Speed of Processing Effects on Spoken Idiom Comprehension

Cristina Cacciari ([email protected])

Dept. of Biomedical Sciences, via Campi 287
41100 Modena ITALY

Paola Corradini ([email protected])

Dept. of Biomedical Sciences, via Campi 287
41100 Modena ITALY

Roberto Padovani ([email protected])

Dept. of Biomedical Sciences, via Campi 287
41100 Modena ITALY

Abstract added to the four components underlying reading

comprehension (i.e., text memory, text inferences,
We report two experiments which investigated the time knowledge access, knowledge integration). However, the
course of spoken idiom comprehension. We also assessed extent to which a specific and independent role for SOP can
whether differences in individual speed of processing be isolated in language comprehension is still controversial.
specifically affected idiom meaning activation. Using a SOP effects might in fact reflect the role of components and
cross-modal lexical decision paradigm, we found activation resources characterizing the human information processing
of ambiguous idioms meaning at offset for idioms identified system (e.g., reading skills, general knowledge level,
early in the string (early identification idioms) and at offset working memory structure, long term memory retrieval,
(late identification idioms) in fast and slow participants. The
two groups instead differed in the amount of time required verbal and non verbal intelligence) (for reviews, see Faust et
for activating the idiomatic meaning. al., 1999; Myerson et al., 2003).

Individual differences and speed of processing Speed of processing and sentence

In recent years, a renewed interest emerged on how comprehension
individual differences modulate spoken and written Most of the evidence on SOP concerns aged participants (for
language comprehension (e.g., Just & Carpenter, 1992; a review, see McCabe & Hartman, 2003). In the present
Gernsbacher & Faust, 1991; McNamara & McDaniel, 2004; study we instead investigated the effect of SOP on language
Hannon & Daneman, 2001; Faust et al., 1999). A specific comprehension in undergraduates.
aspect that has attracted a consistent empirical interest is Language-related SOP differences in non-aged speakers
individual speed of processing (henceforth SOP). This factor have been investigated especially at a word level, a
in fact predicts high level cognitive performances and plays battlefield for alternative models of reading and dyslexia
an important a role in language comprehension in aged and (for reviews, see Seidenberg & McClelland, 1989; Coltheart
non-aged participants. A long tradition of studies exists et al., 2001; Kello & Plaut, 2003). Less attention has been
investigating the speed with which an individual performs instead devoted to the role of SOP differences in modulating
mental operations and several models have been proposed to sentence comprehension in non-aged language unimpaired
explain the determinants and effects of SOP. In a nutshell, a participants. For instance, Breznitz and Berman (2003)
growing consensus exists that SOP might be a stable feature manipulated the reading rate of Hebrew and English non-
that differentiates individuals and groups: individuals would aged speakers systematically accelerating or decelerating the
posses a specific global speed of processing that remains silent reading rate of paragraphs that were followed by
constant across an extensive range of tasks of different comprehension questions. Surprisingly, when participants
nature and complexity (e.g., choice RT, line-length were asked to read faster than at their own self-paced rate,
discrimination, letter classification, mental rotation, visual less decoding errors occurred and the comprehension
search, word judgment) (e.g., Hale & Jansen, 1994; Vernon improved. On the contrary, deceleration improved the
& Jensen, 1984; Myerson et al., 2003; Faust et al., 1999; reading accuracy but impaired comprehension. McNamara
Salthouse, 1994). For each individual, it would be possible and McDaniel (2004) investigated the effect of domain
to define a speed parameter represented by the slope of the knowledge on the resolution of lexical ambiguity in
best fitting linear function of his/her mean response latencies participants with different reading skills and SOP.
(Faust et al., 1999). Participants read sentences ending with an ambiguous
Recently, Hanon and Daneman (2001) showed that the homograph and were asked to judge the relatedness of a
proportion of variance in reading comprehension visual target word that had a contextually irrelevant
performances of university students dramatically increased meaning. A wider relevance effect was found in less skilled
when an autonomous speed of response component was readers that also were slower in responding to the target

