DOI: 10.1093/brain/awg243
Advanced Access publication August 5, 2003
Brain (2003), 126, 2419±2430
Idiom comprehension in Alzheimer's disease: the
role of the central executive
Costanza Papagno,1 Federica Lucchelli,2 Silvia Muggia1 and Silvia Rizzo3
di Psicologia, UniversitaÁ di Milano-Bicocca,
di Neurologia, Azienda Ospedaliera Ospedale
Niguarda Ca' Granda, Milano, 3Dipartimento di
Psicologia, UniversitaÁ di Palermo, Palermo, Italy
1Dipartimento
2Divisione
Summary
Idiom comprehension of 15 patients with mild probable
Alzheimer's disease was examined by means of a sentence-to-picture matching task. Patients had to choose
between two pictures, one representing the ®gurative
and the other the literal interpretation. They were also
submitted to a literal sentence comprehension test and
to a pencil-and-paper dual task. Whereas literal comprehension was normal in seven subjects and mildly
impaired in the others, idiom comprehension was very
poor in all of them and correlated with the performance
on the dual task. When the idiom test was repeated
using an unrelated situation as an alternative to the picture representing the ®gurative meaning, performance
signi®cantly improved. It was hypothesized that the
response in the sentence-to-picture matching task in the
case of idioms requires sentence processing followed by
the suppression of the literal interpretation. Alzheimer's
Correspondence to: Costanza Papagno, Dipartimento di
Psicologia, UniversitaÁ di Milano-Bicocca, Piazza
dell'Ateneo Nuovo 1, Edi®cio U6, 20126 Milano, Italy
E-mail: [email protected]
disease patients proved to be unable to inhibit the literal meaning, although they had not lost the idiomatic
meaning. In a second experiment, 15 Alzheimer's disease patients with a comparable level of cognitive
impairment were submitted to the same idiom comprehension test, and to a test of verbal explanation of the
idioms. The results showed signi®cantly better performance in the oral task than in the sentence-to-picture
matching task. In oral explanation, however,
Alzheimer's disease patients also produced some literal
interpretation whenever this represented a possible situation in the real world. We suggest that, during idiom
interpretation, the literal meaning needs to be suppressed in order to activate the ®gurative meaning, and
we stress the fact that both linguistic and extralinguistic
factors must be taken into account to explain idiom
interpretation.
Keywords: idiom; central executive; suppression; Alzheimer's disease
Abbreviations: MODA = Milan Overall Dementia Assessment
Introduction
An aspect of linguistic communication is the use of ®gurative
expressions, the meaning of which differs from the literal
meaning of their component words; examples of these
expressions are proverbs, hyperbole, metaphors and idioms.
Two aspects concerning ®gurative language have been
studied in detail: linguists have addressed problems related
to the mechanisms involved in the interpretation of ®gurative
language, and neuropsychologists have been concerned with
its neural correlates. These expressions, however, do not
constitute a homogeneous group, and idioms, metaphors,
proverbs etc. have different characteristics. Idioms are among
the most common forms of ®gurative language. They are
typically described as frozen phrases whose meanings are
stipulated directly in a mental lexicon, and the speaker's
meaning cannot be derived from an analysis of the words'
typical meanings. Many idioms, however, are more complex
Brain 126 ã Guarantors of Brain 2003; all rights reserved
and dynamic (Glucksberg et al., 1993). They range from
expressions that are almost like long words (e.g. `by and
large') to expressions that are like metaphors (e.g. `skate on
thin ice'). In between there is the majority of idioms, in
particular verbal idioms, that are processed syntactically and
semantically, and accordingly can undergo syntactic and
semantic variations.
Neuropsychological studies on ®gurative language comprehension in brain-damaged patients concern mainly metaphors. There is, however, some information available on
idioms, suggesting that patients with left hemisphere damage
outperform right-brain-damaged patients in ®gurative sentence comprehension (Foldi et al., 1983; Van Lancker and
Kempler, 1987; Kempler and Van Lancker, 1993; Burgess
and Chiarello, 1996), even though psycholinguistic models
all stress the importance of lexical knowledge (Swinney and
2420
Costanza Papagno et al.
Table 1 Clinical and demographic data of the Alzheimer's disease patients: Experiment 1
Patient
Age
(years)
Education
(years)
Gender
MODA
(normal values >89.1)
Verbal
¯uency
Visuoperceptual
tests
Stroop
test
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
75
83
76
68
73
58
87
55
75
76
69
71
75
59
76
12
8
8
8
5
17
8
5
12
6
8
5
5
5
17
F
F
F
M
F
F
F
F
F
F
F
F
F
M
M
63.9
89.1
73.9
80.2
82.3
81.7
67.3
79.0
78.1
76.0
88.9
85.6
76.2
72.4
83.7
0
4
1
1
2
2
0
2
4
1
4
2
4
2
1
2
3
4
3
4
2
1
4
2
1
2
3
0
3
2
0
4
1
1
0
Equivalent scores are reported for verbal ¯uency, the Stroop test and visuoperceptual abilities (range 0±
4, where 0 corresponds to a score within the lowest 5% of the normal population, 4 is equal to/better
than the median of normal population and 1, 2 and 3 are intermediate between 0 and 4 on a quasiinterval scale). Visuoperceptual abilities were tested in some patients with the Street test and in others
with the Line Orientation Judgement test. M = male; F = female.
Cutler, 1979; Tabossi and Zardon, 1995) and increasing
evidence indicates the involvement of syntactic processing
(Peterson et al., 2001) in the comprehension of idioms.
