditional classification of constructions, Aisenstadt
claims that:
All word-combinations in present-day English can be divided into idioms and
non-idiomatic phrases. The latter constitute the vast majority of the word-
7 See Cowie (1981:226) and Greenbaum (1970:9).
© The Editorial Board o f Studia Linguistics \ 996

Idiomaticiiy and terminology 131
combinations functioning in speech. They can in turn be subdivided into free

phrases and RCs. (Aisenstadt 1979:71).
Cowie (1981) makes another division, namely one between free construc
tions on the one hand and idiomatic expressions (idioms and semi-idioms)
on the other; for a com parison of the two classifications, see A lexander
(1987:109). In this article we have a similar classification. In addition we
introduce a separate classification for the collocability characteristics of
constructions parallel to one for compositionality.
2.3. Transformational deficiency

A type of characteristic of lexicalised expressions that was m ore or less
ignored until the sixties, but which has received a lot of attention since
then, is their limited morpho-syntactic freedom or ‘limited flexibility ’,8 In
addition to limited compositionality and restricted collocability, limited
flexibility is the third linguistic type of characteristic by which lexicalised
expressions differ from other types of construction.
In the same way as restricted collocability is not a characteristic of
idioms only, there are many (more or less) compositional lexicalised
expressions with limited flexibility:
Although an examination of their transformational behaviour offers insights
into idioms as syntactic objects, it does not throw much light on idiomaticiiy
per se, Not only are transformational constraints insufficient to distinguish the
idiomatic from the non-idiomatic, they also fail to distinguish between semi
literal idioms with affective idiomatic adjectives (my dear lady, that wretched
woman, you poor man, etc.), semi-literal idioms with non-affective idiomatic
adjectives ( white lie , blue film, red cent, etc.), completely non-literal idioms
(white elephant, blue stocking, turn a blind eye , etc.) and peculiarities of usage
(criminal law , an utter fool, an outright lie, a real hero, drunken quarrels, mis
cellaneous articles, etc.). All the adjectives in these items share the feature that
they can be used only attributively. Predicative use, nominalisations and
inflection for degree do not generally apply to these adjectives. On the basis of
transformational deficiency alone, a feature, incidentally, which contributes to
the status of these items as institutionalised set phrases, they would be classed
as idioms of equal status (Fernando & Flavell 1981:43,44).
We agree with their conclusion that
... idioms are at best only a sub-class of all transformationally deficient
structures. All idioms are transformationally deficient, but not all transforma
tionally deficient structures are idioms. Therefore any attempt to set up
syntactic deviance as the primary norm to establish idiomaticity fails, since it
cannot distinguish between idioms and these other transformationally
deficient structures, (p. 44).
8 See e.g. the transformational linguists Katz & Postal (1963), Fraser (1970) and
Newmeyer (1974). See also Weinreich (1969), Chafe (1970), Gazdar el al. (1985), Carter
(1987), Healey (1968), Fernando & Flavell (1981), Pawley & Syder (1983) and Cowie et al,
(1983).
©The Editorial B o a rd o f Studia Linguistic« 1996,

132 Henk Barkema
As idiomaticity and limited flexibility are not mutually inclusive, it is

necessary to introduce a separate dimension for flexibility, in addition to
ones for idiomaticity and collocability. W e admit that at first sight a multi
dimensional classification seems less convenient than the traditional
system which applies one label to one construction. However, we believe it
is m ore realistic, and better expresses the complex interaction between the
types of characteristics of such constructions.
2.4. Institutionalisation
The fact that idioms and semi-idioms are institutionalised is usually taken
for granted. A number of linguists, however, have given explicit attention
to the notion of institutionalisation. F or instance, one of M akkai’s (1972)
five criteria for the identification of idioms is the extent to which a
construction has become current .4 Two examples of descriptions of this
type of characteristic of idioms and semi-idioms are:
By the institutionalisation of idiom we mean the regular association in a
speech community of a given signification with a given syntactic unit (a
compound, a phrase or a sentence), such that the resulting expression is
interpreted non-liter ally. In other words, part of the phenomenon of
idiomaticity is the institutionalisation of an asymmetry between sense and
syntax in the case of compound, phrasal and sentential idioms. (Fernando &
FIaveil 1981:44).
What makes an expression a lexical item, what makes it part of the speech
community’s common dictionary, is, firstly, that the meaning of the expres
sion is not (totally) predictable from its form, secondly, that it behaves as a
minimal unit for certain syntactic purposes, and third, that it is a social institu
tion (Pawley & Syder 1983:209; the italics are present in the original text).
Carter (1987:59) uses the term ‘fixed expression 1 to refer to constructions

.. with different degrees of possible fixity or “frozenness”, both syntactic
and semantic’. H e adds that:
... there is no real advantage in drawing strict lines between idioms and non
idioms or in treating collocations separately from idioms; instead it makes
more sense to try to illustrate the different degrees of variability of fixed
expressions (p. 135).
On the other hand there are many ‘familiar utterances’ (see Moore &
Carling 1982) or ‘speech formulae 1 (see Pawley & Syder 1983)* Some
examples from Pawley & Syder (1983:206) are: Can I com e in ?, A re you
ready?, I enjoyed that , D o what you're to ld . Although such constructions
are institutionalised, they cannot be regarded as ones with an ‘asymmetry
between sense and syntax’ or as ones that are ‘semantically and
syntactically fixed’. It seems, therefore, that in this respect there are not
only two types of construction (namely institutionalised constructions
y See also Fernando & Flavell ( 1981:19).
© T h e Editorial Board oFStudui Linguistic» ) 996,

which are, and ones which are not, fixed) but also two parallel types of
process: a socio-linguistic one during which constructions gradually
become m ore current, and a linguistic one during which they gradually
become m ore semantically, syntactically and collocationally fixed.
3. Introduction of a multi-dimensional model

In our discussion o f the literature we distinguished betweeen four types of
characteristics: institutionalisation, collocational restriction, idiomaticity
and transformational deficiency. A t the same time we noted that the rel
ations between these characteristics arc complex. As little is know n about
the exact interaction between the various types of characteristics, it is
impossible to create classes on the basis of this interaction. F o r this reason
we introduce a descriptive model in this section, which is fundam entally
different from the traditional one: instead of providing each construction
with one label which highlights only one striking feature, we classify
expressions at different dim ensions o f description, thereby sketching a
rich and transparent profile with no more than a few terms. In section 4 we
discuss a num ber of examples, and provide them with multi-dimensional
classifications,
In this paper we use the term ‘construction’ to refer to a phrase, clause o r

sentence with at least two grammatical function slots realised by lexical
items other than the definite or indefinite article. Because constructions
have at least two function slots, compounds such as the following are
excluded from the class of lexicallsed expressions: ice-cream , b a b y
carriage, sid ew a lk , shoulderblade , kneecap and sid eca r , These com
pounds have one function slot only, namely that of ‘head o f the noun
phrase’, as opposed to expressions like blind alley , which has two: ‘p re
modifier’ (b lin d ) and ‘head of the noun phrase 1 (alley).
Various linguists have stressed the scalar nature of linguistic character**
istics. Examples are Fernando & Flavell (1981), Alexander (1987) and
Carter (1987); for a detailed discussion: see Wood (1986). In our m odel
each type of characteristic has its own scale. Because of their scalar nature,
an infinite num ber of classes could be distinguished by cach time choosing
slightly different positions on each of the three scales. Each class would
then be only slightly different from another, Obviously such a
classification would be impracticable. Therefore we have divided each
scale into three parts—two extremes and one middle section, For cach
scale we introduce terms for the different parts. We try to m ake these
terms correspond as much as possible with the traditional terminology.
In our view the choice for a multi-dimensional model has two theoretical
advantages: each classification is related to a specific dimension of descrip
tion-syntactic, semantic or other, and is therefore unambiguous. A t the
same time complex relations between various types of characteristics can
OThe Editorial Hoard of Studia Lingüistica 1996.
134 HenkBarkema
be made explicit. A practical advantage is that with a few labels an accurat

and detailed profile can be given of each construction.
In the field of lexicography a num ber of linguists have broken nev
descriptive ground in this respect; for example, in their dictionary Cowic
et al. (1983:xxi-xxxvii) present the collocability characteristics of expres
sions and use separate grammatical codes for their morphological and
syntactic characteristics. Also, the Collins C O B U ILD English Language
Dictionary (Sinclair 1987) provides separate descriptions of the flexibility
characteristics of lexicalised expressions .10
A linguist who has already distinguished between different scalar types
of linguistic characteristics is C arter (1987:63-4). He briefly discusses
three different scales: collocational restriction, syntactic structure and
semantic opacity in the form of \ .. sets of continua with fixed points but
several intermediate categories1; for examples of classes of lexicalised
expressions and for references to articles in which these classes are
discussed: see Carter (1987:59). In C arter’s classifications on the various
scales there appear to be some snags. On his ‘collocational restriction’
scale he distinguishes between ‘unrestricted’, 'semi-restricted’ and
‘restricted’ collocation, but he has an additional type—‘familiar colloca
tio n —between restricted and semi-restricted. It is not clear to us why this
type is necessary. Although it seems that his collocability scale has two
extremes (namely ‘unrestricted’ on the one hand and 'restricted’ on the
other) he remarks that restricted collocations are ‘generally m ore fixed
and closed’, thereby leaving unclear whether such constructions can only
occur at the extreme, or also somewhere near it.
On his ‘syntactic structure’ scale we find both ‘irregular’ and ‘flexible1. It
is true that constructions with idiosyncratic syntax are generally inflexible,
but the reverse is not. For this reason we prefer a distinction between
descriptions of the morpho-syntactic structures of constructions on the
one hand, and descriptions of their (scalar) flexibility characteristics on
the other. Carter’s third scale concerns ‘semantic opacity’: this scale seems
to involve both compositionality and motivation characteristics—two
types we prefer to keep separate (see section 2.1). Nevertheless the ideas
behind these scales form a valuable foundation for the construction of a
solid model. In the rest of this article we give a detailed description of
various types of characteristics of constructions. In addition we introduce
and define terminology which can be used by grammarians and lexi
cographers for the specification of the various (types of) characteristics
(see the appendix for the definitions).
m See Sinclair (1987:1076): the entry of ‘PHR\
© T h e Editorial B o a rd o f Studia Linguistica 1996.