372
word (Experiment 1). The irrelevant meaning of the target and others finding it already at idiom offset (Titone &
interfered less for skilled readers with high general Connine, 1994; Tabossi et al., in press).
knowledge than for both skilled and less skilled participants
with low general knowledge (Experiment 3). The present study
We investigated the comprehension of idioms with an early
Idiomatic expressions: how are they vs. late point of identification (early identification vs. late
represented and understood? identification idioms) with a cross-modal lexical decision
Idiomatic expressions are phrasal units whose meaning paradigm. An idiom-related visual word was presented
generally cannot be derived from the meaning of the before the idiom verb (control position) or at idiom offset
constituent parts. Few studies investigated the role of (Tabossi & Zardon, 1993). A facilitation for the target at
individual differences in idiomatic meaning comprehension offset with respect to the control position indicates whether
in language-unimpaired adults (e.g., Johnson, 1991; Nippold is the idiom meaning activated. Most of the evidence
& Martin, 1989; Nippold, Moran & Schwarz, 2001). These collected so far concerns non ambiguous idioms. This study
results showed that idiom understanding iss associated with instead used strings having both a literal and an idiomatic
measures of intelligence, academic achievement in reading interpretation (ambiguous idioms, e.g., “break the ice”)
and listening comprehension, and with general non-verbal embedded in contexts either biasing the literal or the
measures of mental capacity. None of these studies directly idiomatic meaning of the string. Ambiguous idioms are well
addressed the role of individual SOP in on-line spoken suited for testing alternative predictions on both idiomatic
idiom comprehension. meaning retrieval processes and sentential ambiguity effects.
Idioms are very widespread in language, but the processes Specifically, we aimed at testing: a. whether the dominant
underlying their on-line comprehension are still meaning of an idiom (in our case, the figurative one) is
controversial. For “Lexical look-up” models (Bobrow & activated irrespective from contextual information or only
Bell, 1973; Swinney & Cutler, 1979; Gibbs, 1980), idioms when contextually appropriate (as in Colombo, 1995, 1998),
are multiword lexical units accessed as such from the mental b. the interaction between meaning dominance and idiom
lexicon. Their meaning is apprehended by direct memory identification.
retrieval and not elaborated via linguistic processing. The As recently argued by Hannon and Daneman (2001), the
fast nature of idiomatic meaning retrieval, often faster than presence of individual differences obscured by group data
corresponding literal sentences, is explained postulating an can be one of the causes of inconsistent findings in the
“horse race” between a full compositional analysis of the literature. Therefore, this study tested the extent to which
string, that requires longer reaction times, and the faster individual SOP differences modulated ambiguous idioms
retrieval of its global figurative meaning. This class of comprehension in language unimpaired undergraduates.
models differs in terms of representational assumptions: for This factor might account for some of the incongruencies
Bobrow and Bell (1973), idioms are stored in a separate found in the literature insofar as none of the existing on-line
idiom list in the mental lexicon whereas for Swinney and studies tested, to our knowledge, the effect of SOP.
Cutler (1979) they are represented as non-compositional Evidence exists that non-aged participants with faster SOP
units together with the other items. For “Non-Lexical” (henceforth, fast participants) process sentential information
models (Cacciari & Tabossi, 1988; Tabossi & Zardon, 1993, provided more actively than those with slower SOP
1995; Titone & Connine, 1994; Tabossi, Fanari & Wolf, in (henceforth, slow participants) (e.g., McNamara, 1997;
press), idioms are represented as configurations of lexical Long, Oppy & Seely, 1994; Long, Seely & Oppy, 1999).
items, with no duplication of lexical information and no Hence, we expect fast and slow participants to differ in
separate representation for idioms. The Configuration spoken idiom comprehension as well.
Hypothesis (Cacciari & Tabossi 1988) belongs to this
second class and posits that idiom meanings are associated Experiment 1
with configurations formed by the same words activated We divided our participants into fast and slow using the
during the comprehension of literal sentences. The distribution of the individual mean speed of response in the
idiomatic string is processed word-by-word, as any other cross-modal lexical decision task (see below). It is
piece of language (Peterson et al., 2001), until enough reasonable to assume that both fast and slow participants
information accumulates to render the string identifiable as a would recognize early identification idioms before offset.
memorized figurative string. Only at this moment is the Hence, a facilitation for the visual target with respect to
idiomatic meaning retrieved. Usually, the recognition of the control condition should be observed for early identification
idiomatic nature of the string depends on a specific part of idioms indicating an idiomatic meaning activation at offset.
the idiom that can be located early in the string (early If meaning dominance effects extend to idioms, we might
identification idioms), or later on (late identification idioms), expect to find activation in both literal and idiomatic
for instance at offset. Hence, the point in which the string is contexts. The picture is less clear for late identification
identified as idiomatic should determine how early is idioms for which the existing findings are contrasting. If
activated the idiomatic meaning. However, the evidence on their idiomatic meaning indeed requires time to be identified
idiom meaning activation especially for late identification and emerge (as in Cacciari & Tabossi, 1988; Tabossi &
idioms is still inconclusive with some studies showing that Zardon, 1993, 1995), then no activation should be observed
activation requires time occurring only after idiom offset at the string offset especially in literal contexts. The
(Cacciari & Tabossi, 1988; Tabossi & Zardon, 1993, 1995) activation of late identification idiom meaning might in fact
require an idiomatic context that increases the expectation