There are, in our opinion, many limitations in previous
neuropsychological studies on idioms. First, even if the literal
and non-literal sentences were matched in structure, items
were not chosen on the basis of their intrinsic linguistic
features, and idioms, proverbs and courtesy phrases were
considered together as `familiar language' (Van Lancker and
Kempler, 1987; Kempler et al., 1999). The severity of aphasia
and the modalities of assessment are not always described;
the time of testing with respect to the time of onset of aphasia
was variable and it is not clear whether patients underwent
language rehabilitation; other neuropsychological de®cits are
not reported; and the number of stimuli was limited.
Moreover, in one of these studies (Van Lancker and
Kempler, 1987), the performance of aphasic patients cannot
be considered normal since their score was ~72% correct for
idioms, compared with the controls' performance of 97.3%.
The same aphasic subjects scored 90% correct on word
comprehension.
Papagno and colleagues (Papagno et al., 2002) and
Papagno and Tabossi (2002) produced data which do not
support the right hemisphere hypothesis. In both studies, a
sentence-to-picture matching task was used for idiomatic and
literal sentences. A homogeneous class of expressions was
selected. These were non-ambiguous, opaque idioms; the
meaning could not be directly derived from the ®gure
involved and the literal interpretation was implausible
because it had no meaning (for example, the idiom `venire
alle mani' literally means `to come to the hands') (Nunberg
et al., 1994). In the ®rst study, repetitive transcranial
magnetic stimulation was used to evaluate the neural
correlates of idiom comprehension. It was found that only
left hemisphere stimulation increased reaction times and
reduced accuracy in a task in which participants had to choose
between a literal and a ®gurative representation of the
idiomatic sentence. In the second study (Papagno and
Tabossi, 2002), ten aphasic left-brain-damaged subjects
were submitted to the same tests. They showed a performance
in idiom comprehension that was signi®cantly impaired
compared with that of a matched group of normal controls,
even when aphasic de®cits were mild. These dif®culties seem
to be due, to a large extent, to the fact that patients, unlike
unimpaired listeners, rely on a literal-®rst strategy in the
interpretation of idioms, and access a ®gurative interpretation
only when the linguistic analysis fails to yield acceptable
results (Bobrow and Bell, 1973).
As far as patients with Alzheimer's disease are concerned,
it has been suggested that their behaviour is similar to that of
right-brain-damaged patients (Winner and Gardner, 1977). It
is not clear which expressions were used, how severe their
dementia was and how it was tested. Kempler and colleagues
examined 29 patients with probable Alzheimer's disease
ranging from mild (Mini-Mental State Examination score 28)
to severe (Mini-Mental State Examination score 2); they were
given tests of words, idioms, proverbs (overall, ten stimuli)
and novel phrase comprehension (Kempler et al., 1988). In
the ®gurative language comprehension task, no alternative
corresponding to the literal interpretation was available, but
only a referential representation of one word in the stimulus
(concrete response). The results showed that Alzheimer's
disease patients had dif®culty interpreting abstract meanings:
when faced with alternative interpretations of familiar
phrases, they chose concrete responses, suggesting that they
were using lexical (single-word), referential meaning to
Idiom comprehension in Alzheimer's
2421
Fig. 1 example of the idiom comprehension (sentence-to-picture matching) task. (A) The alternative corresponds to the literal
interpretation. (B) The alternative corresponds to a sentence including a word contained in the idiom. This ®gure can be viewed in colour
as supplementary data at Brain Online.
interpret the phrases (Kempler et al., 1988). Familiar phrases
are considered as psychological `chunks' similar to words,
perceived and understood as wholes, without analysis of
constituent parts taking place at all.
Although we disagree with this interpretation and suggest
that idioms undergo processes very similar to those involved
in the comprehension of literal expressions, in particular
a full syntactic analysis (C.Papagno, P.Tabossi, M.R.Colombo,
P.Zannetti, unpublished results; Peterson et al., 2001), we
accept that idiom comprehension can be impaired for other
reasons, apart from lexical±semantic and syntactic de®cits. In
a recent study on Alzheimer's disease patients, it was shown
that the decline of ®gurative language (assessed by means of
an oral explanation task of metaphors and idioms) is not an
early symptom of dementia, can be preserved longer compared with propositional language, and can occur independently of the impairment of propositional language (Papagno,
2001). Idioms were both ambiguous and non-ambiguous,
decomposable and non-decomposable, frozen and unfrozen.
Performance on the idiom comprehension test was contrasted
with performance on the Token Test. There were, however,
very few patients with an impairment of ®gurative language
but normal propositional language. It was suggested that an
idiom needs to be present as a whole in the subject's semantic
store, and then it must be correctly accessed and retrieved.
Semantic memory and executive functions, therefore, seem to
2422
Costanza Papagno et al.
which corresponds to a mild degree of cognitive impairment,
were included in the study. Other criteria of inclusion were (i)
the absence of previous neurological diseases, (ii) the absence
of lesion on the CT scan, and (iii) a minimum of 3 years of
education. Demographic and clinical data for the 15
Alzheimer's disease patients are reported in Table 1. They
all gave informed consent to participate in the study.
