You are on page 1of 20

This article was downloaded by: [University of Barcelona]

On: 03 January 2012, At: 08:43


Publisher: Psychology Press
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Metaphor and Symbol


Publication details, including instructions for authors and subscription
information:
http://www.tandfonline.com/loi/hmet20

The Effect of Semantic Analyzability of


Idioms in Metalinguistic Tasks
Cristina Cacciari & Maria Chiari Levorato
Available online: 17 Nov 2009

To cite this article: Cristina Cacciari & Maria Chiari Levorato (1998): The Effect of Semantic
Analyzability of Idioms in Metalinguistic Tasks, Metaphor and Symbol, 13:3, 159-177
To link to this article: http://dx.doi.org/10.1207/s15327868ms1303_1

PLEASE SCROLL DOWN FOR ARTICLE


Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions
This article may be used for research, teaching, and private study purposes. Any substantial
or systematic reproduction, redistribution, reselling, loan, sub-licensing, systematic supply,
or distribution in any form to anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation
that the contents will be complete or accurate or up to date. The accuracy of any
instructions, formulae, and drug doses should be independently verified with primary
sources. The publisher shall not be liable for any loss, actions, claims, proceedings, demand,
or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.

METAPHOR AND SYMBOL, 13(3), 159-177


Copyright O 1998, Lawrence Erlbaum Associates, Inc.

The Effect of Semantic Analyzability of


Idioms in Metalinguistic Tasks
Cristina Cacciari
Downloaded by [University of Barcelona] at 08:43 03 January 2012

Universitri di Bologna

Maria Chiara Levorato


Universid di Padova

Fifth graders' and adults' abilities to assign an interpretation and explain and motivate
an idiom's meaning were investigated with idioms characterizedby different degrees
of semantic analyzability. By semantic analyzability, we mean the extent to which
the meanings of the words forming an idiom contribute to its figurative interpretation.
Forty-five Italian 5th graders and 15 adults participated in this study and were asked
to (a) provide a paraphrase of idiom meanings, (b) explain the reasons motivating the
meaning of idioms, (c) answer whether a 1st grader (i.e., a 6-year-old child) might
understand the idioms, and (d) indicate what a 6-year-old can do to understand them.
The results showed that semantic analyzability indeed affected the ways in which
children and adults explained and motivated an idiom's meaning. Children often
paraphrase an idiom by providing a figurative interpretation that is strictly incorrect
but does not reflect the literal meaning of the idiom string. The results suggest that
children possess a figurative competence based on multiple sources of linguistic
information, among which the semantic transparency of the idiom string indeed plays
a major role.
The aim of this study was to investigate the knowledge and strategies children and
adults use to make sense of idioms that vary in their semantic analyzability; that is,
in the extent to which the meanings of the words forming an idiom contribute to its
overall figurative interpretation. More precisely, we investigated the linguistic
knowledge and metalinguistic abilities used by both children and adults when asked

Requests for reprints should be sent to Cristina Cacciari, Dipartimento di Psicologia, Vide BertiPichat 5,40127 Bologna, Italy. E-mail: ccacciari@psibo.unibo.it

Downloaded by [University of Barcelona] at 08:43 03 January 2012

160

CACCIARI AND LEVORATO

to paraphrase, explain, and motivate the meaning of a specific class of linguistic


expressions-namely, idiomatic expressions varying with respect to the perceived
similarity between the "literal" and the idiomatic meanings.
How is the interpretation of a sentence assigned?According to standard models
of language comprehension, the meaning of the words is retrieved from semantic
memory and combined according to morphosyntactic rules of the language. Traditionally, idiomatic expressions have been considered as the exception that confirms
the rule-that is, as long words with semantically empty constituents whose
meaning is directly retrieved from the mental lexicon and not computed afresh, as
it is for literal sentences (e.g., Bobrow & Bell, 1973; Swinney & Cutler, 1979).
Recently, a new wave of studies questioned this standard account of idioms and the
comprehension models derived from it (Cacciari & Glucksberg, 1990;Gibbs, 1987,
1991, 1994; Wasow, Sag, & Nunberg, 1983). A quasi-compositional view has
emerged that primarily concerns idioms with a clear metaphorical origin (e.g., "spill
the beans, " "burn the candle at both ends, " "carry a torch ").
What is the relation between the meaning of a word (e.g., candle, torch, or beans)
when it is part of a literal sentence and when it is, instead, embedded in an idiomatic
expression? For those who consider idioms as semantically empty strings, the
answer is quite straightforward: Phrasal idioms possess a single semantic representation unrelated to the meanings of their parts (e.g., Cruse, 1986). Recent
evidence, however, supports an opposite viewpoint: McGlone, Glucksberg, and
Cacciari (1994; Glucksberg, 1993) investigated how people process variant idioms-that is, new figurative expressions in which one or more constituents of an
original idiom are changed (e.g., "spilling a single bean, " "biting off much less
than one could chew"). Their results showed that not only were variant idioms
easily understood but, more relevantly, they were also understood as rapidly as their
literal paraphrases, suggesting that variant idioms are processed in much the same
way as literal expressions. The proposed explanation is that variant idioms, like
literal expressions, require linguistic processing: The words that form familiar
idioms, by repeated usage, might come to incorporate at least part of the figurative
meaning they have when embedded in idioms. This extended meaning is preserved
even when such words are used out of an original idiom-for instance, in a variant
idiom.
Of course, not all idioms behave in the same way. For instance, Nunberg (1978;
see also Cacciari & Glucksberg, 1990; Gibbs, Nayak, Bolton, & Keppel, 1988)
distinguished among different levels of analyzability-or decomposability--of
idiom meaning: In some of them, figurative meaning is more transparently conveyed by word meanings (e.g., "skate on thin ice, " "burn the candle at both ends ")
than in other cases where the relation is fairly opaque (e.g., "trip the lightfantastic, "
"kick the bucket").