373
toward a figurative interpretation of the string (as in Tabossi non-word by pressing the appropriately labeled response
& Cacciari, 1988; Colombo, 1993, 1998). However, Titone keys.
and Connine (1994) and Tabossi et al. (in press), found that
the meaning of late identification non-ambiguous idioms Results and Discussion
was already available at offset in neutral contexts. If these 35 participants whose mean lexical decision times were
findings extend to ambiguous idioms, then we might find below the 45° percentile were considered as fast participants
activation of late identification idiom irrespective from the and 35 participants whose times exceeded the 55° percentile
contextual bias. If indeed fast participants outperform slow as slow. The statistical analyses were conducted on the log10
participants in processing sentential information (as in transformation of the raw lexical decision times. Table 1
McNamara, 1997; Long et al, 1994, 1999), then differential reports untransformed means. In order to reduce variability,
idiomatic meaning activation patterns should emerge for late data points ±2 SDs from the mean response time of each
identification idioms. participant were excluded from the analyses (1.9%).

Method Table 1: Mean lexical decision times for early identification

Participants. One-hundred forty-one undergraduates of the and late identification idioms, literal and idiomatic contexts,
University of Modena participated in the norming phase and P1/P2 (i.e., control/experimental target positions), and fast
eighty different undergraduates in Experiment 1. All were (F) and slow (S) participants.
Italian native speakers. Early Late
Materials. In the norming phase we assessed: a. the identification identification
familiarity and knowledge of the idiomatic meaning; b. the Literal Idiomatic Literal Idiomatic
idiom Identification point; c. the dominant vs. subordinate Context Context Context Context
nature of the idiomatic vs. literal meaning. We selected 24 P1/P2 P1/P2 P1/P2 P1/P2
familiar idiom strings with a dominant figurative meaning, F 858/829 849/839 850/849 847/836
12 were identifiable as idioms before offset (i.e., early S 1381/1370 1400/1378 1418/1378 1416/1363
identification) and 12 were not (i.e., late identification). The
visual target words were uniquely associated with the
idiomatic meaning of the string and were balanced for Fast and slow participants were equally accurate, as shown
number of characters and syllables, age of acquisition, by the same low mean percentage of errors (1.9% vs. 1.2%,
written frequency, bigram frequency and lexical decision respectively; t<1). Overall, the idiomatic meaning was
times in isolation. For each idiom string, two sentences of already activated at offset [Target: Fp(1,68)=28.81, p <.001;
similar length and structure were constructed (mean number Fi(1,22)=120.04, p <.001], a result consistent with Titone
of words 15.4) that biased toward the literal or the idiomatic and Connine (1994) and Tabossi et al. (in press) but not with
interpretation. In addition, 72 filler sentences without any Cacciari and Tabossi (1988). An effect of the idiom
idiom were prepared together with 48 legal non-words and identification point emerged with the targets associated with
24 words to serve as targets. Four experimental lists were early identification idioms responded to more rapidly than
created each containing 72 fillers, 12 early identification for late identification ones [Identification: Fp(1,68)=7.73, p
and 12 late identification idioms, half embedded in an = .007; Fi(1,22)=4.06, p = .05]. In literal contexts, the targets
idiomatic context and half in a literal one. Half of the targets associated with early identification idioms were responded
in each context condition appeared before the idiom string to faster than for late identification idioms suggesting that
and half at offset. Participants were randomly assigned to their meaning required more time to be retrieved [Context X
one of the four lists. Identification: Fp(1,68)=4.1, p <.05]. No difference was
Procedure. The sentences were digitally recorded by a male observed in idiomatic contexts.
speaker. A cue, inaudible to the participants, was placed in The time course of the meaning activation of early
the sound files to specify the position of the target and identification idiom was faster than that of late identification
caused the program to display it on the screen. In the idioms. This might reflect an earlier availability of the
experimental sentences the target appeared either at idiom idiomatic meaning of early identification idioms that
offset (experimental condition) or at the offset of the word speeded up the decision on the target at offset. Due to our
preceding the idiom (control condition) when no activation experimental paradigm, we did not know whether any
of the idiomatic meaning was expected. The target appeared difference exists in the amount of activation that early
in the center of the screen in black characters on a white identification vs. late identification idioms received during
background and remained on the screen until participant the string processing. That some difference might exist is
response; if no response was given in 4000 ms., the target indirectly suggested by Tabossi et al. (in press) who found
disappeared and the next trial followed. An interval of 900 that late identification idioms produced reliable priming
ms. followed the participant response before the subsequent effect only at the string offset but not earlier on, as instead
sentence begun. A recognition test on the sentences heard happened for early identification idioms.
during the experiment was performed at the end of the As to the SOP, fast and slow participants differed in
experiment to check that the participants actually listened general [Fp(1,68)=136.92, p<.001; Fi(1,22)=1543.52,
for comprehension. Participants were tested individually in a p<.001] and in the specific processing of both early
sound-attenuated room. Their task was to judge whether the identification and late identification idioms: whereas slow
visual target word corresponded to an Italian word or to a participants were faster with early identification than with