Procedure
Fig. 2 Example of the literal sentence comprehension task.
be involved. Consistent results were found by Papagno and
Vallar (2001), investigating the ability of a patient with Down
syndrome to appreciate non-literal and literal aspects of
language. Whereas the latter was largely preserved, the
former was defective. Some executive aspects of working
memory were also impaired. What could be taken for certain
in this patient was that impaired appreciation of metaphor and
idioms could not be interpreted in terms of syntactic and
lexical±semantic aspects of language comprehension, but
might involve executive functions.
Patients with Alzheimer's disease have impaired executive
functions, even in the early stage (Baddeley et al., 1997).
Therefore, we decided to explore idiom comprehension in
this group of subjects, given that Alzheimer's disease patients
with no de®cit in propositional language comprehension but
impaired ®gurative language comprehension have been found
(Papagno, 2001). In these cases (impaired idiom comprehension, normal literal comprehension), dif®culties in idiom
comprehension cannot be attributed to lexical±semantic and
syntactic de®cits.
In particular, we examined the relation between idiomatic
and literal language in Alzheimer's disease patients and the
role of executive functions in idiom comprehension.
Experiment 1
Subjects
Fifteen patients (12 women and three men, mean age 71.73,
SD 8.86; years of education 8.6, SD 4.12) with probable
Alzheimer's disease were selected on the basis of their
clinical history, their CT or MRI ®ndings and their performance on the Milan Overall Dementia Assessment (MODA;
Brazzelli et al., 1994). Only patients with a score >63/100,
Forty unambiguous verbal idioms were selected (see
Appendix). They were rated as familiar by a group of 45
healthy subjects from different parts of Italy (31 women and
14 men, age 22±65 years), who had to judge whether each
idiom was unknown (score 0), known but not sure about the
meaning (score 1), familiar, with a well-known meaning
(score 2), or highly familiar (score 3).
Idioms were unambiguous: either they were ill-formed
from a syntactical point of view (some Italian idioms contain
a syntactic violation) or their literal interpretation was
unlikely, because it made no sense (as in the example
above, `venire alle mani') or the situation was implausible in
the real world (e.g. `spremersi le meningi', which means `to
squeeze one's own meninges'). In a ®rst version, for each
idiom, two line drawings were created, one representing the
idiomatic interpretation and the other representing as well as
possible the literal interpretation (Fig. 1A). In a second
version, the test was administered again with two alternatives,
but the picture representing the literal interpretation was
replaced by one representing an unrelated situation, which
was de®ned by a sentence containing one word of the
idiomatic expression (for example, when the idiomatic
expression was `to have a green thumb', the alternative was
a man hurting his thumb) (Fig. 1B). Each patient performed
both versions.
Tests
Patients performed three tasks. The order of presentation was
counterbalanced across participants.
Idiom comprehension
This was a sentence-to-picture matching task: the examiner
read a sentence corresponding to an idiom to the patient, who
chose between two pictures. Patients' performance on this
task was compared with the performance of a group of 15
elderly healthy participants, matched for age and educational
level.
Literal sentence comprehension (Parisi and
Pizzamiglio, 1970)
This test included 80 items. The patient was shown a pair of
pictures while one sentence was read to him. One of the
Idiom comprehension in Alzheimer's
2423
Table 2 Results of tests in the Alzheimer's disease patients: Experiment 1
Patient
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Dual task
(normal value 91.99 6 11.10)
Literal
sentences
Idioms
Literal/®gurative
Figurative/unrelated
0.37
(±1.24)
0.82
(0.68)
0.89
(0.98)
0.48
(0.76)
0.99
(1.41)
0.86
(0.85)
0.79
(0.55)
0.125
(±2.28)
0.68
(0±08)
0.72
(0.25)
0.65
(±0.04)
0.55
(±0.46)
0.81
(0.64)
0.76
(0.42)
0.40
(±1.11)
60/80
(0.68)
74/80
(0.90)
76/80
(1.1852)
75/80
(1.0463)
69/80
(0.213)
59/80
(±1.1759)
66/80
(±0.203)
51/80
(±2.2871)
72/80
(0.6296)
62/80
(±0.7593)
74/80
(0.907)
71/80
(0.4907)
66/80
(±0.2037)
64/80
(±0.4815)
73/80
(0.7685)
15/40
(±0.71)
29/40
(±1.044)
34/40
(1.671)
18/40
(±0.334)
24/40
(0.417)
34/40
(1.671)
19/40
(±0.208)
9/40
(±1.4622)
18/40
(±0.3342)
15/40
(±0.71)
17/40
(±0.4595)
16/40
(±0.5849)
30/40
(1.1697)
22/40
(0.1671)
10/40
(±1.3368)
18/40
(±2.9057)
34/40
(0.1664)
38/40
(0.9344)
37/40
(0.7424)
32/40
(±0.2176)
37/40
(0.7424)
33/40
(±0.0256)
29/40
(±0.7936)
31/40
(±0.4096)
35/40
(0.3584)
39/40
(1.1265)
38/40
(0.9344)
34/40
(0.1664)
32/40
(±0.2176)
30/40
(±0.6016)
Figures in parentheses are Z scores.
pictures was correctly described by the sentence; the other
corresponded to a sentence that was identical except for a
detail (i.e. `the boy draws' versus `the boy will draw'). The
sentences were matched, as far as possible, in length and
complexity to idiomatic sentences. As no control subject
scored <70 on this task in the original study and in a
subsequent one (Vallar et al., 1988), the score of 70 was taken
as the cut-off point (Fig. 2).