3.1. Familiarity and lexicalisation

In English and other natural languages, constructions are generally
created by combining two or more lexical items in a grammatical
structure. W e call them ‘standard constructions’. Examples are:
(3) a friendly w om an ('a fully grown human female who is acting o r ready
to act as a friend’)
(4) buy a house (‘to obtain a building for people to live in by giving money
(or something else of value)’)
On the other hand many constructions have become institutionalised. The
process of institutionalisation is a sodo-lingu istic one: a (standard) literal
or metaphorical construction gradually becomes m ore current within a
language community because of its specific meaning a n d /o r pragm atic
function." W hen it has a higher than average currency in the language
community, we call it a ‘familiar construction’ or ‘expression’. Exam ples
are paddle o n e ’s own canoe o r b e d a n d breakfast , but also ones as What
time is it? and D o what you're told! Pawley & Syder (1983:205) call such
constructions ‘memorized sequences’:
i
These are strings which the speaker or hearer is capable of consciously

assembling or analysing, but which on most occasions of use arc recalled as
wholes or as automatically chained strings.
From the start of the process of institutionalisation an expression can go
through a gradual process of lexicalisation, which is linguistic in nature:
the application of general gram mar rules to the expression becomes m ore
and more subject to restrictions, with as a result limited flexibility as well
as limited collocability a n d /o r compositionality. An example of an expres
sion which is gradually becoming lexicalised is a new w orld o rd er . Initially
it was used by President Bush to refer to the global balance of power after
the Gulf W ar in 1991, but one year later it was also used to refer to a new
balance of power in m ore general terms, as is illustrated with the following
quote: A new world order is emerging in Paris’s gilded salons (T h e
Independent , 3 0 /7 /1 9 9 2 ).
These restrictions on generally applicable grammar rules are generally
‘idiosyncratic', i.e. they cannot be formulated in terms of (general) restric
tion rules and it is impossible to pinpoint exactly under what conditions
the grammar rules are restricted.
At the beginning of the process of lexicalisation the expression is still a
‘free expression’: the general grammar rules are not subject to idio
syncratic restriction when they are applied to it. As soon as the application
of one type of (syntactic, semantic or lexical) rule to an expression
becomes subject to idiosyncratic restriction, it can be regarded as a
11 For a discussion of the term ‘currency’: see Aarts (1991:57-8).
©The Editorial B oard o f Studia Linguistic« Î 99fi,

136 Henk Barkema
‘restricted expression’. Finally, when it has limited m orphosyntactic

freedom and at the same time is collocationally restricted an d /o r no
longer compositional, it has become a ‘lexicalised expression ’.12 Examples
are:
( 5 ) cup an d saucer ('a cup and a saucer as a pair’)
(6 ) A s thick as two short planks (‘very stupid indeed’)
It should be noted that expressions do not necessarily go through a
process of lexicalisation. For instance, What tim e is it ? and D o what you 're
told! are unlikely to become lexicalised. A t the same time a large number
of constructions are restricted only collocationally, and for this reason
cannot be regarded as lexicalised. A n example is:
( 7 ) to work/operate/run a machine (not e.g.: *to manage a m achine ,
conduct a machine)
Combinatory dictionaries such as BBI (Benson et al. 1986) pay special
attention to such constructions, while lexicalised expressions are dealt
with in handbooks of grammar and dictionaries. The different types of
construction can be classified in the following way:
Sociolinguistic classification:
. ... -------► one option
familiar ,
— — several options
construction <
standard
Linguistic classification:
idiosyncratic compositionality
restricted^ —^-idiosyncratic flexibility ►
lexicalised
construction ^idiosyncratic collocability
^ "free
Expression:
familiar construction: expression

lexicalised familiar construction: lexicalised expression
3.2. Com positionality

The notion of compositionality is based on that of ‘sense’. We distinguish
between three types: ‘basic’, ‘extended’ and ‘derived’. W e use the term
‘sense 5for one or more concepts which can be denoted by one lexical item.
12 In Barkema (1993, 1994a and 1994b) we used the term ‘received1 instead of
‘lexicalised’ because we did not yet distinguish between institutionalisation and lexicalisation.
© T h e Editorial Board o f Studia Linguistica 1996.

The basic sense of a lexical item is the one from which other senses can be
s y s t e m a t i c a l l y derived, i.e. by means of semantic rules (see A arts &
Calbert 1979). It is its most obvious sense, that is: it is the first that comes
to a speaker’s m ind when s/he comes across a lexical item as a single word,
and not a text, sentence, clause or phrase containing it - for example when
s / h e sees it on a blank sheet of paper, or, to quote Burgess (1969:298): \ ..
as words which carry meaning if chalked up singly or written in the sky by
sky-writing aircraft’. It is often the most concrete sense, but not necessarily
the one that is used most frequently. F or instance, the basic sense of house
is ‘a building for people to live in’.
An extended sense of a lexical item is a temporary extension of its basic
sense. It is the unique result of a violation of restriction rules. Such
violation can only occur in context. As a result of the violation, sense
adaptation rules are activated. There are three factors which play an
important role in the activation of adaptation rules: (a) the basic sense of
the lexical item; (b) the speaker’s stereotyped beliefs about the referents of
this basic sense, and (c) the syntactic environment in which it occurs.
This notion of violation is the central point in this description. In every
syntactico-semantic construction lexical items impose restrictions on the
selection of others. These selection restrictions are determ ined by the
basic senses of lexical items. For example, such restrictions are violated in
(8) But most Czechs believed until now that Slovakia would still be in
favour of a federal h ou se . ( The E co n o m ist , 1/8/1992)
Here house no longer means ‘building1, but: ‘country’. The result of the
violation of restriction rules is an extension of the basic sense of house.
A derived sense of a lexical item is an extended sense which has becom e
established. Like the basic sense, it can also be reproduced by a speaker
when s/he comes across a lexical item as a single word. However, a derived
sense is less obvious than a basic one; it generally is the second or
following sense that comes to a speaker’s mind when s/h e is confronted
with a lexical item as a single word. As opposed to extended senses,
derived ones can be found in dictionaries. A n example of a lexical item
with a derived sense is h ou se ; one of its derived senses in L D O E L C is:
‘(the members of) a law-making body, esp, when one of two1; an example
is:
(9) Sir Marcus divided the house and did not divide without organisation.
Hughes’s motion was roundly defeated. (N ew Statesm an & Society,
2/8/1991).
From a diachronic point of view on language, derivation is a continuous
process: gradually new senses are introduced into the language, while
existing basic and derived senses may disappear, and derived senses can
gradually become basic senses.
On the basis of the notions of basic, extended and derived sense we can
© The Editorial B oard of Studia Llnguistica 1996.
138 Henk B arkema
now describe ‘compositionality’. The compositionality of a construction is

the extent to which its meaning is the combinatorial result of the basic or
derived senses of the lexical items in the construction and the syntactic
relations in the constituent that contains these lexical items.
The meaning of a standard literal construction originates from a
combination of the basic or derived senses of the lexical items in that
construction and the syntactic relations in the constituent which contains
these lexical items. A standard literal construction does not contain lexical
items with extended senses. As its meaning is the combinatorial result of
the senses and the syntactic relations we call it ‘fully compositional’.
Examples are:
(10) That girl is young (basic senses only)
(11) That girl is a nasty cat (basic and derived senses)
In example (11) the derived sense of cat (‘nasty woman’) is selected.
The meaning of a standard m etaphorical construction originates from a
combination of at least one lexical item with an extended sense and one or
m ore lexical items with basic, derived or extended senses, and the
syntactic structure of the constituent which contains these lexical items -
exactly the same goes for metonymous and synecdochical constructions .13
As is the case with standard literal constructions, metaphorical ones are
fully compositional because their meanings are the combinatorial results
of the senses of their lexical items and the syntactic structure of the con
stituents which contain them. However, as three types of sense can be
combined instead of two, the compositionality of metaphorical construc
tions is more complicated than that of literal ones. A n example of a
standard metaphorical construction is:
(12) These are early days for M r Major, and some argue that his decision
to put his most obvious potential successor, Kenneth Clarke, into
the H om e Office, that graveyard of political careers, shows he is not
wholly self-confident. ( The E conom ist, 1/8/92).
Some literal lexicalised expressions are fully compositional too. Some
examples are; Ladies and Gentlemen, if I m ay say so, putting it bluntly and
having said that . As a consequence we only find descriptions of the way in
which they are used (not of their meanings) in dictionaries such as
L D O E L C or handbooks of grammar such as Quirk et a l (1985) (hence
forth: CGEL), e.g.:
(13) Ladies and Gentlemen (‘USAG E: when speaking to a gathering of
people’)
(14) if I may say so (style disjunct; see C G E L §8.124)
13 For reasons of conciseness we use the term ‘metaphorical’ in contexts where it can be
substituted with ‘metonymous’ and ‘synecdochical’.
© T h e Editorial B oard o f Studia Linguistica 1996,