374
late identification idioms, no difference emerged for fast employed. At the beginning of the experiment, participants
participants [SOP X Identification: Fp(1,68)=4.17, p= .04; were advised that it was composed by two sessions and that
Fi(1,22)=5.53, p = .03]. Both fast and slow participants were in the second a response deadline was used. The instructions
faster when the targets were presented at offset than at were repeated at the beginning of each session. In Session
control position [SOP X Target: Fp(1,68)=3.78, p=.05]. The 2, a warning message appeared on the screen when the
idiomatic meaning of early identification idioms was deadline was over. Each new trial begun 900 ms after the
activated by both fast and slow participants. Despite what response (or the deadline end). The distribution of the
we expected, slow participants showed a full-fledged response times obtained in Session 1 was used to divide the
activation of the idiomatic meaning of late identification participants into fast vs. slow in both sessions.
idioms, whereas for fast participants the result only
approached significance (p=.07) [SOP X Target X Results and Discussion
Identification: Fp(1,68)=5.33, p=.02]. Why? One possibility Based on the results of Session 1, we split the participants
has to do with the slow response pace of slow participants: into fast and slow ones (with 37 participants per group) with
early identification and late identification idioms in fact the same procedure of Experiment 1. We present the results
differ in the amount of perceptual input required for the of the two sessions separately (Table 2). Session 1: A high
recognition of their idiomatic nature, with the first accuracy level was again observed with no fast vs. slow
recognized as idioms well before their offset (early difference (2.2% vs. 1.3% of errors respectively, t<1). The
identification point), and the second only when the last word idiomatic meaning of early identification idioms was already
has occurred (late identification point). The response pace of activated at offset, as in Experiment 1 [Target:
slow participants was so slow (see the 544 ms delay with Fp(1,72)=114.23, p<.001; Fi(1,11)=124.63, p<.001], by both
respect to fast ones) that the difference between early and fast and slow participants [SOP X Target: Fp(1,72)=10.02,
late identification idioms presumably was annulled: when p=.002; Fi(1,11)=25.64, p<.001]. The mean decision times
slow participants responded to the target associated with a of fast vs. slow participants again significantly differed
late identification idiom, enough perceptual input had [Fp(1,72)=166.75, p<.001; Fi(1,11)=403.79, p<.001].
already arrived to transform a late identification idiom into
an early identification one. Thereby Experiment 2 was Table 2: Mean decision times in literal and idiomatic
designed in which a response deadline was introduced. contexts, for P1/P2 (i.e., control vs. experimental positions),
fast (F) and slow (S) participants in Session 1 (early
Experiment 2 identification idioms) and Session 2 (late identification