Pencil-and-paper dual task (Baddeley et al.,
1997)
After being tested for a baseline span (the maximum length at
which the patient recalled three lists without errors), patients
were presented with lists of digits at their own span for a
period of 2 min. Patients were then required to cross out, in a
speci®c sequence, boxes (1 cm square) which had been linked
to form a path laid out on an A4-size sheet of white paper over
a 2-min interval. Finally, the two tasks were combined (digit
span and tracking) for 2 min. A score combining the
performance on the two tasks was given (control mean =
91.99 6 11.10). This test was chosen because it proved to be
particularly sensitive to early dysexecutive changes in
Alzheimer's disease patients.
Results
The results for the 15 patients in the three tests are shown in
Table 2. The mean number of correct responses in idiom
comprehension when a literal foil was used was 20.67 (SD
7.97), whereas for the 15 healthy participants the mean
number of correct responses was 34.53 (4.37). The t-test
showed that the difference between Alzheimer's disease and
control subjects was highly signi®cant [t(14) = ±5.963,
P < 0.0001]. Sentence comprehension proved to be preserved
in seven patients and mildly impaired in the others, with the
exception of Patients 6 and 8, whose impairment was
moderate. Performance on idiom comprehension was correlated with that on the dual-task (rS = 0.89, P < 0.001), whereas
there was no correlation between the former and the degree of
cognitive decline, as assessed by the MODA (rS = ±0.26,
P = 0.92) or sentence comprehension (rS = 0.24, P = 0.37).
Finally, sentence comprehension was not correlated either
2424
Costanza Papagno et al.
Table 3 Clinical and demographic data for the Alzheimer's disease patients: Experiment 2
Patient
Age
(years)
Education
(years)
Gender
MODA
(normal values >89.1)
Verbal
¯uency
Visuoperceptual
tests
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
76
79
84
82
85
79
78
78
67
69
56
76
65
77
82
5
17
3
3
5
5
12
5
13
5
8
8
13
11
6
M
F
F
F
F
M
M
F
F
F
M
M
F
F
F
80.6
77.8
82.7
85.4
77.9
85.6
85.6
86.4
79.9
82.2
78.7
83.9
79.8
86.5
74.3
4
1
4
1
0
4
4
4
4
3
1
1
0
3
1
4
3
4
2
1
4
4
4
3
4
4
1
0
2
1
Stroop
test
0
For explanation see footnote of Table 1.
with dual-task performance (rS = 0.16, P = 0.54) or with
cognitive decline (rS = 0.41, P = 0.13). We submitted 20
normal subjects to a rating task to check the plausibility of the
pictures representing the literal interpretations. The subjects
were asked to rate how well the picture corresponding to the
literal interpretation of the idiom matched the sentence read
by the examiner. A score ranging from 0 to 7 was assigned,
where 0 meant `not at all' and 7 `perfectly'. A correlation
analysis indicated that the more implausible an idiomatic
sentence is as a linguistic description of the literal picture, the
more likely it is that, upon presentation of that sentence, the
patient will select the idiomatic picture (rS = ±0.32, P < 0.05).
In contrast, there was no effect of familiarity in this task
(rS = 0.02, P = 0.87).
When the test was re-administered in the second version
(one picture representing the idiomatic interpretation and the
other representing an unrelated situation including a word
contained in the idiom), performance signi®cantly improved
(mean of correct responses 33.13, SD 5.20) [t(14) = ±6.38,
P < 0.0001], showing that patients had a good knowledge of
the meaning of the idiom, despite their poor performance in
the previous task. Every item that was selected correctly
when the alternative was the literal interpretation was also
chosen correctly when the alternative was the unrelated
situation, demonstrating a consistent response of the patient,
who was not performing at chance.
Discussion
Patients chose the literal interpretation when the corresponding picture was highly plausible but chose the idiomatic
interpretation when the alternative represented an unrelated
situation, even when the distractor corresponded to a sentence
including a word contained in the idiom. This means that
Alzheimer's disease patients know the meaning of the idiom,
but the literal interpretation, which is also activated, strongly
interferes with it; it also means that patients are not picking
the literal interpretation on the basis of a single word in the
idiom. It seems as if patients are unable to suppress the literal
interpretation when there is an overt representation of it. We
suggest that the activation of the literal interpretation is
stronger or perhaps quicker than the activation of the
idiomatic meaning, as hypothesized by Bobrow and Bell
(1973), in contrast with the conclusion drawn with normal
subjects (Gibbs, 1980). The interference produced by the
literal interpretation is similar to that observed in the Stroop
test: the choice of the literal representation can be due to the
greater automaticity of the inhibition of the response with
respect to that actually requested (MacLeod, 1991). It has
been demonstrated that Alzheimer's disease patients produce
signi®cantly larger interference effects with the Stroop ColorWord Test than comparable normal people (Bondi et al.,
2002)
To further investigate this possibility (interference of the
literal interpretation), we selected another group of
Alzheimer's disease patients, comparable with respect to
their degree of cognitive impairment, and we submitted them
to a second experiment. The hypothesis was that when a
verbal explanation is required, people are not `offered' an
external literal interpretation but generate their own response
internally. Indeed, the representation of the literal meaning
could reinforce the literal interpretation, which seems to take
place. If the idiomatic meaning were unknown, a greater
number of unrelated responses would be expected, the literal
ones being produced only in the case of a highly plausible
meaning.