Downloaded by [University of Barcelona] at 08:43 03 January 2012

The specific role of the information conveyed by the words forming familiar
idioms has also been investigated in a developmental framework (Abkarian, Jones,
& West, 1992; Cacciari, 1993; Cacciari & Levorato, 1989; Gibbs, 1987, 1991;
Levorato & Cacciari, 1995, in press; Nippold & Rudzinski, 1993; Nippold &
Taylor, 1995).Gibbs (1987,1991) investigated the comprehension of idioms where
the meanings of the parts contributed independently to their overall figurative
meaning (e.g., "putyourfoot down, " "wait on hand andfoot") and nonanalyzable
idioms (e.g., "throw in the sponge, " "beat around the bush") in children ranging
from kindergartners to first, third, and fourth graders. Gibbs found that idiom
analyzability accounted for the ease with which children comprehended an idiom.
Younger children understood analyzable idioms better than nonanalyzable idioms.
Older children understood both kinds of idioms equally well in context. Without
context, children provided more accurate interpretationsfor analyzable idioms than
for nonanalyzable idioms. Levorato and Cacciari (in press) investigated the extent
to which children's comprehension of idioms was sensitive to semantic analyzability in participants from Grade 1 through Grade 4. The results suggest that to
comprehend idioms, younger children use both contextual information and semantic analyzability of the idiom strings, whereas older children mainly use semantic
analyzability .
The role of the semantic analyzability of idioms has been investigated mainly
to test possible effects on children's comprehension of idioms. What has been
relatively ignored (with the important exception of Nippold & Rudzinski, 1993;
Nippold & Taylor, 1995) is the ability of children to explain and motivate idioms
that differ in terms of semantic analyzability and the development of children's
argumentative structures. Would children be able to capitalize on the resemblance
of literal and figurative meanings and to attribute a rationale whereby an idiom
means what it says? Nippold and Rudzinski presented some evidence in favor of a
role of semantic analyzability using an explanation task. Further support for the
idea that the ability to analyze an idiom is related to the development of figurative
language understanding was provided by Nippold and Taylor, who used an explanation task and a forced-choicetask with idioms varying in terms of familiarity and
semantic transparency with school-age children and adolescents.
Our study extends and goes beyond these developmental studies in that it used
a set of related metalinguistictasks for investigating the argumentativestructure of
explanationsand motivations for idiom meanings. The children's answers also were
compared with those of a group of adults presented with the same tasks and idioms.
A large corpus of evidence already exists suggesting that adults have intuitions
concerning the rationale underlying idiom meaning (Cacciari & Glucksberg, 1990;
Gibbs, 1994; Gibbs et al., 1988; Nunberg, 1978) and use such intuitions when
necessary for discourse purposes. Less evidence exists, however, on the reasoning

Downloaded by [University of Barcelona] at 08:43 03 January 2012

162

CACCIARI AND LEVORATO

strategies people use to explain why an idiom means what it says (Cacciari, 1993;
Gibbs, 1994). When an idiom provides an apt example-such as "as alike as two
peas in a pod9'-this can be helpful in comprehension because it provides a clear
example of the kind of situations in which it might be appropriate to use such an
idiomatic expression. At the same time, children should be able to go beyond a
mere restatement of the idiom when asked to explain it. Metalinguistic skills are
needed to go beyond a simple restatement and even more so with less transparent
idioms, where a restatement strategy is less appropriate-if not impossible-and
more elaborate reasoning capacities are required.
To summarize, this study investigated the following two topics: (a) children's
and adults' ability to paraphrase, explain, and motivate the meaning of three
different types of idiomatic expressions that differ in terms of semantic analyzability and (b) the differences between children's and adults' metalinguistic abilities.
We predict that the analyzable idioms should be the easiest to paraphrase, explain,
and motivate for both groups of participants because of the straightforward relation
between the constituent word meanings and the overall figurative interpretation.

METHOD
Participants
Forty-five Italian fifth graders (age range = 10.3to 11.2, mean age = 10.1)attending
a primary school in Reggio Emilia, serving middle-class families, and 15 students
from the University of Bologna volunteered to participate in this experiment. An
almost equal number of male and female participants in both age groups were
interviewed.

Materials and Procedure


Three different types of idioms that varied along a continuum of semantic analyzability were used as experimental materials. Because no norming study is available
on the semantic analyzability of Italian idioms for children, we used judgments
obtained in a previous study conducted on Italian adults (Cacciari & Glucksberg,
1995). Twelve booklets containing 9 idioms each were prepared and given to an
independent group of 12 participants who were asked to (a) write a paraphrase of
each idiom's meaning; (b) explain the origin of the idiom (if they knew it); and (c)
rate on a 7-point scale the extent to which the literal action or mental state denoted
by the idiom was related to, or similar to, the figurative meaning of the idiom. The

IDIOM ANALYZABILITY

163

Downloaded by [University of Barcelona] at 08:43 03 January 2012

scale ranged from 1 (the literal meaning is dissimilarfrom the idiomatic meaning)
to 7 (the two meanings are related). The three types of idioms (listed in the
Appendix together with English word-by-word and meaning translations) include
the following:

1. Quasi-metaphorical idioms. These should be the easiest to paraphrase and


explain so long as they use the strategy followed by good metaphors-namely, they
"call to mind aprototypical or stereotypicalinstance of an entire category of people,
events, situations or actions" (Glucksberg, 1993, p. 18; e.g., "jeeling like a caged
animal" for feeling constrained).
2. Transparent idioms. These should be easy to understand because they are
characterized by a discernible relation between an idiom's component words and
its stipulated meaning. They are nonetheless less analyzable than quasi-metaphorical idioms, in which the metaphorical ground is explicitly described by the vehicle
(e.g., "making a hole in the water" for being unsuccessful).
3. Opaque idioms. No apparent relation is perceivable between literal and
idiomatic meanings. Children might understand them as conventional familiar
formulas but be unable to explain them (e.g., "breaking the ice ").
We selected idioms very familiar to children. Idiom familiarity was tested in a
previous study (Levorato & Cacciari, 1992) in which 152 Italian primary school
teachers were presented with 85 common Italian idioms and were asked to rate on
a 5-point scale ranging from 1 (never heard or found) to 5 (very often heard or
found) the frequency with which children may have experienced each idiom in
conversations, textbooks, or television programs for children. The mean ratings
(and standard deviations) for semantic analyzability and familiarity were then
calculated. (For quasi-metaphorical idioms, semantic analyzability mean = 5.6, SD
= 0.9, and familiarity mean = 3.5, SD = 0.7. For transparent idioms, semantic
analyzability mean = 4.3, SD = 0.8, and familiaritymean = 3.1, SD = 0.9.For opaque
idioms, semantic analyzability mean = 3.2, SD = 1.3, and familiarity mean = 3.3,
SD = 0.7.)
Three lists of 10 familiar idioms each, one formed by quasi-metaphorical,one
by transparent, and one by opaque idioms, were used (see the Appendix). For each
child, a questionnaire was prepared that contained one of the three lists with the 10
idioms in a randomized order. For each idiom, the child was asked to answer, in
written form, four questions that were reported on the same page with each idiom.
The four questions concerned (a) the paraphrase of the idiom's meaning ("'What
does the idiom mean?"); (b) the explanation-that is, the reasons motivating the
meaning of the idiom ("'Why do people use the idiom?'); (c) a judgment of
understandability concerning a younger child's ability to understand an idiom
("Can a first grader understand the idiom?'); and (d) a metalinguistic judgment

164

CACCIARI AND LEVORATO

Downloaded by [University of Barcelona] at 08:43 03 January 2012

concerning the ways in which a younger child might understand an idiom ("'What
can a 6-year-old child do in order to understand the idiom?').
Each list and questionnairewas submitted to agroup of 15 children. The children
completed their questionnaires individually while seated in a quiet room of the
school. Each experimental session lasted 20 to 30 min. A questionnaire containing
the entire set of 45 idioms, in a randomized order, was presented individually to
the university students, who were advised that the questionnairewas part of a study
on children's acquisition of figurative language.
The answers given to the four questions by both the children and the adults were
classified independently by three independent judges. Interjudge agreement ranged
from 94% to 98%.

RESULTS AND DISCUSSION


Paraphrases
We examine first the answers given by adults and children to the question "What
does the idiom mean?'The paraphrases obtained were classified as follows:
1. Idiomatic paraphrase. The participant provided the correct idiomatic meaning.
2. Literal paraphrase. The participant interpreted the idiom literally (e.g., for
the idiom "lookingfor a needle in a haystack, " one answer was "looking for a very
small thing in the straw").
3. Figurative paraphrase. A figurative meaning, different from the conventional idiomatic meaning, was proposed that expressed a plausible interpretation of
the idiom (e.g., for "breaking the ice, " a child wrote, "overcoming something very
unpleasant"; for "lookingfor a needle in a haystack", another suggested, "looking
for answers to very difficult questions"; for "feeling like a caged animal, " which
means "feeling very uneasy," a child wrote, "feeling without feelings").
A ceiling effect emerged for adults, who gave idiomatic paraphrases in 90.6%
of their responses. Such paraphrases represented 60.6% of the children's answers
(vs. 11.3% literal paraphrases), showing once again (cf. Cacciari & Levorato, 1989;
Gibbs, 1987, 1991; Levorato & Cacciari, 1995) that 11-year-old children already
knew the meaning of many familiar idioms and were able to paraphrase them even
in an out-of-context condition.
An interesting result extending previous evidence was the percentage of figurative paraphrases (28%) given by children. In previous reports, such an answer was
obtained when idioms were presented in rich informative contexts (Cacciari &

Downloaded by [University of Barcelona] at 08:43 03 January 2012

IDIOM ANALYZABILITY

165

Levorato, 1989; Levorato & Cacciari, 1995, in press). In this study, figurative
paraphrases were provided when children perceived that the expression was not
literally meant but did not yet know its conventionalized idiomatic form. Literal
paraphrases represented only 11.3% of the answers the children gave, suggesting
again that processing language on a strictly literal level is almost a peripheral
strategy for most I 1-year-old children. The percentages of idiomatic, figurative,
and literal paraphrases according to idiom type and age level (i.e., children vs.
adults) are presented in Table 1.
The children's paraphrases, unlike those of the adults, were influenced by idiom
type. A log-linear model for the analysis of the association between type of
paraphrase and type of idiom showed the reliability of the type of paraphrase factor,
x2(2,N = 45) = 171.05,p < .0001, and of the interaction between these two factors,
x2(4,N = 45) = 17.45, p < .002. Idiomatic paraphrases were significantly more
frequent for quasi-metaphorical idioms (72.6%) than for transparent and opaque
idioms (52% and 57.3%, respectively; in the saturated model, z = 3.045, p < .01).
No difference emerged between transparent and opaque idioms.
These results suggest that when the literal referent of the idiom is itself an
instance of the idiomatic meaning, as in quasi-metaphorical idioms (e.g., in "being
like two drops of water" for "being very similar"), children were easily able to
paraphrase it idiomatically. Semantic analyzability can be characterized as a
continuum whose end poles are quasi-metaphorical idioms on one side and opaque
idioms on the other, with transparent idioms located in the middle. Hence, the lack
of a significant difference between transparent and opaque idioms is problematic,
suggesting that, developmentally, the difference between transparent and opaque
idioms might not always be so clear-cut. The answers obtained to the following
questions further clarify this point.
The next analysis concerns the answers given by the children and adults to the
question "Why does one use the idiom?"