Experiment 2 was divided into two sessions: in Session 1 idioms and response deadline).
only early identification idioms were used with the same Session 1 Session 2
experimental procedure of Experiment 1. This was Literal Idiomatic Literal Idiomatic
motivated by the need of clear cut time measures to be used Context Context Context Context
to divide participants into fast vs. slow (see below. In P1/P2 P1/P2 P1/P2 P1/P2
Session 2 only late identification idioms were employed and F 877/719 847/729 671/628 683/594
a response deadline introduced (see below). The presence of S 1416/1080 1467/1020 749/661 754/664
a response deadline should clarify the extent to which slow
participants indeed activated the idiomatic meaning of late
identification idioms, or whether they simply annulled the Session 2: The idiomatic meaning of late identification
difference between early vs. late identification due to their idioms was already activated at offset, as in Experiment 1
slowed response rate. Breznitz and Berman (2003) recently [Target: Fp(1,72)=80.306, p<.001; Fi(1,11)=82.170, p<.001]
documented a beneficial effect on sentence comprehension in both types of context [Context X Target: Fp(1,72)=2.87,
of a reading rate acceleration for slow readers, normal p=.094; Fi(1,11)=5.65, p=.03]. Despite the deadline, fast and
readers and dyslexic. Hence, we might expecta general slow participants overall significantly differed [105 ms,
speeding up of idiomatic meaning activation. Fp(1,72)=19.77, p<.001; Fi(1,11)=51.4, p<.001], although to
a lesser extent (105 ms) than in Session 1 (452 ms) and in
Method Experiment 1 (544 ms). No specific effect of SOP emerged.
Participants. Eighty-five undergraduates of the University However, if the time pressure speeded up the lexical
of Modena participated in Experiment 2. They were native decision on idiom-related targets fostering idiomatic
speakers of Italian and had not participate in any phase of meaning activation, it also had an accuracy cost as shown by
Experiment 1. the higher mean error rate of Session 2 (Session 2: 5%;
Materials and Procedure. The experimental materials were Session 1: 1.8%), with more errors in literal than in
those used in Experiment 1 but presented in two sessions idiomatic contexts [Fp(1,72)=3.9, p=.05; Fi(1,11)=4.14,
divided by a short rest. In Session 1 only early identification p=.06]. Slow participants were as much accurate as fast ones
idioms (plus appropriate fillers) were used with the same (5.4% vs. 4.7%, respectively). Unsurprisingly, slow
procedure of Experiment 1. In Session 2 only late participants gave more post-deadline responses than fast
identification idioms (plus fillers) were employed and a ones [3.7% vs. 2.9%, respectively; Fp(1,72)=3.59, p=.06].
response deadline introduced whose length (1007 ms) More post-deadline responses were provided by both slow
resulted from summing up the mean decision time of fast and fast participants when the targets were presented at
participants in Experiment 1 plus 1.5 SD. The same cross- control position than at idiom offset [Fp(1,72)=17.03,
modal lexical decision paradigm of Session 1 was