Experiment 2
Subjects
Fifteen Alzheimer's disease patients, ®ve men and 10 women
(mean age 75.53 years, SD 7.98, range 56±85; years of
education 7.93, SD 4.26, range 3±17), entered the study. They
Idiom comprehension in Alzheimer's
2425
Table 4 Results of the tests in Alzheimer's disease patients: Experiment 2
Patient
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Dual task
(normal value 91.99 6 11.10)
Idioms
String-to-picture
Oral explanation
0.69 (±0.49)
0.88 (0.59)
0.74 (±0.207)
0.39 (±2.248)
0.79 (0.08)
0.73 (±0.2651)
0.90 (0.7145)
0.82 (0.2536)
0.73 (±0.2651)
0.80 (0.1386)
0.96 (1.0605)
0.87 (0.5418)
0.91(0.7723)
0.43 (±1.9942)
1 (1.2911)
6/40
22/40
24/40
9/40
11/40
10/40
16/40
14/40
27/40
13/40
37/40
27/40
22/40
25/40
28/40
21/40
25/40
25/40
26/40
14/40
32/40
29/40
33/40
33/40
19/40
26/40
28/40
33/40
25/40
31/40
(±1.5)
(0.29)
(0.52)
(±1.2)
(±1)
(±1.1)
(±0.4)
(±0.6)
(0.86)
(±0.7)
(1.99)
(0.86)
(0.29)
(0.63)
(0.97)
(±1)
(±0.3)
(±0.3)
(±0.1)
(±2.3)
(0.96)
(0.42)
(1.14)
(1.14)
(±1.4)
(±0.1)
(0.24)
(1.14)
(±0.3)
(0.78)
Values in parentheses are Z scores
were comparable to the patients of the previous group in
severity of disease [t(14)= ±1.353, P = 0.19], mean age
[t(14)= ±1.225, P = 0.24] and education [t(14) = 0.406,
P = 0.69]. The same criteria of inclusion were followed.
Demographic and clinical data of the patients are reported in
Table 3. All patients gave informed consent to participation in
the study.
Procedure
In this experiment patients were subjected to the dual task, to
the idiom comprehension test in the sentence-to-picture
matching modality, and to a test of oral explanation of idioms.
In this last task, the same idioms were presented verbally to
the patient, who was asked to explain their meanings (after
being instructed that he/she was expected to give a ®gurative
interpretation of the sentence). The oral explanation was
accepted when at least two out of three independent judges
considered it adequate. The order of tasks was counterbalanced across subjects. We assumed that when Alzheimer's
disease patients have simply to explain the meaning of an
idiom, they attain a better performance than when they have
to operate a choice between two alternatives, where the
concrete representation interferes with the retrieval of the
abstract interpretation. In other words, the lexical knowledge
of idioms is preserved, but patients are not able to reject the
literal meaning when this is offered to them.
Results
The results for the 15 patients in the three tests are shown in
Table 4. Their performance was signi®cantly worse than that
of 15 healthy participants, matched for age and educational
level, both in the sentence-to-picture matching task [t(14) =
±5.377, P < 0.0001] and in the oral task [t(14) = ±6.994,
P < 0.0001]. However, the performance of the Alzheimer's
disease patients was signi®cantly better in the oral explanation task (mean number of correct responses 26.27, SD 5.57)
compared with their performance in the sentence-to-picture
matching task (mean number of correct responses 19.4, SD
8.84) [t(14) = ±3.17, P < 0. 01]. Whereas for 35.5% of the
total answers the patients gave a correct response in both
tasks, 31.2% of answers were correct in the oral task and
wrong in the sentence-to-picture matching task. Responses
were wrong in both tasks in 20.5%, and only 12.8% of
responses were correct in the sentence-to-picture and wrong
in the oral task.
As in the previous experiment, the performance on idiom
comprehension in the sentence-to-picture matching task
correlated with that on the dual task (rS = 0.54, P < 0.05),
while there was no correlation between the former and the
degree of cognitive decline (rS = ±0.29, P = 0.28). Also,
performance in the oral task did not correlate with cognitive
decline (rS = 0.22, P = 0.41) or dual-task performance
(rS = 0.27, P = 0.30). The number of correct responses
correlated signi®cantly with literal picture plausibility in the
sentence-to-picture matching task (rS = ±0.35, P = 0.02). The
number of correct responses in the oral task correlated
signi®cantly with familiarity (rS = 0. 42, P < 0.01).
We also checked sentence plausibility, because some oral
responses corresponded to the literal interpretation (25%).
Ten normal controls were asked to rate how well an idiomatic
string would also evoke a literal interpretation. For example:
the sentence `to go to the cool' means in Italian `to be put in
jail', but one could accept the possibility that someone is
moving towards a cool place, while the sentence `to squeeze
his/her own meninges' does not allow any literal explanation.
Controls had to give a rating of 0 (no acceptable literal
meaning), 1 (mild literal meaning) or 2 (absolutely acceptable
literal meaning). A correlation analysis showed that
Alzheimer's disease patients produced the literal interpret-
2426
Costanza Papagno et al.
ation whenever the sentence offered this possibility: this
means that, whenever plausible, the literal meaning overpasses the ®gurative one (rS = 0.63, P < 0.0001). The
remaining errors in the oral task were mainly partial or wrong
interpretation (57%), no response (17%) and opposite meaning (1%).
Discussion
The performance of Alzheimer's disease patients was
signi®cantly better in the oral task than in the sentence-topicture matching task. In this last task performance correlated
with the plausibility of the pictures, whereas in the oral task
performance correlated with familiarity. A correlation was
also found between the literal errors produced in the oral task
and the degree of plausibility of the sentence, suggesting that,
at least out of context, the literal interpretation precedes the
®gurative one for patients with Alzheimer's disease, inhibiting further analysis whenever the result of sentence processing is plausible.