TABLE 1
Percentage of Idiomatic (I), Figurative (F), and Literal (L) Paraphrases
According to Idiom Type and Age Level
Idiom Type

Age Level

Quasi-metaphorical

Children
Adults
Children
Adults
Children
Adults

Transparent
Opaque

166

CACCIARI AND LEVORATO

Explanations

Downloaded by [University of Barcelona] at 08:43 03 January 2012

The answers given by children and adults were categorized according to the
following five categories:
1. Properties of the literal referent. Participants mentioned the properties that
typically characterize the person, object, or action depicted in the idiom. For
instance, for "being like two drops of water, " a child wrote that the drops of water
are all alike; another mentioned for "being like cat and dog" ("being enemies")
that cats and dogs are always enemies; and another wrote for "being afraid of one's
own shadow" ("'being afraid of everything") that a shadow is harmless.
2. Properties of the idiomatic referent. Participants pointed to the characteristics
that a person (or event) has when he or she is described by using an idiom. This
type of explanation seems to reflect the idea of adirect and perceivable link between
an idiom's meaning and its linguistic realization, reflecting a nominal realism
applied to idiomatic meanings. For instance, for "lookingfor a needle in haystack, "
a child wrote, "there are people that complicate things so much and want to look
for something that cannot be found7'; for "being like dogs and cats," a child
suggested that it was "because two people hate each other"; for "crying over spilt
milk, " a child pointed out that it is "because one cries after having done something
and not before."
3. Analogies between the literal referent and the idiomatic referent. Participants
elaborated on the analogy between the action or character described by the words
forming an idiom and its figurative meaning. These answers show a high degree of
elaboration on the relation between the literal and idiomatic meanings of the idiom
string. For "breaking the ice, " one child wrote that "the ice is hard to break and the
more time that passes, the more icy the silence gets"; another pointed out that "if
something wrong happens to you, you break the ice when you get over your
sadness"; for "beingat the seventh sky" ("being in seventh heaven "), a child wrote
that it is "because she is so happy that she jumps up to the seventh sky"; another
child pointed out that "everyone knows that the sky is wonderful, so just imagine
if such a thing as a seventh sky existed!"
4. Reference to use. Participants reported that idioms are formulaic expressions
commonly used in everyday conversations. For instance, one child wrote that "it's
a way of saying," and another wrote that "it's a proverb."
5. Missing answers. Participants were unable to give an explanation.

The frequencies of these five types of explanations for children and adults are
shown in Table 2. The answers given by participants were submitted to a log-linear
model for analysis of the association between type of explanation and type of idiom.
The type of explanation factor yielded significance for the children, X2(4,N = 45)
= 121.45, p < .0001, as well as the interaction with the type of idiom factor, ~ ' ( 8 ,

IDIOM ANALYZABKITY

167

TABLE 2
Percentage of Different Types of Explanations According to Idiom Type and Age Level

Explanation Type

Age Level

QuasiMetaphorical

Transparent

Opaque

Literal referent

Children
Adults
Children
Adults
Children
Adults
Children
Adults
Children
Adults

42.0
78.6
34.0
0.6
12.6
14.6
11.3
5.3
0.6

22.0
20.6
35.3
11.3
20.0
58.0
1.3
0.6
21.3
9.3

15.3
14.0
41.3
9.3
26.0
48.0
4.6
2.0
22.6
26.6

Idiomatic referent
Analogy

Downloaded by [University of Barcelona] at 08:43 03 January 2012

Reference to use
Missing

N = 45) = 118.52, p < .0001. The saturated model analysis showed that the
explanations based on the "properties of the idiomatic referent" (z = 8.34, p < .01)
were more frequent than expected and were given regardless of the idiom type.
Two types of answers were responsible for this interaction: (a) the "properties
of the literal referent" answer that was-unsurprisingly-more
frequent for quasimetaphorical idioms (z = 4.97, p < .01), less frequent for opaque idioms, and
intermediate for transparent idioms (z = -5.67, p < .01) and (b) "missing answers,"
more frequent for transparent and opaque idioms (z = 3.16, p < .01, and z = 3.08, p
< .01, respectively). This result replicated the result obtained in the paraphrases of
transparentand opaque idioms. These two types of idioms obtained almost the same
pattern of results with the exception of the explanations based on the "properties
of the literal referent" (22% and 15.3% for transparent and opaque idioms, respectively).
Overall, quasi-metaphorical idioms markedly differed from the other two types
of idioms, suggesting that these idioms were indeed the easiest to understand and
explain. It should be noted that they often explicitly mention part of their idiomatic
meaning, making the restatement strategy fairly easy to adopt.
As to the answers given by the adults, a significant effect of the type of answer
factor was obtained, x2(4, N = 15) = 294.02, p < .0001, as was a significant
interaction with the type of idiom factor, x2(8, N = 15) = 207.4, p < .0001. The
saturated model analysis showed that explanations based on the "properties of the
literal referent" and on "analogies" were significantly more frequent than expected
~ .01, and z = 10.05, p < .01, respectively). Unlike what happened with
(z = 8 . 3 1 , <
children, the "properties of the idiomatic referent" answers were less frequent than
statisticallyexpected (z = -2.6, p < .05). Quasi-metaphorical idioms elicited mostly
the "properties of the literal referent" explanations (z = 7.4, p < .01), presumably

Downloaded by [University of Barcelona] at 08:43 03 January 2012

168

CACCIARI AND LEVORATO

because of the straightforward nature of the relation between the expression and its
figurative meaning. Such answers were instead-and unsurprisingly-negatively
associated with opaque idioms (z = - 5 . 7 , ~< .Ol) that obtained a positive association
with answers such as "I don't know" (z = 4.01, p < .01).
A comparison using a log-linear model was performed between children's and
adults' explanations, showing that the two types of answers did, in fact, differ: A
developmental tendency toward an increase in analogies was observed (z = -4.6
for children and z = 4.6 for adults, p < .01) as well as a decrease in explanations
based on the idiomatic referent (z = 6.1 for children and z = -6.1 for adults, p <
.01). The adults used more analogies that consisted in identifying the semantic
relation-the figurative similarity-between the literal content of the idiom and
its figurative meaning: Analogies increased when passing from quasi-metaphorical to transparent and opaque idioms. On the other hand, the adults rarely gave
explanations centered on the idiomatic referent (only 7.1% of answers), presumably because the mention of the idiomatic referent does not contain any information on why a linguistic community has selected a certain word string to express
a figurative action or state. Such explanations, on the contrary, were preferred by
children. This suggests that the children often acted as if the meaning was
transparent and the significance of a situation or sentence was self-evident,
implicitly assuming a direct and transparent link between form and meaning.
When passing from quasi-metaphorical to transparent and opaque idioms, a
decrease in the explanations based on the literal referent was observed for both
children and adults. Moreover, both groups were unable to explain a number of
transparent and opaque idioms: Adults mostly failed with opaque idioms (26.6%)
and children with both transparent and opaque idioms (21.3% and 22.6%, respectively).