375
p<.0001; Fi(1,11)=17.74, p=.001], and more so in idiomatic meaning. In contrast, fast participants processed sentential
than in literal contexts [Context X Target: Fp(1,72)=5.07, information more actively, consistently with the literature.
p=.03]. The response deadline introduced in Experiment 2
Under time pressure, a recruitment of attentional and affected both groups of participants with a more striking
processing resources speeded up the response times of both effect on slow ones. With a deadline, the magnitude of the
fast and slow participants. The response deadline accelerated facilitation for the idiomatic target was in fact quite similar
slow participants more than fast ones, as suggested by the for fast and slow participants (66 ms vs. 89 ms,
583 ms mean between-session time difference for slow respectively). That imposing a response deadline speeded up
participants (i.e., a decrease of 46.8%) as compared with the mental processing and changed the allocation of time and
149 ms difference for fast participants (i.e., 19%). processing resources among different task components had
already been documented in the literature on judgment and
Conclusion decision-making (for a review, see Svenson & Maule, 1993)
The idiomatic meaning of ambiguous spoken idioms was and on reading comprehension (Breznitz & Berman, 2003).
activated at offset for both early and late identification According to Breznitz and Bearman (2003), who
idioms, as in Titone and Connine (1994) and Tabossi et al. extensively investigated this “acceleration phenomenon”,
(in press) but not in Cacciari and Tabossi (1988) and reading acceleration extends attention span, reduces
Colombo (1993, 1998). Then the effects of context of the distractibility, helps to overcome some capacity limitations
present study are certainly less clear-cut than those reported of short term memory and of working memory and increases
in Tabossi and Cacciari (1988) and Colombo (1993, 1998). word retrieval from the mental lexicon. Identifying the
This can at least in part be due to a meaning dominance factors underlying the differences between fast and slow
effect analogous to that reported in studies on semantic participants is the major problem we are currently
ambiguities: we in fact selected idiom strings with a investigating in a set of experiments designed to assess the
dominant figurative meaning. specific role of sensorimotor speed components, cognitive
How to reconcile these different findings? We can discard speed components and personality-based components in
an idiom familiarity effect since the expressions employed accounting for SOP differences in spoken idiom
in these studies were all familiar. More sound candidates for comprehension.
explaining these discrepancies lie in methodological Overall, the results of this study are more compatible with
differences. Titone and Connine, Cacciari and Tabossi, Non-Lexical models of idiom comprehension than with
Tabossi et al. used a cross-modal lexical decision paradigm Lexical-look up ones. In fact we concurred with other
and Colombo a self-paced reading times paradigm. Then, studies in showing that idioms do not behave as long words,
the cross-modal studies employed different control since their meaning activation requires time and can start
conditions: here we compared the lexical decision time for a only after the string is identified as idiomatic. The SOP
same target word presented either before the idiom, where effects we found on idiomatic meaning activation are best
no facilitation at all is expected, or at the idiom offset. Such accommodated by Non-Lexical models that posit that the
a modality can be more sensitive toward smaller amount of idiomatic sentence is fully processed up to recognition of its
activation than when an idiom-related target is compared figurative nature (and syntactically, even later on). Only
with an idiom-unrelated word, as in Cacciari and Tabossi, then can the idiomatic meaning be retrieved. In contrast,
for instance. Idioms also vary in the extent to which the Lexical-look up models assume direct retrieval of an idiom
strings also have a literal meaning: when an idiom string has meaning, a cognition operation that might prove less
two possible interpretations they rarely are equally frequent sensitive toward SOP differences being less demanding even
with the dominance of the idiomatic meaning affecting for slow participants. In sum, idiom comprehension cannot
meaning activation (Popiel & McRae, 1988). A word should be reduced to mere meaning retrieval from the mental
then be said on the notion of identification point (or lexicon.
predictability) that has been differently operationalized.
Furthermore the decision on an idiom identification point Acknowledgments
relies on off-line, paper-and-pencil measures whose The present study was funded by a grant COFIN 2003
predictive effect on on-line measures should be further (2003119330_005) to the first author.
assessed.
Last, and not at all least, we took individual differences References
into serious account. Did SOP differences modulate spoken Breznitz, Z. & Berman, L. (2003). The underlying factors
idiom comprehension? Although further investigations are of word reading rate. Educational Psychology Review, 15,
needed (and are in progress), we observed a clear-cut SOP 3, 247-265.
effect. Specifically, in Experiment 1, slow participants were Bobrow, S. & Bell, B. (1973). On catching on to idiomatic
more sensitive to the difference between early identification expressions. Memory & Cognition, 1, 343-346.
and late identification idioms than fast ones, and processed Cacciari, C. & Tabossi, P. (1988). The comprehension of
the first faster than the second. Spoken idiom identification idioms. Journal of Memory and Language, 27, 668-683.
occurs word-by-word and is successfully achieved when Cacciari, C. & Tabossi, P. (in prep.). Factors affecting
enough information is build up to render an idiom idiomatic meaning time course: the role of idiom length
recognizable: only then can its meaning be retrieved. Slow and context.
participants needed more perceptual input than fast Colombo, L. (1993). The comprehension of ambiguous
participants to identify a string as idiomatic and activate its idioms in context. In Cacciari, C., Tabossi, P. (Eds.),