General discussion
Two experiments were conducted with Alzheimer's disease
patients. In Experiment 1, an analysis of the relationship
between idiomatic and literal language was performed.
Idiomatic comprehension was tested in two conditions, both
using a sentence-to-picture matching task paradigm: a picture
representing the idiomatic meaning was contrasted, in one
condition, with a picture representing the literal meaning,
whereas in the other condition it was contrasted with a picture
representing an unrelated situation, including a word contained in the idiom. There was no correlation between
idiomatic and literal language comprehension, as already
found in a previous study (Papagno, 2001). However,
participants had better idiomatic comprehension than is
re¯ected in the ®rst condition (idiomatic versus literal
interpretation): indeed, their performance signi®cantly improved with the unrelated alternative.
In Experiment 2, idiom comprehension was assessed in
two modalities: patients performed a sentence-to-picture
matching task (two alternatives: correct ®gurative meaning
versus literal representation) and an oral task in which they
produced a verbal explanation of the idioms. This latter task
was signi®cantly better performed than the former, but
patients still produced a literal interpretation whenever the
literal meaning of the sentence was plausible.
In both experiments, performance on idiomatic expressions
correlated with performance on the dual task.
The conclusions of this study are straightforward. First, our
data con®rm those of a previous study (Papagno, 2001), in
which ®gurative language comprehension was dissociated
from literal language comprehension, but also suggest that the
modality of testing is very important, as already outlined by
Tompkins and colleagues (Tompkins et al., 1992). In
Papagno's (2001) paper, idioms were both ambiguous and
non-ambiguous and varied in transparency, leading to more
literal interpretations in the oral modality. In the present
study, idioms were chosen in order to be non-ambiguous, and
the literal interpretation in the oral task is therefore restricted
to the few plausible ones. Kempler and colleagues used ten
stimuli, which they called `familiar phrases', generically
referring to overlearned expressions, including proverbs,
idioms and social interaction formulae, some of the items
being full sentences and others being phrases (Kempler et al.,
1988). We have already discussed the fact that even restricted
categories of idioms can be divided into subgroups with
different features, and therefore it is dangerous to consider
them together. There is another point to discuss: Kempler and
colleagues stated that the oral type of testing often presents
problems because responses can be judged to be poor, as
Alzheimer's disease patients may choose the wrong word or
may not respond to the question asked, or the response may be
uninterpretable (Kempler et al., 1988). For this reason these
authors required a picture pointing response, suggesting that
de®cient performance can be interpreted more clearly with
this type of response. Our patients, however, had only mild
cognitive impairment and, more importantly, they had no
remarkable language de®cits. Moreover, we accepted all
responses, that showed that the core abstract meaning of the
idiom was preserved. On the other hand, we have found that a
sentence-to-picture matching task can produce other problems.
Another point deserves further discussion. In the previous
study by Papagno (2001), out of 39 Alzheimer's disease
patients, only two were found with an impairment of
metaphor and idiom comprehension and preserved literal
comprehension, while six showed the opposite pattern. This
could be due to the fact that poor literal comprehension is not
suf®cient or necessary to reduce ®gurative interpretation.
Idiomatic language can be impaired not only because of
lexical±semantic and syntactic de®cits, as observed in
aphasic patients (Papagno and Tabossi, 2002), but can be
produced by different mechanisms. One possibility is inadequate suppression, as seen in the Stroop task. Brega and
Healy (1999) showed that sentence processing can be
obligatory when the component words are highly relevant
to the task. In their experiment, participants showed longer
reaction times and more errors in naming the colours in which
words were printed when the words were included in a
sentence and were colour words or colour-related words. By
analogy, one can assume that idioms are processed syntactically and semantically and that this produces the interfering
effect. Gernsbacher and Robertson (1999) had already
described the crucial role that suppression plays in many
aspects of language comprehension. They de®ne suppression
as a general, cognitive mechanism, the purpose of which is to
attenuate the interference caused by the activation of
extraneous, unnecessary or inappropriate information.
Sometimes this super¯uous activation arises from the external environment, as when we conduct a conversation in a
noisy place. At other times this super¯uous information is
activated internally, as when we have to deal with the
Idiom comprehension in Alzheimer's
competing meanings of a word or phrase. In the case of
idiomatic expressions, the literal and ®gurative interpretations can be activated in parallel, so that a mechanism for
attenuating the activation of the inappropriate interpretation
is needed. However, it has been proposed that there are some
asymmetries in what gets activated and what gets suppressed.
Giora and Fein (1999) suggested that hearing familiar idioms
should lead to both idiomatic and literal interpretations
becoming activated, because both interpretations are salient
out of context. In contrast, less familiar idioms are more
likely to activate a literal interpretation, out of context.
According to their view, the literal interpretation of an idiom
is functional for idiom interpretation. Therefore they predict
that, with familiar idioms, deriving the literal interpretation
does indeed involve suppressing the idiomatic interpretation;
however, they propose that deriving the idiomatic interpretation requires the literal interpretation to be retained. Our data
do not support this hypothesis, given that familiar idioms
need suppression of the literal interpretation in order to derive
the idiomatic meaning. It seems as if familiar idioms, out of
context, are also more likely to activate the literal meaning, so
that suppression is necessary.