Judgment of Understandability
A judgment of understandability was solicited to determine whether children
realized that the semantic analyzability of an idiom can affect comprehensibility-that is, to test the metalinguistic awareness of fifth graders. Table 3 shows the
percentage of affirmative ("yes") answers to the question "Can a first grader
understand the idiom?'according to idiom type and age level.
Children did not differentiatebetween types of idioms when theyjudged whether
they were comprehensible by younger children and mostly answered negatively
(55.1%). A log-linear analysis of the association between the factors type of answer
~
and type of idiom showed the statistical reliability of only the first factor, X 2 ( =
45) = 2 4 6 . 3 , ~
< .0001, due to the high frequency of the answer "No" (z = 10.34,p
< .001). The children were unable to appreciate the difference between types of

IDIOM ANALYZABILITY

169

Downloaded by [University of Barcelona] at 08:43 03 January 2012

TABLE 3
Percentage of Affirmative Answers According to Idiom Type and Age Level
Age Level

Quasi-Metaphorical

Transparent

Opaque

Children
Adults

38.6
82.0

44.0
38.6

20.6
25.3

idioms and likely answered on the basis of their negative evaluation of younger
children's cognitive abilities.
In the opinion of the adults, on the contrary, 6-year-old children could understand
quasi-metaphorical idioms (82%) more easily than transparent idioms (38.6%),
which in turn were judged as more comprehensible than opaque idioms (25.3%),
the least understandable of all. In fact, the log-linear model applied to the adults'
answers showed that the reliability of the interaction between the type of answer
and the type of idiom factors, X2(4,N = 15) = 110.9, p < .0001, was due to the
preference for the answer "Yes" for quasi-metaphorical idioms (z = 5.74, p < .01)
and "No7' for opaque idioms (z = 2.83, p < .01). Affirmative and negative answers
did not reliably differ for transparent idioms, which were in an intermediateposition
between quasi-metaphorical and opaque idioms.
These results suggest that whereas the adults were sensitive to the structure of
idioms, the children did not distinguish between the demands made by different
kinds of idioms on 6-year-olds.
The next question, "What can a 6-year-old child do to understand the idiom?,"
can illustrate more thoroughly the strategieschildren and adults attribute to a young
child faced with the task of comprehending an idiomatic expression. The aim of
this question was to induce children and adults to state explicitly the interpretive
strategies by which they identify the semantic relation between the linguistic form
of the string and the idiomatic meaning.

Metalinguistic Strategies

The children's and adults' answers were coded using the following categories:
1. Asking adults. Participants stated that in order to understand an idiom's
meaning, one should ask adults.
2. Observing (or pe@orming) what is literally expressed by the idiom. Participants maintained that the observation, or performance in the case of actions, of what
was literally meant by the string was enough to clarify an idiom's figurative
meaning. For instance, for the idiom "drowning in a glass of water" ("getting lost

Downloaded by [University of Barcelona] at 08:43 03 January 2012

170

CACCIARI AND LEVORATO

over nothing"), one child wrote that "it is impossible to drown in a glass of water
because a glass is too small, so it means that you are worried for nothing"; for
"crying over spilt milk, " one child pointed out that "one should spill milk and then
despair"; for "being afraid of one's own shadow, " a child wrote that one should
"put himiherself in front of a lamp and watch hislher shadow"; for "being as mute
as afish" ("being silent and keeping secrets"), a child wrote that one should "look
at an aquarium and observe the great conversations of fishes!"
3. Observing (or performing) the idiomatic referent expressed by the idiom.
Participants reported that indicating or recalling to the child situations in which the
idiom was appropriate was a viable way of making him or her understand the idiom.
For instance, for "cutting the rope" ("escaping"), a child wrote, "He or she should
feel afraid and then escape very quickly"; for "being on the clouds" ("being
absentminded"), a child pointed out that "you are in the classroom, the teacher is
explaining and you are thinking of something else"; another child suggested, "to
remind him or her that it had happened to him or her too, to think of something else
while someone was asking him or her something"; for "costing an eye of the head"
("costing a lot of money"), one child wrote that one should "watch a person who
is spending a lot of money ."
4. Metalinguistic activities. The appreciation of or reflection on the meaning of
the words was seen as sufficiently clear to grasp the figurative meaning. For
instance, one child wrote that "one can examine the words very carefully"; another
child pointed out that "one should think about the words."
5. Missing answers. Participants simply said, "I don't know."
Table 4 reports the percentages of answers according to idiom type, for children
and adults, respectively. Even though half of the children (55.1%) judged that a
TABLE 4
Percentage of Different Types of Metalinguistic Judgments According to
Idiom Type and Age Level
-