The meanings of some literal lexicalised expressions encom pass m ore

than the combinatorial results of the senses in their lexical items and their
syntactic structures. In other words: the basic senses of the lexical items in
these expressions play a role in, but form only parts of, their meanings.
Take for example:
(15) bed and. breakfast (‘a system of accommodation in a hotel o r guest
house in which you pay for a room for the night and breakfast the
following morning1, COBUILD)
Both bed and breakfast have their basic senses in this expression. H ow
ever, these contribute only partly to the meaning of the expression, and
other meaning properties such as system o f accom m odation , p a y , r o o m ,
hotel or guest house and the follow in g m orning cannot be inferred from
them.14 The same goes for e.g.:
(16) X ime reference Y after A7*'”*refere,,ve Y (‘continuously’: e.g. day after day)
We call such expressions ‘pseudo-compositional’. Some literal lexicalised
expressions have pseudo-compositional meanings because the referents
of their lexical items form a unit, as in:
(17) cup and saucer
(18) man and wife
Example 18 is pseudo-compositional for another reason: in the expres
sion man means h u sb a n d .
There is an im portant distinction between standard metaphorical con
structions and ones that have become lexicalised, such as the following:
(19) p u t the cat am on g the pigeons
(20) a thorn in one's side
Standard metaphorical constructions have meanings which arise from
extended senses (possibly combined with basic a n d /o r derived senses).
Rules for sense adaptation are activated as a result of violation of
restriction rules, Lexicalised expressions which were originally standard
metaphorical constructions, on the other hand, have established meanings
that are not inferred from the senses of their lexical items. A s opposed to
many lexicalised expressions which were originally standard literal
constructions, they do not have lexical items with basic senses that form
(part of) their meanings. Therefore they are ‘fully non-com positional 1.15
We use the terms idiom atic’ and ‘idiom’ to refer to lexicalised expressions
which are non-com positional
Many idiomatic expressions have equivalents in the form of 'counter
feit form’. Such a form has the same syntactic form and contains the same
14 See Makkai (1972) for the term ‘idiom of encoding’. See Mel’cuk (1995) for the term
kquasi-idiom\
15 Makkai (1972) uses the term ‘idiom of decoding’ for such expressions.
©The Editorial B oard of Sludia 1-inguisticu 1996.

140 HenkBarkema
lexical expression, but, because of the way in which it is used, has a

meaning that is the combinatorial result of the meanings of the lexical
items in the construction. A n example of a counterfeit form is:
(2 1 ) the helicopter circled round the tip o f the iceberg
Some lexicalised expressions are partly idiom atic (i.e. partly non -
com positional ), while they contain at least one lexical item with a basic
sense. In the following examples the idiomatic parts are underscored:
( 22 ) a blind alley (‘a little narrow street with no way out at the other end’)
(23) rain cats and dogs (‘to rain very heavily5)
Examples 22 and 23 are not pseudo-compositional because the basic or
derived senses of only some lexical items in the expressions contribute to
their meanings .16 In pseudo-compositional expressions all basic or
derived senses play a role in their meanings. W e call examples such as 19
and 20 , which only contain lexical items without basic or derived senses,
entirely idiom atic .
In many lexicalised expressions the senses are ‘dorm ant’ instead of
absent; in other words: the (basic) senses of idioms can be re-activated in
certain contexts and by means of certain syntactic variations. A n example
is:
(24) H e and the equally shark-like Gardini p u lled what lo o k ed like badly
burned chestnuts out o f the N e w Z ealan d fire in the final of the Louis
Vuitton Cup for the challengers. (T h e Independent, Sport, 9/5/92)
In section 2.1 we already referred to this notion with the term ‘motivation’.
M ost (if not all) speakers of English are able to re-establish the relations
between the institutionalised meanings of the idiomatic expressions to
play second fiddle (‘to play a less im portant part’), to keep the ball rolling
(‘to continue something, such as a conversation or a plan’) and to clip
som eone's wings (‘to prevent someone from being as active or powerful as
before’) and the compositional meanings of their (literal) counterfeit
forms. The only requirement is the creation (or the imagination) of a
context which corresponds with the institutionalised meaning of the
idiom. In Cowie et al.’s (1983:xiii) classification such idioms are therefore
labelled as ‘figurative idioms’. The same is m ore difficult with expressions
like eat humble p ie (‘to have to admit that one was wrong or that one has
failed5), by an d large (‘on the whole, in general’) and red herring (‘a fact or
subject which is introduced to draw people’s attention away from the main
point’). For this reason Cowie et al. (1983:xii) label such expressions as
‘pure idioms’).
Some lexicalised expressions are idiomatic because they contain one or
m ore lexical items which have no (basic, derived or extended) senses.
16 Example 22 is polysemous; its second sense is entirely idiomatic (see §4, examples 77a
and 77b).
© T he Editorial B oard o f Studia Linguistica 1996.

These ‘absent sense’ lexical items only occur as part of lexicalised

expressions. Some examples are: spick an d span , in lieu of, the whole
caboodle , on beh a lf of, like wildfire , in fine/good fettle, a t loggerheads w ith ,
kith and kin and on tenterhooks. For instance, in L D O E L C no separate
senses are provided for the underscored lexical items. A s each of these
expressions contain at least one lexical item without a (basic) sense we call
them ‘absent sense’ expressions. For such expressions speakers require
special (etymological) knowledge to be able to recover the original basic
senses. A n example of an expression which was originally com positional is
by and large which, according to Simpson & W einer (1989) means ‘to the
wind (within six points) and off it*. The ability of speakers to re-establish
relations between idiosyncratic meanings of idiomatic expressions and the
meanings of their counterfeit forms on the one hand, and to recover the
basic senses in 'absent sense’ expressions on the other, depends on their
knowledge, but also on their imagination! Therefore it is impossible to give
a linguistic description of recoverability in terms of a relation between
senses and meanings.
It is possible that there is a consensus among speakers about the recover
ability of meanings of idiomatic expressions. In addition, it is possible that
the recoverability of the meaning of an expression determines the degree of
its flexibility. W hether o r not this is true should be established through
systematic examination of the knowledge and insights of a large num ber of
speakers. As this leads us from the field of linguistics into that of psycho
linguistics, however, we do not further elaborate on this topic, F or an
example of psycholinguistic research into the recoverability of the m ean
ings of lexicalised expressions, see Van de V oort & Vonk (1.995).
3.3. Flexibility
Lexicalised expressions minimally require the presence of specific lexical
items and a specific syntactic structure for their meanings a n d /o r
pragmatic functions. F or example, neither you r uncle nor a n d B o b have a
sufficient num ber of items to carry the idiosyncratic meaning of an d B ob *s
your uncle . Therefore a n d B o b ’s yo u r uncle is the base fo rm of the expres
sion.17 The base form is the simplest morpho-syntactic form that the
expression can take. Dictionaries such as COBUILD (Sinclair 1987) and
LDOELC (Summers 1992) list base forms as main or sub-entries;
examples are:
(25) rack and ruin (‘a ruined state, esp. of a building, caused by lack of
care’)
(26) and all that ja z z (‘and other things like that’)
17 See Jespersen (1924), Healey (1968: 72) and Tovena (1992).
©The Editorial B oard o f SluUia U nguistf ca 1 9 % .

142 Henk Barkema
Of the majority of base forms, all the grammatical function slots are filled,
for example:
(27) tip o f the iceberg
(28) sitting du ck
A nother set consists of base forms which have open function slots.
Examples are:
(29) what the X wun (‘used with various words, such as hell, d e vil , blazes ,
etc. when asking angry or surprised questions: what’)
(30) X ime reiercnce Y after X imc ,efemice Y (‘continuously’; e.g. day after day)
Everaert (1993) uses the term ‘idiomatic pattern’, Fillmore et al. (1988)
the term ‘formal idiom’. Pawley & Syder (1983:211) give examples of
‘sentence stems 1 with open slots such as N P be-T E N SE sorry to keep -
TENSE you waiting and Who ( the E X P L E T IV E ) do-P R E S N P think Pro1
be-PRES. Nattinger (1980) gives examples such as D ow n with . . . The
term we use for such expressions is ‘template f o r m \1K
The base pattern of a lexicalised expression is the morpho-syntactic
structure of its base form, expressed in terms of syntactic functions,
categories and features,lt; For instance, sitting du ck has the following base
pattern:
(31) D E T E R M IN E R + P R E M O D IF IE R (non-gradable adjective) +
H EA D (singular count noun)
Because the base pattern is just the morpho-syntactic structure of an
expression, several different bas e fo rm s can have the same base pattern ,
e.g.:
(32) a sitting duck
(33) a standing jo k e (‘a joke or humorous subject that is so well-known to
a group of people that they laugh whenever it is mentioned’)
(34) a shooting star (‘a small piece of material from space which burns
brightly as it passes through the E a rth ’s air’)
We use the term "variant form ’ to refer to a form of a lexicalised expression
which is not the base form. To be able to refer to a single morphological or
syntactic alteration on the base form we use the term ‘variation’. Thus the
following variant form (in italics) contains only one variation (under
scored).
(35) I f you asked the man in the m odern street for his opinion of
homosexuality, he would probably reply, ‘I’ve nothing against queers
. . (Birmingham Collection of English Texts; see Renouf 1984)
IK We do not regard base forms with open initial determiner slots as template forms (e.g.
DETERMINER lost cause).
See Aarts & Aarts (1982) for a descriptive model of functions and categories.
© The Editorial Board of Stud in Linguistics 1 996.