We suggest that suppression is mediated by the central
executive. In the case of an absolutely implausible expression
or an expression that contains a syntactic violation (Papagno
and Tabossi, 2002), the literal meaning is less salient and can
be easily rejected even when the central executive is
damaged, but when the literal interpretation is acceptable or
is overtly suggested, suppression is necessary. As the central
executive is damaged in Alzheimer's disease patients, as
already observed in previous studies (Baddeley et al., 1991,
1997), inhibition of the literal meaning does not take place
ef®ciently and the ®gurative meaning does not get suf®cient
activation.
Finally, as a consequence of these observations, we stress
that several factors have to be taken into account in studying
idiom comprehension: their intrinsic features, such as their
compositionality, transparency (Nunberg et al., 1994) and
familiarity; linguistic aspects, such as semantic and syntactic
processing; and extralinguistic abilities, such as executive
and visuoperceptual functions.
Acknowledgements
The study was funded through a grant of the MIUR
(Ministero Istruzione, Universita Ricerca) and FAR
(Finanziamento di Ateneo per la Ricerca) (60%) to C.P.
References
Baddeley AD, Bressi S, Della Sala S, Logie RH, Spinnler H. The
decline of working memory in Alzheimer's disease: a longitudinal
study. Brain 1991; 114: 2521±42.
Baddeley AD, Della Sala S, Gray C, Papagno C, Spinnler H.
Testing central executive function with a pencil-and-paper test. In:
Rabbitt P, editor. Methodology of frontal and executive function.
Hove (UK): Psychology Press; 1997. p. 61±80.
2427
Bobrow S, Bell S. On catching on to idiomatic expressions. Mem
Cognit 1973; 19: 295±308.
Bondi MW, Serody AB, Chan AS, Eberson-Shumate SC, Delis DC,
Hansen LA, et al. Cognitive and neuropathologic correlates of
Stroop Color-Word Test performance in Alzheimer's disease.
Neuropsychology 2002; 16: 335±43
Brazzelli M, Capitani E, Della Sala S, Spinnler H, Zuf® M. A
neuropsychological instrument adding to the description of
patients with suspected cortical dementia: the Milan Overall
Dementia Assessment. J Neurol Neurosurg Psychiatry 1994; 57:
1510±7.
Brega AG, Healy AF. Sentence interference in the Stroop task.
Mem Cognit 1999; 27: 768±78.
Burgess C, Chiarello C. Neurocognitive mechanisms underlying
metaphor comprehension and other ®gurative language. Metaphor
Symbolic Activity 1996; 11: 67±84.
Foldi NS, Cicone M, Gardner H. Pragmatic aspects of
communication in brain damaged patients. In: Segalowitz SJ,
editor. Language functions and brain organization. New York:
Academic Press; 1983. p. 51±86.
Gernsbacher MA, Robertson RRW. The role of suppression in
®gurative language comprehension. J Pragm 1999; 31: 1619±30.
Gibbs RW. Spilling the beans on understanding and memory for
idioms in conversation. Mem Cognit 1980; 8: 449±56.
Giora R, Fein O. On understanding familiar and less-familiar
®gurative language. J Pragm 1999; 31: 1601±18.
Glucksberg S, Brown M, McGlone MS. Conceptual metaphors are
not automatically accessed during idiom comprehension. Mem
Cognit 1993; 21: 711±9.
Kempler D, Van Lancker D. Acquisition and loss of familiar
language: idiom and proverb comprehension. In: Eckman FR,
editor. Con¯uence: linguistics, L2 acquisition, and speech
pathology. Language acquisition and language disorders, Vol. 4.
Amsterdam: John Benjamins; 1993. p. 249±57.
Kempler D, Van Lancker D, Read S. Proverb and idiom
comprehension in Alzheimer disease. Alzheimer Dis Assoc Dis
1988; 2: 38±49.
Kempler D, Van Lancker D, V. M, Bates E. Idiom comprehension
in children and adults with unilateral brain damage. Dev
Neuropsychol 1999; 15: 327±49.
MacLeod CM. Half a century of research on the Stroop effect: an
integrative review. Psychol Bull 1991; 109: 163±203.
Nunberg G, Sag IA, Wasow T. Idioms. Language 1994; 70: 491±
538.
Papagno C. Comprehension of metaphors and idioms in patients
with Alzheimer's disease. A longitudinal study. Brain 2001; 124:
1450±60.
Papagno C, Tabossi P. idiom comprehension in aphasic patients.
Brain Lang 2002; 83: 78±81.
Papagno C, Vallar G. Understanding metaphors and idioms: a
single-case neuropsychological study in a person with Down
syndrome. J Int Neuropsychol Soc 2001; 7: 516±27.
2428
Costanza Papagno et al.
Papagno C, Oliveri M, Romero L. Neural correlates of idiom
comprehension. Cortex 2002; 38: 895±8.
familiar idioms by brain-damaged and normally aging adults. J
Speech Hear Res 1992; 35: 626±37.
Parisi D, Pizzamiglio L. Syntactic comprehension in aphasia.
Cortex 1970; 6: 204±215.
Vallar G, Papagno C, Cappa S. Latent dysphasia after left
hemisphere lesions: a lexical±semantic and verbal memory de®cit.
Aphasiology 1988; 2: 463±78.
Peterson RR, Burgess C, Dell GS, Eberhard K. Dissociation between
syntactic and semantic processing during idiom comprehension. J
Exp Psychol Learn Mem Cogn 2001; 27: 1223±37.