Judgment Type

Age Level

QuasiMetaphorical

Asking

Children
Adults
Children
Adults
Children
Adults
Children
Adults
Children
Adults

28.6
11.6
49.3
73.5
11.3
2.0
6.0
5.4
4.6
7.5

Literal referent
Idiomatic referent
Metalinguistic activity
Missing

Transparent

Opaque

8.6
24.2
34.6
39.6
32.6
3.3
4.0
4.7
20.0
28.2

30.6
25.7
9.3
22.3
22.0
3.4
14.6
4.0
23.3
44.6

Downloaded by [University of Barcelona] at 08:43 03 January 2012

IDIOM ANALYZABILITY

171

young child was unable to understand idioms, they nonetheless listed several ways
in which he or she could grasp idiomatic meaning. The answer "I don't know"
following a negative judgment was in fact rare (12.9%).The most frequent answers
for children were as follows: "Observing what is literally expressed by the idiom7'
(31.1%), "Observing the idiomatic referent of the idiom" (22%), and "Asking
adults7'(22.6%). The reference to metalinguistic activities was the least frequent
(8.2%). A log-linear analysis showed that both the type of answer factor and its
interaction with type of idiom yielded significance, x2(4,N = 45) = 68.9, p < .0001,
and x2(8,N = 45) = 116.8,p < .0001, respectively.
The frequencies of the five types of answers for the three types of idioms
markedly differed. "Observing the literal action" (49.3%) was given for mostly
quasi-metaphorical idioms (z = 6.02, p < .01), less for transparent idioms (34.6%),
and the least for opaque idioms (9.3%, z = -6.9, p < .01). Not surprisingly, and
consistent with the results obtained for the explanations, this answer represented
almost half of the total answers given by children for quasi-metaphorical idioms.
Here, in fact, the literal referent of the idiom is in itself an instance of the idiomatic
meaning and exemplifies the action expressed by it. Children seemed to imply that
if one carefully observes the surrounding world (cats and dogs, two drops of water,
etc.), then one could understand what an idiom typifies. Children who were unable
to indicate this strategy suggested "asking adults" (28.6%, z = 3.5 1,p < .01).
The answers associated with transparent idioms showed that these were perceived as more semantically transparent than opaque idioms, although much less
so than quasi-metaphorical idioms. Transparent idioms elicited a high percentage
of "Observing the idiomatic action" answers (32.6%), which were much less
frequent for quasi-metaphorical idioms (11.3%, z = 3.7, p < .01), and a high
percentage of "Observing the literal action" answers (34.6%, z = 2.51, p = .01).
However, unlike quasi-metaphorical idioms (4.6%), transparent idioms elicited a
percentage of "I don't know" answers (20%, z = 2.7, p < .05) that was also very
close to that of opaque idioms (23.3%,z = 2.57, p < .05). Transparent idioms were
perceived as intermediate between quasi-metaphorical idioms, cognitively the
easiest, and opaque idioms, the most difficult.
It should be noted that the answer "Observing the idiomatic action" can be
interpreted in two ways: It may indicate the perception of a link, if not a direct
mapping, from literal to idiomatic referents for transparent idioms. But it could also
be a "way out" type of answer when given for opaque idioms, as it reflects a "naive"
belief that word meanings are self-evident. The presence of such a belief is also
suggested by metalinguisticanswers such as "one should reflect on word meanings"
that were positively associated with only opaque idioms (z = 2.86, p < .05).
Opaque idioms differed from both quasi-metaphorical and transparent idioms
in that they obtained the highest percentages of "Asking adults" answers (30.6%)
and "I don't know" (23.3%) answers. This is not surprising because, by definition,
there is no detectable relation between literal and figurative meaning for opaque

Downloaded by [University of Barcelona] at 08:43 03 January 2012

172

CACCIARI AND LEVORATO

idioms. The answer "Observe the idiomatic action" qualitatively differs when given
for opaque (22%), transparent (32.6%), or quasi-metaphorical idioms (1 1.3%). In
fact, children are very unlikely to use this information to infer the meaning of an
opaque idiom, whereas they possibly would for transparent and quasi-metaphorical
idioms.
Let us examine the adults' answers. Surprisingly, they gave a remarkable
percentage of "I don't know" answers (z = 4.71, p < .01)-26.8% overall, but
note the 44.6% given for opaque versus the 7.5% for quasi-metaphorical idioms.
It is reasonable to assume that the adults meant something different from the
children. The children's use of this answer (15.6% vs. 26.8% by adults) might
reflect a real lack of knowledge (and, in fact, it was more frequent with
transparent and opaque idioms). The adults, on the contrary, might endorse an
implicit theory about children, according to which children cannot come to
know a figurative meaning that is, by definition, conventionally established.
The most frequent answer for adults was "Observing the literal action" (45.1%,
z = 12.14, p < .001), which-unsurprisingly-was
overwhelmingly used for
quasi-metaphorical idioms (73.5%); a bit less often, but still quite frequently, for
transparent idioms (39.6%); and much less frequently for opaque idioms (22.3%).
The second most frequent answer was "Asking adults" (20.5%, z = 4.08, p < .01),
which was especially used for transparent (24.2%) and opaque idioms (25.7%, z =
4.72, p < .01) and much less for quasi-metaphorical idioms (11.6%). This answer
can be interpreted in two ways: Either adults had little confidence in the ability of
young children, or they were convinced of the difficulty implied in the acquisition
of figurative expressions. The reliability of the interaction between the two factors-type of answer and type of idiom-x2(8, N = 15) = 96.55, p < .0001, was
primarily due to two answers: "Observing the literal action," which was positively
associated with quasi-metaphorical idioms (z = 5.9, p < .01) and negatively
associated with opaque idioms (z = -4.63, p < .01), and "I don't know," which was
negatively associated with quasi-metaphorical idioms (z = -4.04, p < .01) and
positively associated with opaque idioms (z = 4 . 2 7 , ~< .01).
"Observing the idiomatic referent" was used much less by adults (z = -6.05, p
< .Ol) than by children (2.9% and 22%, respectively). The same difference was also
found in the production of explanations, where adults mentioned the idiomatic
referent much less often than children. It is worth noting that for adults, the
frequency of two answers, "Asking adults" and "I don't know," increased according
to the extent to which idioms were close to the opaqueness end pole (i.e., when one
passed from quasi-metaphorical to transparent and then to opaque idioms). On the
other hand, a decrease in the "Observing the literal referent" answer was observed
when one got closer to the analyzability end pole, that is, when one passed from
quasi-metaphorical to transparent and then to opaque idioms. This again suggests
that the meanings of the constituent words contribute differently to the overall
figurative meaning.