And the following two variant forms (in italics) contain several variations
(underscored ):20
(36) In frank terms, what has been revealed about the current scandals
involving Kyowa and Sagawa are only the tips of tw o o f m an y icebergs
floating in Japanese political waters.
(37 ) . . . a deal to supply five aircraft to Aeroflot as a hard-fought fo o t in
the d o o r o f one o f the aviation world's least exploited m a rk ets ,
Some expressions have m ore than one institutionalised form: one base
form and one or m ore forms which are longer. These larger forms we call
‘lexicalised variant form s’. In dictionaries the parts of these forms that are
not required for the base forms are usually put between brackets. Two
examples are:
(38) a bird in the hand ( is worth two in the bush)
(39) a hair o f the dog (that bit you)\ (‘an alcoholic drink taken in the
morning because it is said to cure illness caused by drinking too
much alcohol the night before')
We distinguish between four types of variation: perm utation, term
selection, addition and interruption. ‘Perm utation’ implies that elements
that are present in the base form are moved in the construction; their
functions remain m ore or less the same / 1 Examples are:
(40) heir presum ptive vs, presum ptive heir
(41) the sum. total vs. the total sum
' ^ 4 m mk *< » < >>a*»»1*»»
In cases like these it is difficult to decide which form is the base form, In
relation to the flexibility of these expressions it is more im portant which
variation is possible (in this case permutation) than what the direction of
this variation is (from pre- to postmodifier or vice versa).
Term selection* implies that an element from a closed system is
replaced with an element from the same system. Examples are:
(42) sitting du ck vs. sitting clucks
(43) the straw that breaks the earners back vs. the straw that bro k e the
camel's back
(44) do som ething by fits an d starts vs. do something in fits a n d starts
(45) to be at a crossroads vs. to be at the crossroads
‘Addition’ is the introduction of modifying or quantifying elements or of
elements that do not interrupt the syntactic structure of the construction :22
2,1 These examples are from The Guardian 1992.

21 An example is modification. For instance, a premodifying adjective is used instead of a
postmodifying adjective.
22 See Nicolas (1992) for a semantic classification of additional modifiers in lexicalised
expressions.
©TheEdiioriul Hoard o f Studin Unguisticu I W 6 .

144 Henk Barkema
(46) a ruling passion vs. a ruling passion to discover new facts

(4 7 ) a flash in the pan vs. a flash in the econ om ic pan
(48) the tip o f the iceberg vs. the tip as well as the rest o f the iceberg
The fourth type of variation is ‘interruption’. It is the insertion of a foreign
element (that is not an immediate constituent) into the base form of a
lexicalised expression:
(49) (It was) the talk o f the town vs. (It was) the ta lk , yesterday , o f the town
(interruption by adverbial)
We use the term ‘flexibility1 to refer to the extent to which constructions
can take different grammatically possible forms. ‘F ree’ constructions can
take all forms that are grammatically possible. A n example is:
(50) a book and a pen
This construction can take forms like the following:
(51) a pen and a hook (permutation of NPs)
(52) books and pens (term selections: plural nouns instead of singular
ones)
(53) an expensive book and a cheap pen (addition of premodifiers)
(54) (They bought) a b o o k yesterday , a n d a pen too (interruption by an
adverbial)
It is evident that one would need several pages to discuss all grammatically
possible forms. However, there is no reason to repeat here what various
handbooks of grammar have described quite thoroughly already.
The most important characteristic of lexicalised expressions is that they
cannot take all forms that are grammatically possible. To put it differently:
when they are subjected to free grammatical variation, this variation will
yield several pseudo-variants, i.e.: forms that can only be counterfeit
forms .23 Some lexicalised expressions have pseudo-variants only. Take for
example the base form of the lexicalised expression man and boy .
Grammatically possible pseudo-variants are:
(5 5) # men and boy (term selection: plural form)
(56) # boy an d man (permutation: conjoins)
(57) # friendly man an d naughty boy (addition: two premodifiers)
(58) # the m an and the boy (addition: two central determiners)
O n the other hand there are lexicalized expressions that have a limited
num ber of variant forms, for instance: a pain in the n eck . Some examples
of variant forms are:
(5 9) a great pain in the neck
(6 0) a pain in the neck o f the trade unions
2,1 In Barkema (1994a) we used the terms ‘unrelated variant’ for pseudo-variant and
‘related variant' for variant form.

Examples of pseudo-variants are:

(61) # . . . a p a in , yesterday, in the neck
(62) # a p a in in a neck
The various examples we have given in this section show that there are
three types of construction: ones that have base and variant forms, ones
that have base forms as well as variant forms and pseudo-variants, and
ones with only base forms and pseudo-variants. W e use the term fu lly
flexible , semi-flexible and inflexible constructions respectively to refer to
these three types. We only discuss the internal flexibility characteristics of
constructions and not external ones such as the functions that construc
tions can take.
In some cases nominal heads which form part of lexicalised expressions
can be premodified by adjectives such as proverbial and o ld (see e.g,
Nicolas 1992). However, the addition of such elements has nothing to do
with flexibility. It can be seen as a form of self-editing that has no effect on
the established meaning at all, but only stresses the fact that the expression
is lexicalised. A n example is:
(63) an edifice of decisions as fragile as the proverbial card house
(Birmingham Collection of English Texts).
3.4. Collocability
We use the term 'collocability’ to refer to the degree to which it is possible
to substitute a lexical item from an open class in a construction with
alternatives from the same class: thus a noun is substituted by other nouns,
a verb by other verbs, etc .24 These alternatives can be synonyms, near
synonyms and antonyms. If the collocability of a construction is restricted,
this is a m atter of arbitrariness, i.e. the restriction is not linguistically
motivated. In other words: a combination of lexical items in a construction
may be in accordance with the morphological, syntactic and semantic
rules of the language and at the same time be unacceptable as a form of a
lexicalised expression. F o r instance, the su m m it o f the iceberg is perfectly
grammatical, but it is a counterfeit form of the tip o f the iceberg (‘the small
part of a problem that is obvious, when the problem is much more serious
and widespread than it seems to be\ COBUILD, Sinclair 1987).
The greater the difference in size is between the set of theoretically
possible alternatives and the subset of acceptable ones, the m ore the
collocability of a construction is limited.
In ‘free’ constructions all lexical items from open classes can be substi
tuted by any num ber of alternatives from the same classes. This num ber
24 See also Carter (1987), chaptcr II. Firth (1951:123) already used the term
‘collocability* more than forty years ago.
CThe Editorial Board of Studtu Linguistica l ‘J % .

146 Henk B arkema
can be extremely large, depending on the ‘ranges’ of the lexical items - see
Carter (1987, chapter III) or Cowie (1981:226). Take for example:
(64) H e observes the wom an
Both items from open classes - observes and w om an - can be substituted
by any number of alternatives from the classes of verbs and nouns. For
example, observe can be replaced with synonyms such as notice or look at ,
while wom an can be replaced with ones like lady or missus. The same
goes for the template form X ime n’lhvncc y after X inw rcfcrence where time
references like the following can be used: d a y , week, m inute and yea r . We
label such constructions as ‘collocationally open 4.25
On the other hand, the lexical items from open classes in the majority of
lexicalised expressions cannot be replaced with alternatives from the same
class. Take for instance:
(65) by and large (‘on the whole, in general1); (not e.g.: # by a n d big, and
# near and large)
(66 ) red tape (‘silly detailed unnecessary official rules that delay action’);
(not: # green tape, red ribbon , etc.)
The alternatives in examples 65 and 66 (preceded by a # ) we call ‘pseudo-
alternants’: with these alternative synonyms, near-synonyms or antonyms
they can no longer be regarded as forms of lexicalised expressions but
should be labelled as ‘standard’ constructions (or in the case of by and big
as ungrammatical) instead. We call constructions in which none of the
lexical items from open classes can be replaced ‘collocationally closed’.
In some lexicalised expressions one or m ore open-class items can be
replaced with (nearly) synonymous or antonymous items from the same
class. Some examples are:
(67) shanks's mare/pony, (‘walking’)
( 68 ) generally/strictly speaking
In example 67 shanks's mare and shanks's p o n y are alternants, while
shanks's horse is a pseudo-alternant. In the same way generally speaking
and strictly speaking in example 68 are alternants, while elaborately
speaking is a pseudo-alternant.
If a lexical item from an open class in a lexicalised expression is replaced
by an antonym, the resulting expression is generally an antonym of the
original expression too, e.g. a g o o d sailor (‘one who enjoys travel by water
without being sick’) versus a b ad sailor .
The number of alternatives in expressions like 67 and 68 is limited in an
arbitrary way to a few or to only one. The alternative lexical items in these
expressions are subject to the same selection restriction rules that apply to
the ones they substitute for. We label such expressions as ‘collocationally
25
E.g. Carter (1987:63) uses the term ‘unrestricted collocation’.
© T he Editorial Board of Studia Linguistica 1996.