Swinney DA, Cutler A. The access and processing of idiomatic
expression. J Verb Learn Verb Behav 1979; 18: 523±34.
Tabossi P, Zardon F. The activation of idiomatic meaning. In:
Everaert M, van der Linden EJ, Schenk A, Schreuder R, editors.
Idioms: structural and psychological perspectives. Hillsdale (NJ):
Lawrence Erlbaum; 1995. p. 273±82.
Tompkins CA, Boada R, McGarry K. The access and processing of
Van Lancker D, Kempler D. Comprehension of familiar phrases by
left but not by right hemisphere damaged patients. Brain Lang 1987;
32: 265±77.
Winner E, Gardner H. The comprehension of metaphor in braindamaged patients. Brain 1977; 100: 717±29.
Received January 30, 2003. Revised March 26, 2003.
Second revision May 22, 2003. Accepted May 26, 2003
Idiom comprehension in Alzheimer's
Appendix
Where possible the equivalent English idiom is given, otherwise an
explanation of the ®gurative meaning is given.
Table 5 List of idioms
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
avere il pollice verde
andare al fresco
avere la puzza sotto il naso
dare del ®lo da torcere
essere in forma
andare in bestia
fare ®asco
essere al settimo cielo
mandare a monte
essere di facili costumi
prendere ®schi per ®aschi
montarsi la testa
dare i numeri
farsene un baffo
avere i minuti contati
stare alle costole
avere il cuore in mano
vendere cara la pelle
venire alle mani
mordere il freno
parlare al muro
essere sul viale del tramonto
spremersi le meningi
essere di manica larga
far cadere le braccia
non vedere l'ora
perdere la faccia
stare in campana
avere poco sale in zucca
tendere le orecchie
essere uno stinco di santo
avere le mani in pasta
prendere in castagna
lasciarci le penne
far venire il latte alle ginocchia
essere a piede libero
tenere banco
essere sulla cattiva strada
restare con un palmo di naso
mettersi le gambe in spalla
NS
NS
SV
SV
NS
SV
SV
NS
I
NS
SV
NS
NS
NS
I
SV
NS
NS
NS
NS
SV
I
NS
I
SV
SV
NS
NS
SV
SV
NS
NS
SV
SV
NS
NS
NS
IF (0±3)
PP (0±7)
SP (0±2)
2.4
2.3
2.5
2.6
2.7
2.5
2.3
2.5
2.3
2.3
2.3
2.7
2.6
2.4
2.5
2.3
2.1
1.9
2.4
1.4
2.7
1.9
2.4
2.2
2.3
2.8
2.4
2.1
2.3
2.2
2.3
1.8
2
2.1
1.9
2.1
1.9
2.4
2
1.5
4.3
2.3
4.8
3.3
1.3
2.2
1.9
2.8
3.2
1.0
1.7
2.8
3.9
1.8
1.4
2.2
5.5
3.1
0.8
3.3
4.4
3.7
1.9
2.9
3.0
3.7
3.4
2.3
2.0
3.3
1.1
2.4
2.4
2.2
1.9
2.6
2.4
2.0
1.6
3.4
0.2
1.4
0.4
0.1
0
0
0
0.1
0.3
0
0
0.1
1.1
0
0.3
0
0.3
0.7
0.1
0.2
0.6
0.7
0
0.1
0.4
1
0
0
0
0.6
0
0.5
0
0.2
0
0
0.4
0.7
0
0.1
IF = idiom familiarity; PP = literal picture plausibility; SP = literal meaning plausibility; SV = syntactic
violation; NS = lack of sense; I = real-world implausibility of the sentence. The ®rst three idioms could
have a plausible literal interpretation and were excluded from further studies.
2429
2430
Costanza Papagno et al.
Table 6 Idiom translation
Idiom
Literal meaning
Figurative meaning
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
To be good at gardening
To go in the cool
to have a bad smell under the nose
to give some thread to twist
to be in shape
to go in beast
to make ¯ask
to be at the seventh sky
to send to mountain
to be of easy customs
to take whistles for ¯asks
to mount the head
to give the numbers
to make a moustache of it
to have the minutes counted
to stay at the ribs
to have the heart in hand
to sell the skin at a high price
to come to the hands
to bite the brake
to talk to the wall
to be on the sunset boulevard
to squeeze one's meninges
to be of large sleeve
to make the arms fall
to not see the hour
to lose the face
to stay in bell
to have few salt in pumpkin
to stretch the ears
to be the shin-bone of a saint
to have the hands in dough
to take in chestnut
to leave the feathers
to make the milk come to the knees
to be at free foot
to hold bench
to be on the bad path
to remain with a span of nose
to put the legs on shoulder
to have a green thumb
to go to jail
to be snooty
to make things hard for someone
to be very well
to become very angry
to fail badly
to be in the seventh heaven
to make something fail
to be of easy virtue
to misunderstand
to become swollen-headed
to talk nonsense
not to give a damn
to have very little time
to dog someone's heels
to be very generous
to ®ght hard
to come to blows
to champ at the bit
to talk to a brick wall
to be on the wane
to rack one's brains
to be indulgent
to have one's heart sink to one's boots
to look forward
to lose face
to be very careful
to have nonsense in one's head
to prick up one's ears
to be an angel
to have one's ®nger in the pie
to catch someone out
to lose life
to be particularly boring
to be on bail
to be the person who talks a lot in a group of people paying attention to one
to make bad things
to be badly disappointed
to ¯ee