IDIOM ANALYZABILITY

173

Downloaded by [University of Barcelona] at 08:43 03 January 2012

GENERAL DISCUSSION
Let us concentrate first on the children's performance. They proved to be remarkably good at paraphrasing idioms (60.6%).If idiomatic paraphrases were considered together with figurative paraphrases, they were close to the ceiling (88.6%),
suggesting that 11-year-old children had already abandoned a literal strategy in
interpreting idioms, even when idioms were presented out of context.
According to the hypothesis we tested, idioms did not form a homogeneous class
of linguistic entities and are located along a continuum whose end poles are
quasi-metaphorical idioms on the one hand and opaque idioms on the other, with
transparentidioms in an intermediate position. From adevelopmental point of view,
this hypothesis did not gain full support. In fact, transparent and opaque idioms did
not obtain a significantly different percentage of idiomatic paraphrases. Idiomatic
paraphrases were instead significantly more frequent for quasi-metaphorical idioms.
The hypothesis did, however, gain some support from the explanations children
gave of the idiom's meaning and from the strategiesindicated for interpreting them.
The explanations focused on the properties of the literal referent were used in an
interesting way that varied according to the idiom type: They were more frequent
for quasi-metaphorical idioms, less so for transparent idioms, and rare for opaque
idioms. Further support to the hypothesis also came from the answers given to the
question concerning what a young child could do to interpret an idiom. Reference
to the literal action was indicated much more for quasi-metaphorical idioms than
for transparent idioms, and even less for opaque idioms. If the explanations
elaborated by our participants can provide some hints about the comprehension
strategies, then these results suggest that the three types of idioms do indeed differ.
When asked about younger children's ability to understand idioms, the children
answered negatively. Such an answer mightreflect an implicittheory about younger
children. This is supported by the fact that although children mostly disputed the
idea that younger children could understand idioms, they nonetheless answered the
question concerning how a young child might come to understand an idiom. The
children indicated strategies that varied according to the type of idiom, with a
polarization of quasi-metaphorical idioms on the one hand and transparent and
opaque idioms on the other hand. The children seemed to be aware that when an
idiom was a clear instance of a more general class of actions, as with many good
metaphorical vehicles (cf. Glucksberg, 1993), its explanation and possible understanding might be straightforwardly derived by restating the referent, a strategy by
definition useless for opaque idioms and unhelpful for transparent idioms, where
what counts is the web of related meanings and connotations that a speaker or reader
can associate with the constituent word meanings.
Let us now examine the adults' answers and compare them with those of the
children. Children's and adults' answers differed not only with respect to the

174

CACCIARI AND LEVORATO

Downloaded by [University of Barcelona] at 08:43 03 January 2012

amount of idiomatic paraphrases produced (90.6% for adults vs. 60.6% for children)
but also in the explanations, understandability judgments, and strategies they
proposed. Such differences can be summarized as follows:
1. The adults generally proved to be more sensitive to the semantic analyzability
of idioms.
2. The adults used many more analogies than the children for explaining
idiomatic meanings (40.2% vs. 19.5%, respectively)-a result consistent with
previous evidence that analogies (and particularly those based on a relation between
systems of properties) are cognitively difficult for children and are acquired later
than other figurative expressions (Gentner, 1989).
3. The adults used fewer explanations centered on the idiomatic referent (only
7.1% as opposed to 37% for children).
4. Adults and children shared the same preference for the characteristics of
the literal referent to explain quasi-metaphorical idioms, but the percentage of
choice for the adults was almost twice that of the children (78.6% vs. 42%,
respectively).
- 5. The adults considered younger children able to understand idioms. The
adult group judged quasi-metaphorical idioms as the clearest, followed by
transparent and then by opaque idioms, but they showed a relative uncertainty
about the strategies younger children might successfully use, as shown by the
high percentages of the answers "I don't know7' (26.8%) or "Asking adults"
(20.5%).
6. When asked about possible strategies for younger children's understanding of idioms, the adults modulated their answers according to the type of
idiom. They mentioned-as did the children-that quasi-metaphorical idioms
might be understood by focusing on the literal properties of the referent. But as
far as explanations were concerned, the adults relied much less on the idea that
idiomatic meanings are transparently conveyed by the words forming transparent
and opaque idioms.
Quasi-metaphorical idioms were the easiest to paraphrase, explain, and interpret
for both the children and the adults. This is not surprising if one reflects on the
structure of these idioms: An instance of an event or situation is used to describe
new instances of such events or situations.Furthermore, quasi-metaphorical idioms
usually mention, as typical instances of the issue at hand, objects and actions present
in the everyday life of children (dogs, cats, drops of water, etc.), which allows ready
entrance to interpretation. The literal referents are relevant to the idiom's interpretation, and they therefore can be used in providing interpretations, as in pointing
out analogies.