limited’ and such replacements by (near-)synonyms or antonyms ‘regular

substitutions ’.26 There are exceptions: in a small num ber of expressions
one or more lexical items from open classes can be replaced by ones from
the same class, but these alternatives are no (near-) synonyms or antonym s.
The semantic relations between the lexical items are arbitrary. F o r
example:
(69) m on key business/tricks
We call such expressions collocationally limited as well, and call such

replacements irregular substitutions'.
Different open-class items in lexicalised expressions can have different
degrees of collocability. For the sake of completeness, we use the terms
‘partly’ and ‘entirely’ in combination with collocability in the same way as
we use them in relation to idiomaticity. For instance, the template form
gitnc tvfirentr y ap e r jpum* reference v js eollocationally ‘entirely open’, because
both lexical items from open classes can be freely substituted, while
shanks's m a r e /p o n y is both ‘partly limited’ ( mare can be replaced with
pony ) and ‘partly closed’ (shanks's cannot be substituted).
In a small num ber of cases it is possible to replace a whole expression by
a synonymous one in a num ber of steps. An example is: from near miss to
narrow shave: near m iss ~+near thing-*close thing-*close shave -»narrow
shave. The same goes for o ld stager and g o o d h a n d : o ld stager-* o ld hand.
-+good h a n d . The insoluble question in relation to dominoes like this is:
when are two forms one and the same lexicalised expressions, and when
are they two o r more expressions which are synonymous?
With the examples given in this section so far, we do not want to suggest
that all tton-lexicalised contructions are collocationally open. In section
3.1 we gave the following example of the restricted construction:
(70) to work/operate/run a m achine (not e.g.: *to manage a m a ch in e , *to

conduct a m achin e )
Another example of such a collocationally limited construction is:
(71) to place a bet (not e.g.: * to ¡mt a b e t , *to set a b e t , *to station a b e t , *to
pose a b e t )
Carter (1987, chapter II) refers to Firth’s (1957) examples ‘strong tea’ vs.
‘powerful tea’(see also Greenbaum 1970:9). Such restricted constructions
do not require special attention in dictionaries if their meanings are
completely inferred from the basic and/or derived senses of the lexical
items and the syntactic relations in these constructions and if their
pragmatic functions can be determined on the basis of the context in which
2(1 See Carter (1987:63) for the term ‘semi-restricted collocation1. His class of ‘familiar
collocation' seems to be more or less similar to that o f‘semi-restricted' ones.
©The Editorial Board o f Stud ia Linguistic» I yyfi.

148 Henk Barkema
they are used. However, as they are subject to collocational limitation,

they are discussed in combinatory dictionaries such as BBI (Benson et al.
1986).
With a construction such as crashing b o re , the inevitable question is: is it
a lexicalised expression? According to L D O E L C , crashing has a derived
sense, Very greatV'complete’; on the other hand, however, it only has this
meaning in combination with bore and f o o l , and not in e.g. * crashing liar,
*a crashing idiot , * crashing n in com poop. In our opinion crashing bore is a
partly idiomatic lexicalised expression.
A particular type of substitution can be illustrated with the following
examples:
(72) But since the p r o o f o f the beer is in the drinking, we suggest that you
try a can or two of authentic Draught Bass in the comfort of your
home (advertisement in The Independent, 2 3 /5 /1 9 9 2 )
(73) They therefore feel they are entitled to sing, throw plates and trip the
light fandango during your act if they so choose. (Dillie Keane,
Punch, 25/2/1992).
(74) A chip off the o ld shock (heading in The Independent, 9/5/1992;
reference to Robert Hughes’ The Shock o f the New)
Although the lexicalised expression the p r o o f o f the pu ddin g is in the
eatings trip the light fantastic and a chip o ff the olcl block are collocation
ally closed, the underscored parts in the examples above show that under
special circumstances a particular type of replacement is even possible in
closed expressions .27 The rules for this type of replacem ent seem to be the
following: the form is not an institutionalized alternant, and the basic
sense(s) of the alternative lexical item(s) correspond(s) with the meaning
of the context outside the expression. The substitution is either regular
(beer/pudding; eating/drinking) or irregular. W hen it is irregular, the
alternative is similar to the lexical item it replaces in another way, for
example it rhymes (e.g. block/sh ock); it is also possible that the two lexical
items have the same morpheme(s) (e.g. fantastic/fandango). It does not
seem to be essential that the alternative lexical item belongs to the same
class as the one it substitutes for, as one of the examples shows: in trip the
light fantastic/fandango an adjective is replaced with a noun. We call this
type of substitution ‘punning’ and a form of a lexicalised expression
containing such a pun: ‘a pun alternant’.
The effect punning has on the meaning of a lexicalised expression is
similar to that of a variation, as in both cases the meaning is modified.
However, as one lexical item is replaced by another, the extent to which
pun alternants of lexicalised expressions are possible is not indicative of
their flexibility. As the selection of alternatives cannot be regarded as an
established one, but strongly depends on the context in which the expres-
27 See also Fernando & Flavell (1981:38).
© T h e Editorial Board of Sludia Linguistic« 1996.

sion is used, the extent to which pun alternants are possible is not indica
tive of its coilocability either.
3.5. Other types o f characteristics

So far we have described the three main linguistic types of characteristics
that all constructions have: compositionality, coilocability and flexibility.
In addition, subsets of lexicalised expressions have other characteristics.
Although most lexicalised expressions have base patterns similar to
those of ‘standard’ constructions, a relatively small set have idiosyncratic
syntactic structures and therefore do not have counterfeit forms. We call
these ‘idiosyntactic constructions’. Some examples are: no can d o , long
time no see , m ake d o , com e to think o f it, trip the light fantastic, by a n d
large, gin a n d it and easy does i t p Cruse (1986:38) uses the term
‘asyntactic idiom ’ for such expressions. The disadvantage of this term is
that it suggests that they do not have any syntactic characteristics.
However, an expression like by and large can still be regarded as a form of
coordination, be it an idiosyncratic one.
Some expressions are given special attention in dictionaries and hand
books of grammar because they have specific pragmatic characteristics,
e.g.: Ladies a n d G entlem en , G o o d M orning , H o w d o you d o ? and H appy
Birthday ƒ29 Fillmore el al. (1988) use the term ‘idiom with pragmatic
point’, while Altenberg & Eeg-OIofsson (1990) label them as ‘interactive
expressions’. W e call such constructions with institutionalised pragmatic
functions ‘formulaic expressions’. On the other hand, examples of
pragmatically fixed expressions which are not (consciously) interactive,
but expressive are: G o d knows, What a perform ance , G o o d L o r d and
poor thing (see e.g. C G EL, §11.53). W e call these ‘exclamatory expres
sions’.
Although most lexicalised expressions are monosemous, some are
polysemous, for example blind alley: ( 1 ) ‘a little narrow street with no way
out at the other end1; ( 2 ) i f a method of working or thinking is leading you
up a blind alley o r down a blind alley , it is turning out to be useless or not
leading to good results’ (COBUILD).
Finally, a num ber of lexicalised expressions have specific semantic
characteristics; an example is lift a finger , which can only be used in
negative contexts. For a description of such ‘negative polarity items1: see
Tovena (1992). Louw (1993) uses the term ‘semantic prosody' for the
semantic context in which a word or construction is generally used.
2K See Fillmore et al, (1988) for the term ‘extra-grammatical idiom’. Carter (1987, chapter
III) refers to irregular syntactic structures such as to go one better , to h o ld true and to go it
alone.
2y See Quirk et al. (1985), §§11.38,3 1.54 and 19,48.
©The Editorial B o a rd o f Studitt Linguistic» 1996.

150 Henk Barkema
4. Examples of multi-dimensional classifications

As we said in section 1, die differences between the multi-dimensional
model introduced in this article and the traditional one-dimensional
classifications are explicitness and precision, uniformity in description,
and the possibility to describe both types and tokens. To further illustrate
the difference between the two classificatory methods, we discuss five
expressions (i.e. types) and one example from a corpus (i.e. a token). In
addition, we show how the expressions are treated in the following
dictionaries:
COBUILD: see Sinclair (1987);

LDOELC: see Summers (1992);
ODOCIE: see Cowie et al. (1983);
O EED : see Hawkins and Allen (1991).
( 7 5 ) the bottom /first/highest rung o f the ladder (‘the lowest/etc. level in

the organization’):
Multi-dimensional classification:
— entirely idiomatic;
— semi-flexible (e.g. some variant forms are possible, e.g.... political ladder);
— collocationally partly open (because of bottom/first/highest / p a r t l y
closed (because of rung and ladder),
One-dimensional classifications:
non-compositional metaphor (Wood 1986);

idiom (v/d Linden 1992);
figurative idiom (Cowie j 983).
W ood’s definition of the term ‘idiom 1 does not allow its application to this
expression, because it is not collocationally closed (see section 2.1). The
fact that the expression is lexicalised is implied by the various traditional
terms. None of these terms show whether the expression is semi-flexible or
inflexible.
The dictionaries provide the following information: in COBUILD the
expression is not listed as such; rung has a derived sense and the
expression is used in an example. This gives us the wrong impression that
the expression is a ‘standard’ construction in which rung has a derived
sense and ladder an extended sense. In L D O E L C the expression is a
subentry of rung , which makes sense. In both O E E D and O D O C IE the
expression is absent.

(76) If 1 m ay say so:

—fully compositional;
—inflexible;
—collocationally closed;
formulaic
Fully compositional lexicalised expressions as example 76 are generally
ignored in the literature on idioms, but they are discussed in handbooks of
grammar because they are formulaic. In C G E L (§8.124) the expression is
labelled as ‘style disjunct1, The fact that it is lexicalised, inflexible and
collocationaly closed is implied by the handbook, not described. A lthough
example 76 is a lexicalised expression, none of the dictionaries pays any
attention to it.
(7 7 a) blind, alley ( 1 ) (la little narrow street with no way out at the other
end1):
NB: this expression is polysemous: its second sense is discussed in
example 77b.
— partly idiomatic: because of blind;
—• semi-flexible (e.g. plural is possible);
— collocationally closed;
polysemous.
semi-idiom (Cowie et al. 1983);
collocation (Wood 1986);
idiomatic expression (v/d Linden 1992).
Both Cowie et al. (1983) and van der Linden (1992) indicate with their
terminology that blind alley is a partly idiomatic lexicalised expression,
while W ood (1986) shows that it is collocationally closed. The fact that it
is semi-flexible into the bargain, however, is not expressed in their
terminology.
In COBUILD the sense of example 77a is absent; only its second sense
is described (see example 77b). In L D O E L C and O D O C IE the expres
sion is a main entry. In O E E D it is a subentry of blin d .
(77b) blind alley ( 2 ) (‘if a method of working or thinking is leading you up
a blind alley or dow n a blind alley, it is turning out to be useless or
not leading to good results’; COBUILD):
©Tho Editorial B oard o f Studio U nguislica 1990.

152 Henk Barkema
— entirely idiomatic;
— semi-flexible (plural is possible);
polysemous.
figurative idiom (Cowie et al. 1983);
idiom (Wood 1986);
idiom (v/d Linden 1992).
All terms indicate that the expression is idiomatic. In addition, Wood’s

(1986) label implies that the expression is collocationally closed. None of
the labels indicate that it is semi-flexible.
In COBUILD the expression is listed as a main entry; the dictionary
indicates that it is semi-flexible and that the noun is countable. LDOELC
only indicates that blind alley ( 1 ) can be used figuratively, thereby imply
ing blind alley (2) is a "standard’ construction. O E E D lists the expression
as a subentry of blin d , O D O C IE as a main entry.
(78) bed and breakfast (v(a private house or small hotel that provides) a
place to sleep for the night and breakfast the next morning for a fixed
price’):
— pseudo-compositional;
— inflexible;
— collocationally partly limited (because of breakfast/board), partly closed
(because of bed ).
polysemous.
binomial (Quirk et al. 1985);
collocation (Wood 1986);
collocation (Cowie et al. 1983).
N one of these classifications shows that the expression is lexicalised,
pseudo-compositional and inflexible. The fact that it is collocationally
limited is expressed by the classifications of W ood (1986) and Cowie et al.
(1983), and implied by Quirk et al.’s (1985).
COBUILD gives the wrong impression that the expression (main entry)
is monosemous. It is correct in indicating that it is semi-flexible and that
the nouns are uncountable. L D O E L C regards the expression as
polysemous and provides two meanings. In O D O C IE the expression is
absent, which makes sense in view of the fact that the dictionary focuses on
© The Editorial B oard of Siudia Linguistic« 1996,
idiomatic’ English, O E E D , finally, has entered the expression as a

subentry of b e d . It is polysemous with three meanings; the first two
meanings have to do with a night’s lodging o r the establishment w here one
can lodge, while the third is: ‘an operation in which a shareholder sells a
holding one evening while agreeing to buy it back again next morning,
realizing either a gain or a loss in order to suit a tax requirem ent’.
(79) W hat/ W h y/ Who the X I0(in (‘used with various words, such as h ell ,
devil, blazes , etc. when asking angry or surprised questions’);
— partly idiomatic (because of the X>,(>nn)\
— inflexible;
— collocationally open (because of X ,HW W). NB: the other items are from closed
classes;
idiosyntactic;
template form;
exclamatory.
colligation (Wood 1986);
exclamatory phrase (see CGEL, §11.53); emphasizer (CGEL, §8.100); also
given as an example of a postmodifier of wh- words (CGEL, §7.64);
expletive (CGEL, §18.59n).
Wood’s (1986) classification indicates that the expression is a template

form, but does not express that it is partly idiomatic, inflexible,
collocationally open and idiosyncratic. Because it is a collocationally open
template form, it is typically an expression that should receive extensive
attention in handbooks of grammar. Unfortunately, C G E L only pays
attention to the exclamatory nature of the expression, while it ignores all
other (idiosyncratic) characteristics.
For dictionaries the expression is problematic because it is
collocationally open, so that none of the lexical items can be chosen as
‘anchor’. In C O B U IL D the expression is listed as a subentry of hell ( why
the hell , who the hell , w hat the hell), of blazes {the blazes) and of dickens
(the dickens). T he dictionary indicates that the expression is inflexible. In
LDOELC it is listed as a subentry of what, but not of who or w h y , etc.
OEED lists it as a main entry (w h a tth e dickens) and as subentries of blaze
(what the blazes ), hell (w h a tth e hell) and devil (the devil), In O D O C IE the
expression is a main entry ( what the h e ll/d e v il/b la ze s/h e c k /d ic k e n s),
while it is indicated that what can be replaced by why , etc.
So far, we have given examples of the characteristics of expressions, i.e. of

types. On the basis of one example we show that our descriptive model can
also be used for discussions of the characteristics of tokens:
0 The Editorial B oard of Siudta Linguistic» 1 9 % .

154 Henk Barkema
(80) A nd actually researching the current situation with floating

exchange rates with all the differences that have cropped up over the
past few years, there’s very little written about it (Birmingham
Collection of English Texts, spoken part).
Type: rate o f exchange
— pseudo-compositional;
— semi-flexible (e.g. rate can be pluralised).
Token: floating exchange rates
base form: rate of exchange;
vari an t fo rm: floa ting exch ange rates :
three variations: — one permutation: premodifier exchange vs. postmodifier of
exchange;
— one addition; premodifier floating ;
one term selection: pluralised head of the noun phrase
(rates).
5. Conclusion
In the introduction to this article we discussed the disadvantages of the
traditional terminology, which only highlights one striking feature of a
lexicalised expression, and leaves various others implicit. We concluded
that for detailed examination and discussion of the various linguistic
dimensions of lexicalised expressions a well-defined model is required
that distinguishes between various descriptive dimensions and at the same
time pays heed to the scalar nature of the different types of characteristics.
In the remainder of the article we described a multi-dimensional model
with which lexicalised expressions can be given as many labels as are
required for exhaustive descriptions of their characteristics. In section 4
we illustrated the descriptive power of the model with multi-dimensional
classifications of five types and one token.
The model can be used for classifications and descriptions in the
literature, in handbooks of grammar and in dictionaries. Finally, it can be
used for the annotation of the characteristics of lexicalised expressions in
text corpora and for corpus-linguistic research into the relations between
these characteristics. It was for this very purpose that the model was
designed (see Barkema 1 9 9 3 ,1994a, 1994b, 1996).
References
A arts, F. &A a r t s , J., 1982. English syntactic structures. Functions and categories
in sentence analysis. Oxford: Pergamon.
A a r t s , J. & C a l b e r t , J. 1979. Metaphor and non-metaphor The semantics of
adjective-noun combinations. Tübingen: Max Niemeyer.
© T h e Editorial B oard of Studia Linguisticn 1996.
A arts, J. & M e i j s , W. (eds.). 1984. Corpus linguistics I. Recent developments in