IDIOM ANALYZABILITY

175

Downloaded by [University of Barcelona] at 08:43 03 January 2012

REFERENCES
Abkarian, G. G., Jones, A,, & West, G. (1992). Young children's idiom comprehension: Trying to get
the picture. Journal ($Speech and Hearing Research, 35, 580-587.
Bobrow, S., & Bell, S. (1973). On catching on to idiomatic expressions. Memory & Cognition, 1,
343-346.
Cacciari, C. (1993).The place of idioms in a literal and metaphorical world. In C. Cacciari & P. Tabossi
(Eds.), Idioms. Processing, structure and interpretation (pp. 27-55). Hillsdale, NJ: Lawrence
Erlbaum Associates, Inc.
Cacciari, C., & Glucksberg, S. (1990). Understanding idiomatic expressions:The contribution of word
meanings. In G. B. Simpson (Ed.), Understanding word and sentence (pp. 217-240). Amsterdam:
Elsevier.
Cacciari, C., & Glucksberg, S. (1995). Understanding idioms: Do visual images reflect figurative
meanings? European Journal of Cognitive Psychology, 7,283-305.
Cacciari, C., & Levorato, M. C. (1989). How children understand idioms in discourse.Journal of Child
Language, 16,387-405.
Cruse, D. A. (1986). Lexical semantics. New York: Cambridge University Press.
Gentner, D. (1989). The mechanisms of analogical learning. In S. Vosniadou & A. Ortony (Eds.),
Similarity and analogical reasoning (pp. 19S241). Cambridge, England: Cambridge University
Press.
Gibbs, W. R. (1987). Linguistic factors in children's understanding of idioms. Journal of Child
Language, 14, 569-586.
Gibbs, W. R. (1991). Semantic analyzability in children's understanding of idioms. Journal of Speech
and Hearing Research, 34, 613-620.
Gibbs, W. R. (1994). Thepoetics of mind. Figurative thought, language and understanding.Cambridge,
England: Cambridge University Press.
Gibbs, W. R., Nayak, N. P., Bolton, J. L., & Keppel, M. (1988). Speakers' assumptionsabout the lexical
flexibility of idioms. Memory & Cognition, 17, 58-68.
Glucksberg,S. (1993). Idiommeaning and allusionalcontent. In C. Cacciari & P. Tabossi (Eds.), Idioms.
Processing, structure and interpretation (pp. 3-26). Hillsdale, NJ: Lawrence Erlbaum Associates,
Inc.
Levorato, M. C., & Cacciari, C. (1992). Children's comprehension and production of idioms: The role
of context and familiarity. Journal of Child Language, 19,415-433.
Levorato, M. C., & Cacciari, C. (1995). The effects of different tasks on the comprehension and
production of idioms in children. Journal of Experimental Child Psychology, 60,261-283.
Levorato, M. C., & Cacciari, C. (in press). Idiom comprehension in children: Are the effects of semantic
analyzability and context separable? European Journal of Cognitive Psychology.
McGlone, M., Glucksberg, S., & Cacciari, C. (1994). Semantic productivity and idiom comprehension.
Discourse Processes, 17, 167-190.
Nippold, M. A., & Rudzinski, M. (1993). Familiarity and transparency in idiom explanations: A
developmental study of children and adolescents. Journal of Speech and Hearing Research, 36,
728-737.
Nippold, M. A,, & Taylor, L. C. (1995). Idiom understanding in youth: Further examination of
familiarity and transparency. Journal of Speech and Hearing Research, 38,426-433.
Nunberg, G. (1978). The pragmatics of reference. Bloomington: Indiana Linguistic Club.
Swinney, D. A,, & Cutler, A. (1979). The access and processing of idiomatic expressions. Journal of
Verbal Learning and Verbal Behavior, 18, 523-534.

176

CACCIARI AND LEVORATO

Wasow, T., Sag, I., & Nunberg, G . (1983). Idioms: An interim report. In S. Hattori & K. Inoue (Eds.),
Proceedings of the XIIIth International Congress of Linguistics @p. 102-105). Tokyo: CIPL.

APPENDIX
The following sections list experimental idioms used in their original Italian form
with word-by-word and meaning translations.

Downloaded by [University of Barcelona] at 08:43 03 January 2012

Quasi-Metaphorical Idioms
1. Avere paura anche della propria ombra 'to be afraid also of one's own
shadow' (to be afraid of everything)
2. Essere come due goccie d'acqua 'to be like two drops of water' (to be very
similar)
3. Essere noioso come una mosca 'to be as annoying as a fly' (to be annoying)
4. Essere come cane e gatto 'to be like dog and cat' (to be enemies)
5. Essere muto come un pesce 'to be as silent as a fish' (to be silent and keep
secrets)
6. Essere in un mare di guai 'to be in a sea of troubles' (to be in great trouble)
7. Dormire come un ghiro 'to sleep like a dormouse' (to sleep very soundly)
8. Sentirsi come un animale in gabbia 'to feel like an animal in a cage' (to feel
constrained, limited in one's freedom)
9. Andare indietro come i gamberi 'to go backward like lobster' (to not make
progress)
10. Sentirsi come unpesce fuord'acqua 'to feel like a fish out of water' (to feel
uneasy)

Transparent Idioms
1. Costare un occhio della testa 'to cost an eye of the head' (to cost a lot of
money)
2. Piangere sul latte versato 'to cry over spilled milk'
3. Fare un buco nell'acqua 'to make a hole in the water' (to be unsuccessful)
4. Cercare un ago in un pagliaio 'to look for a needle in a haystack'
5. Essere sulle nuvole 'to be on the clouds' (to be absent-minded)
6. Affogare in un bicchier d'acqua 'to drown oneself in a glass of water' (to
get lost over nothing)
7. Fare ridere ipolli 'to make the chickens laugh' (to be socially unsuccessful)
8. Chiudere bottega 'to close up shop'

IDIOM ANALYZABILITY

177

9. Non sapere chepesciprendere 'to not know which fish to catch' (to be very
uncertain)
10. Essere sulle spine 'to be on the thorns' (to be very anxious)

Opaque Idioms

Downloaded by [University of Barcelona] at 08:43 03 January 2012

1.
2.
3.
4.

5.
6.
7.
8.

9.
10.

Darsi della arie 'to give oneself airs' (to boast)


Prendere per il naso 'to take someone by the nose' (to tease someone)
Essere a1 settimo cielo 'to be at the seventh sky' (to be in seventh heaven)
Avere del sale in zucca 'to have some salt in the pumpkin' (to be intelligent)
Rompere il ghiaccio 'to break the ice'
Toccare il cielo con un dito 'to touch the sky with a finger' (to be extremely
happy
Mangiare la foglia 'to eat the leaf' (to understand a secret)
Essere a1 verde 'to be in the green' (to be broke)
Cadere dalle nuvole 'to fall from the clouds' (to be astounded)
Tagliare la corda 'to cut the rope' (to escape)

You might also like