the use of computer corpora in English language research. Amsterdam, Rodopi.
A a r t s , J. &M e i j s , W. (eds.) 1986, Corpus linguistics 1I. New studies in the analysis
and exploitation of computer corpora. Amsterdam: Rodopi.
A a r t s , J. 1991. Intuition-based and observation-based grammars. English corpus
linguistics, eds. K. Aijmer & B. Altenberg, 44-63. London: Longman.
A a r t s , J. & M e i j s , W. (eds.). .1990, Theory and practice in corpus linguistics.
Amsterdam: Rodopi.
A a r t s J., d e H a a n , P, & O o s t d i j k , N. (eds.), 1993. English language corpora:
design, analysis and exploitation , Amsterdam: Rodopi.
Aijmer, K. & A men berg, B. (eds.). 1991. English corpus linguistics, London:
Longman.
Aisenstadt, E. 1979. Collocability restrictions in dictionaries. Dictionaries and
their users, ed. Hartmann, R. 71-4. Exeter.
Aisenstadt, E. 1981. Restricted collocations in lexicology and lexicography. ITL.
Review of applied linguistics, 53, 53-61.
A l e x a n d e r , R. 1987. Problems in understanding and teaching idiomaticity in
English. Anglistik & Englischiinterricht, 32. Wortschatzarheit. Heidelberg:
Carl Winter.
A l t e n b e r g , B . & E e g - O l o f s s o n , M. 1990. Phraseology in spoken English:
presentation of a project. English corpus linguistics, ed. J. Aarts & W. Meijs.
Amsterdam, Rodopi.
Altenberg, B. 1993. Recurrent verb-complement constructions in the London-
Lund corpus. English language corpora: design, analysis, and exploitation , ed. J.
Aarts & W. Meijs.
Backlund, U. 1976» Frozen adjective-noun collocations in English. Cahiers de
lexicologie, 2 8 ,7 4 -8 8 .
Barkhma, H. 1993. Idiomaticity in English NPs. English language corpora: design,
analysist and exploitation ed. Aarts et al. 257-78.
Barkema, I-I. 1994a. Determining the syntactic flexibility of idioms. Creating and
using English language corpora ed. U. Fries, G. Tottie & P. Schneider, 39-52.
Barkema, H. 1.994b. The idiomatic, syntactic and collocational characteristics of
received NPs: some basic statistics. Hermes journal of linguistics, 13:19-40.
Arhus: Arhus School of Business.
Barkema, FL 1996. The effect of inherent and contextual factors on the grammati
cal flexibility o f idioms. Synchronic corpus linguistics. Papers from the sixteenth
international conference on English language research on computerized corpora ,
Toronto, 1995 ed. L Lancashire, C. Meyer & C. Percy, 217-31. Amsterdam:
Rodopi.
B e n s o n , M., B e n s o n , E. & I l s o n , R. 1986. The BBl combinatory dictionary of
English. Amsterdam: John Benjamins.
B u r g e s s , A, 1969. Language made plain, The English Language ed. W. Boulton &
D. Crystal, Vol. II. London: Cambridge U. Press.
C a r t e r , R. 1987. Vocabulary: applied linguistic perspectives. Aspects of English
ed. J. Sinclair, London: Allen & Unwin.
C h a f e , W. 1970. Meaning and the Structure of Language. Chicago: U, of Chicago
Press.
C o w i e , A, 1981. The treatment o f collocations and idioms in learner’s dictiona
ries. Applied L ingu istics ,2 ,2 2 3 -3 5 .
C o w i e , A., M a c k i n , R. &M c C a i g , I. 1983. Oxford dictionary of current idiomatic
English. Oxford: Oxford U. Press.
C r u s e , D. 1986. Lexical semantics . Cambridge: Cambridge U. Press.
E v e r a e r t , M. & v a n d e r L i n d e n , E. 1989. Proceedings of the first Tilburg
workshop on idioms. Tilburg: ITK Proceedings.
©The Editorial H oard of Studia Linguistic» 1996.
156 HenkBarkema
E v eraert, M. 1989, Verbal idioms and the lexicon. Proceedings of (he first Tilburg
workshop on idioms ed. M. Everaert & E . van der Linden, 1989, 57-78.
E v e r a e r t , M. 1993, Vaste verbindingen (in woordenboeken). Spektator 22,1.
E v e r a e r t , M., E , V a n d e r L i n d e n , A. S c h e n k , & R. S c h r e u d e r (eds.) 1992,
Proceedings of idioms, Tilburg: ITK-publication.
E v e r a e r t , ML, E . V a n d e r L i n d e n , A. S c h e n k & R , S c h r e u d e r (eds) 1995,
Idioms: structural and psychological perspectives. Hillsdale (New Jersey):
Lawrence Erlbaum Associates.
F e r n a n d o , C . & F l a v e l l , R. 1981. On idiom: critical views and perspectives,
Exeter linguistic studies, 5, ed. R. Hartmann. Exeter: U. of Exeter.
F i l l m o r e , C., K a y , P. &O ’C o n n e r , M. 1988. Regularity and idiomaticity in gram
matical constructions: the case of let alone . Language 64, 501-38.
F i r t h , R. 1 9 5 1 . Modes of meaning. Essays and studies , 4 . The English Association.
F r a s e r , B . 1970. Idioms within a transformational grammar. Foundations of
Language 6, 22-42.
F r a s e r , B. 1976. The verb-particle combination in English. New York: Academic
Press.
F r i e s , U . , G . T o t t i e & P . S c h n e i d e r (eds.) 1994, Creating and using English
language corpora. Papers from the fourteenth international conference on
English language research on computerized corpora. Zurich 1993. Series: Lan
guage and Computers, 13. Amsterdam: Rodopi.
G a z d a r , G., K l e i n , E.. P u l l u m , G. & S a g , I, 1985. Generalized phrase structure
grammar . Cambridge, Mass.: Harvard U. Press.
G e e r a e r t s , D. 1992. Specialisation and reinterpretation in idioms. Idioms:
structural and psychological perspectives, ed. Evaraert et al,
G r e e n b a u m , S. 1970. Verh-Intensifier collocations in English. The Hague:
Mouton.
H a w k i n s , J. & A l l e n , R. 1991. The Oxford encyclopedic English dictionary.
Oxford: Clarendon Press.
H e a l e y , A . 1968. English idioms. Kivung , 1, 71-108.
H o c k e t t , C. 1958. A course in modern linguistics. New York: Macmillan.
J e s p e r s e n , 0 . 1 9 2 4 . The philosophy of grammar. London: Allen & Unwin.
K a t z , J. & P o s t a l , P. 1963. Semantic interpretation of idioms and sentences
containing them. MIT Quarterly Progress Report , 70: 275-82.
Louw, B. 1993, Irony in the text or insincerity in the writer? The diagnostic
potential of semantic prosodies. Text and technology ed, M. Baker, G. Francis &
E. Tognini-Bonelli. Amsterdam: John Benjamins.
M a k k a i , A» 1972. Idiom structure in English. The Hague: Mouton,
M el’cuk, I. 1995. Phrasemes in language and phraseology in linguistics, Idioms:
structural; and psychological perspectives ed. M. Everaert et al. 167-233.
M o o r e , T. & C a r l i n g , C. 1982. Language understanding: towards a post-
Chomskyan linguistics, New York: St. Martin’s Press.
N a t t i n g e r , J. 1980. A l e x i c a l p h r a s e - g r a m m a r f o r ESL. TESOL Quarterly, 14,3,
337-44.
N e w m e y e r , F. 1974, The regularity of idiom behavior. Lingua 34, 307-42,
N i c o l a s , T. 1992. Internal modification of English V-NP idioms. Idioms: struc
tural and psychological perspectives, ed. Everaert et al. 85-96.
N i c o l a s , T. 1995. Semantics of idiom modification. In: Everaert et at. (eds,) 1995:
223-52.
P a w l e y , A. & S y d e r , F. 1983. Two p u zzles for linguistic theory: nativelike
selectio n and nativelike fluency. Language and communication , ed. J. Richards
& R. Schmidt, 191-227, London: Longman.
Q u i r k , R., G r e e n b a u m , S., L e e c h , G . & S v a r t v i k , J, 1985. A comprehensive
grammar of the English language. London: Longman.
© T h e Editorial Board o fS ludia Linguistica 1996.
Renouf, A. 1984. Corpus development at Birmingham U. Corpus linguistics If ,

ed. J. Aarts & W. Meijs. 1984; 3-39.
S a u s s u r e , F. d e . 1916. Cours de linguistique générale. Paris: Payot. Translation:
Course in gen era I linguistics . L o nd on : Fo n tana. 1974.
S i m p s o n , J. & E. W e i n e r . 1989. The Oxford English dictionary. Oxford:
Clarendon Press.
S i n c l a i r , J. 1987. Collins COBUILD English language dictionary, Glasgow:
Collins.
S u m m e r s , D. 1992. Longman dictionary o f English language and culture . Harlow:
Longman.
S w e e t , H. 1889. The practical study of languages. London: Oxford U. Press,
Reprinted in 1964.
T o v e n a , L. 1992. Polarity sensitive idioms in translation. Proceedings o f idioms.
ed. Everaet et al. 127-41. Tilburg: ITK Proceedings.
U l l m a n n , S. 1973. Semantics. An introduction to the science of meaning. Oxford:
Blackwell.
v a n d e r L i n d e n , E. 1992. idioms, non-literal language and knowledge rep resen ta-
lion, Tilburg: ITK research memo, 12,
v a n d e r V o o r t , M. & W . V o n k . 1995. You don't immediately die when you kick
an empty bucket. Idioms: structural and psychological perspectives , ed. M. Ever
aert et al. 283-300.
W e i n r e i c h , U. 1969. Problems in the analysis of idioms. Substance and structure
of language, ed. J. Puhvel. 23-81. Berkeley: U. of California Press,
W i l l i a m s , E. Unpublished manuscript. Remarks on lexical knowledge. Princeton
U.
W o o d , M. 1986. A définition of idiom, Indiana, Bloomington: Indiana U. Linguis
tics Club.
Henk Barkema
TOSCA Research Group
Dept, of Language and Speech
Corpus Linguistics Section
Faculty of Arts
Uni varsity of N ijm egei i
P.O. Box 9103
6500 HI) Nijmegen
The Netherlands
e-mail: barkema® let.kun.nl
Appendix; alphabetical list of terms and their definitions

— absent sense: an absent sense expression is a lexicalised expression which con
tains at least one lexical item which does not have a (basic) sense in isolation
(see also: basic sense).
— alternant: the different lexical realizations of the base form of one and the same
collocationally limited or open lexicalised expression are its alternants (see
also: pseudo-alternant and pun alternant).
~ base form: a base form is the simplest form of an expression which contains the
lexical items and exhibits the minimal syntactic structure required for the speci
fic meaning and/or pragmatic function of the expression. A base form is always
a constituent (see also: template form, variant form and base pattern).
©The Editorial B oard of Studia Linguistics 1096.
158 Henk Barkema
base pattern: a base pattern is the syntactic structure of a base form, expressed
in terms of syntactic functions, categories and features (see also: base form),
basic sense: the basic sense of a lexical item is the sense from which other senses
can be systematically derived by means of extension. It is usually the first sense
that comes to a speaker’s mind when s/he comes across a lexical item in isola
tion (see also: extended and derived sense),
collocability: the coUocability of a construction is the degree to which it is pos
sible to replace the open-class lexical items in the construction with (nearly)
synonymous or antonymous items from the same class. The degree of colloca
bility depends on the number of open class items in the construction that can be
replaced as well as on the number of items that can take their places as alterna
tives (see also: collocationally open , coUocationally limited and collocationally
closed),
collocationally closed: a collocationally closed construction is a construction in
which none of the open-class items can be replaced with (nearly) synonymous
or antonymous items from the same class; a construction can be collocationally
entirely closed, but also partly (see also: collocationally open , collocationally
limited , entirely and partly).
collocationally limited : a collocationally limited construction is a construction
in which at least one open-class lexical item can be replaced with at least one
other (nearly) synonymous or antonymous item from this class; a construction
can be collocationally entirely limited, but also partly (see also: collocationally
open , collocationally closed , entirely and partly),
collocationally open : a collocationally open construction is one in which all
open-class items in the base form can be replaced with any number of (nearly)
synonymous or antonymous items from the same class. Substitution is only sub
ject to selection restrictions; a construction can be collocationally entirely open,
but also partly (see also: collocationally limited , collocationally closed , entirely
and partly).
compositionality: the compositionality of a construction is the extent to which
its meaning is the combinatorial result of the basic or derived senses of its lexical
items and the syntactic structure of the constituent that contains these lexical
items (see also: fully compositional, pseudo-compositional , partly composi
tional and non-compositional).
construction: a construction is a phrase, clause or sentence with (a) at least two
function slots, and in addition (b) at least two lexical items from open classes, or
(c) at least one lexical item from an open class and a grammatical function word
other than the definite or indefinite article (see also: expression and standard
construction).
counterfeit form: a counterfeit form is a standard construction with the same
syntactic form and lexical items as the base form or a variant form of a lexica
lised expression. However, it does not have the same idiosyncratic meaning
and/or the same specific pragmatic function as this lexicalised expression (see
also: standard construction and lexicalised expression),
derived sense: a derived sense of a lexical item is every sense, other than the
basic sense, that a word has in isolation (see also: basic and extended sense).
entirely: an expression is entirely idiomatic or collocationally entirely open/
closed/limited if the whole expression has this characteristic, and not only part
of it (see also: partly).
exclamatory: an exclamatory expression is an expression with an institutiona
lised expressive pragmatic function: it is used by the speaker to express strong
feeling (see also:formulaic).
expression: an expression is a familiar construction (set: fam iliar construction
and lexicalised expression),
© The Editorial Board o f Studia Linguistica 1 9 % .
— extended sense ; an extended sense of a lexical item is a sense which is the unique
result of a violation of restriction rules that results in the activation of sense
adaptation rules. The following factors play a role in the sense adaptation: (a)
the basic sense of the lexical item, (b) the speaker’s stereotyped beliefs about the
referents of this basic sense and (c) the syntactic environment in which it occurs
(see also: basic and derived sense).
—familiar, a familiar construction is one which has gone through a gradual pro
cess of institutionalisation. As a result it has become current within the language
community with a specific meaning and/or pragmatic function (see also: expres
sion and lexicalised expression).
—flexibility: the flexibility of a construction is the extent to which it can take dif
ferent grammatically possible forms (see also: fully flexible, semi-flexible and
inflexible).
—formulaic: a formulaic expression is an expression with an institutionalised
interactive pragmatic function; it is addressee-oriented (see also: exclamatory),
— pee: a construction is free if the generally applicable rules can be applied to it
without any (idiosyncratic) restriction (see also: idiosyncratic and restricted).
— fidly compositional: a fully compositional construction has a meaning that is
entirely the combinatorial result of the senses of its lexical items and the syntac
tic structure of the constituent that contains these lexical items (see also:
pseudopartly -, and non-compositional).
— fully flexible: a fully flexible construction is one which can take all variations
that are grammatically possible (see also: semi-flexible and inflexible).
— idiomatic expression/idiom: an idiomatic expression or idiom is a lexicalised
expression which is non-compositional (see also: lexicalised expressions,
entirely idiomatic and partly idiomatic).
— idiosyncratic: restrictions on generally applicable rules or structures are idio
syncratic if they cannot be formulated in terms of (general) rules.
— idiosyntactic: an idiosyntactic construction is one with an idiosyncratic syntac
tic structure (see idiosyncratic).
— inflexible: an inflexible construction is one which can take no variations that are
grammatically possible, i.e. it has pseudo-variants only (see also: fully flexible
and semi-flexible).
— lexicalised expression: a lexicalised expression is a familiar construction with at
least two idiosyncratic linguistic characteristics: it has limited flexibility and in
addition it is collocationally limited and/or not fully compositional (see also:
expression and familiar construction)♦
— lexicalised variant form: if a lexicalised expression has two or more
institutionalised forms, we call the simplest form its base form and the longer
form(s) its lexicalised variant form(s) (see also: variant form).
— monosemous: an expression is monosemous if it has only one institutionalised
meaning (see also: polysemous ),
— non-compositional: a non-compositional construction has a meaning that is
not the combinatorial result of the senses of its lexical items and the syntactic
structure of the constituent that contains these lexical items (see also: fully com
positional, pseudo-compositional , partly-compositional and {partly/entirely)
idiomatic).
— partly: an expression is partly compositional partly idiomatic or collocationally
partly open/closed/limited>if one part of the expression has this characteristic,
and another part has another characteristic (see entirely).
— polysemous: an expression is polysemous if it has two or more institutionalised
meanings (see also: monosemous).
— pseudo-alternant: if in the base form of a lexicalised expression one or more lex-
ical items have been replaced by one or more (nearly) synonymous or antony-
©The Editorial Hoard of Studia Linguistic» 1996.
160 Henk Barkema
mous terms while this alternative does not have the same idiosyncratic meaning
and/or pragmatic function as the lexicalised expression or its antonym, this
construction is a pseudo-alternant (see also: alternant and pun alternant),
pseudo ■‘compositional: in a pseudo-compositional construction only part of its
meaning is the combinatorial result of the senses of all of its lexical items and the
syntactic structure of the constituent that contains these lexical items (see also:
fully compositional, partly compositional and non-compositional),
pseudo-variant: if a construction shows lexical resemblance to, and is gram
matically a variant form of, a lexicalised expression, but does not have its idi
osyncratic meaning and/or pragmatic function, it is a pseudo-variant (see also:
variation and variant form).
pun alternant: a pun alternant is a base or variant form of a lexicalised expres
sion with one or more regular or irregular substitutions that are not established
alternatives but puns (see also: alternant and pseudo-alternant).
restricted: an expression is restricted if not all generally applicable grammar
rules can successfully be applied to it. An expression can be restricted in its
compositionality, its collocability and in its flexibility. These restrictions are
generally idiosyncratic (see idiosyncratic and free).
semi-flexible: a semi-flexible construction is one which can take a limited num
ber of variations that are grammatically possible, i.e. it has both variant forms
and pseudo-variants (see also: fully flexible and inflexible),
standard a standard construction is one which has been formed by means of the
general rules of the grammar of a language (see also: expression, free and coun
terfeit form).
template form: a template form is a base form with one or more open function
slots (see also: base form).
token: a token is a realisation of an expression in context (see also: type),
type: a type is an expression, abstracted from its realisation in context (see also:
token).
variant form: a variant form is a form of a lexicalised expression with one or
more variations that has basically the same idiosyncratic meaning and/or prag
matic function as the base form (see also: variation and pseudo-variant),
variation: a variation is a single alteration in the base form of a lexicalised
expression. It involves permutation, term selection, addition or interruption
(see also: variant form and pseudo-variant).
© T h e Editorial Bonrd of Studia Linguistica 1996.

Barkema, H. (1996)

Uploaded by

Copyright:

Available Formats

Barkema, H. (1996)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Barkema, H. (1996)

Uploaded by

Copyright:

Available Formats

PDF hosted at the Radboud Repository of the Radboud University

The following full text is a publisher's version.

For additional information about this publication click this link.

This expression is different from example 1 because the meanings of its

the traditional classifications, one would be bound to sweep a lot of

2. Discussion of the literature

©The Editorial Board nf Studia Linguistica 1996.

definition is not restricted to expressions with at least two lexical items,

© T he Editorinl Board of Studia Lingüistica 1996.

Restricted collocability in English typically takes the form of word-

However, this characteristic is typical of idioms too. For example, Carter

refer to such constructions with the term 'bound collocation’). On the

© The Editorial Board o f Studia Linguistics \ 996

combinations functioning in speech. They can in turn be subdivided into free

2.3. Transformational deficiency

©The Editorial B o a rd o f Studia Linguistic« 1996,

As idiomaticity and limited flexibility are not mutually inclusive, it is

Carter (1987:59) uses the term ‘fixed expression 1 to refer to constructions

y See also Fernando & Flavell ( 1981:19).

© T h e Editorial Board oFStudui Linguistic» ) 996,

3. Introduction of a multi-dimensional model

In this paper we use the term ‘construction’ to refer to a phrase, clause o r

be made explicit. A practical advantage is that with a few labels an accurat

m See Sinclair (1987:1076): the entry of ‘PHR\

© T h e Editorial B o a rd o f Studia Linguistica 1996.

3.1. Familiarity and lexicalisation

These are strings which the speaker or hearer is capable of consciously

11 For a discussion of the term ‘currency’: see Aarts (1991:57-8).

©The Editorial B oard o f Studia Linguistic« Î 99fi,

‘restricted expression’. Finally, when it has limited m orphosyntactic

familiar construction: expression

3.2. Com positionality

© T h e Editorial Board o f Studia Linguistica 1996.

now describe ‘compositionality’. The compositionality of a construction is

© T h e Editorial B oard o f Studia Linguistica 1996,

The meanings of some literal lexicalised expressions encom pass m ore

©The Editorial B oard of Sludia 1-inguisticu 1996.

lexical expression, but, because of the way in which it is used, has a

© T he Editorial B oard o f Studia Linguistica 1996.

These ‘absent sense’ lexical items only occur as part of lexicalised

17 See Jespersen (1924), Healey (1968: 72) and Tovena (1992).

©The Editorial B oard o f SluUia U nguistf ca 1 9 % .

© The Editorial Board of Stud in Linguistics 1 996.

2,1 These examples are from The Guardian 1992.

©TheEdiioriul Hoard o f Studin Unguisticu I W 6 .

(46) a ruling passion vs. a ruling passion to discover new facts

© T h e Editorial Board o f Studia Linguistica 1996.

Examples of pseudo-variants are:

CThe Editorial Board of Studtu Linguistica l ‘J % .

© T he Editorial Board of Studia Linguistica 1996.

limited’ and such replacements by (near-)synonyms or antonyms ‘regular

(69) m on key business/tricks

We call such expressions collocationally limited as well, and call such

(70) to work/operate/run a m achine (not e.g.: *to manage a m a ch in e , *to

Another example of such a collocationally limited construction is:

©The Editorial Board o f Stud ia Linguistic» I yyfi.

they are used. However, as they are subject to collocational limitation,

27 See also Fernando & Flavell (1981:38).

© T h e Editorial Board of Sludia Linguistic« 1996.

3.5. Other types o f characteristics

(70) to work/operate/run a m achine (not e.g.: to manage a m a ch in e , to