376
 Idioms. Processing, Structure and Interpretation (pp. 3- Myerson, J., Hale, S., Zheng, Y., Jenkins, L. & Widaman,
26). Hillsdale, NJ: Laurence Erlbaum. K. F. (2003). The difference engine: a model of diversity
Colombo, L. (1998). Role of context in the comprehension in speeded cognition. Psychonomic Bulletin & Review, 10,
of ambiguous Italian idioms. In D. Hillert (Ed.), Sentence 2, 262-288.
processing: a cross-linguistic perspective. Syntax and Nippold, M. A., & Martin, S. T. (1989). Idiom interpretation
Semantics, vol. 31. (pp. 405-425). New York: Academic in isolation versus context. A developmental study with
Press. adolescents. Journal of Speech and Hearing Research, 32,
Coltheart, M., Rastle, K., Perry, C., Langdon, R. & Ziegler, 59-66.
J. (2001). DRC: a dual-route cascaded model of visual Nippold, M. A., Moran, C., & Schwarz, I. E. (2001). Idiom
word recognition and reading aloud. Psychological understanding in preadolescents: Synergy in action.
Review, 108, 204-256. American Journal of Speech-Language Pathology, 10, 2,
Faust, M. E., Balota, D. A., Spieler, D. H. & Ferraro, F. R. 169-180.
(1999). Individual differences in information-processing Peterson, R. R., Burgess, C., Dell, G. S., & Eberhard, K.L.
rate and amount: implications for group differences in (2001). Dissociation between syntactic and semantic
response latency. Psychological Bulletin, 125, 6, 777-799. processing during idiom comprehension. Journal of
Gernsbacher, M.A. & Faust, M. (1991). The mechanism of Experimental Psychology: Learning, Memory and
suppression: a component of general comprehension skill. Cognition, 90, 227-234.
Journal of Experimental Psychology: Learning, Memory Popiel, S. J., & McRae, K. (1988). The figurative and literal
and Cognition, 17, 245-262. senses of idioms, or all idioms are not used equally.
Gibbs, R.B. (1980). Spilling the beans on understanding and Journal of Psycholinguistic Research, 17, 475-487.
memory for idioms. Memory & Cognition, 8, 449-456. Salthouse, T. A. (1994). The nature of the influence of speed
Hale, S. & Jensen, J. (1994). Global processing-time on adult age differences in cognition. Developmental
coefficients characterize individual differences in Psychology, 30, 2, 240-259.
cognitive speed. Psychological Science, 5, 384-389. Seidenberg, M. S. & McClelland (1989). A distributed
Hannon, B. & Daneman, M. (2001). A new tool for developmental model of word recognition and naming.
measuring and understanding individual differences in the Psychological Review, 96, 523-568.
component processes of reading comprehension. Journal Svenson, O. & Maule, A. J. (1993). (Eds.), Time pressure
of Educational Psychology, 93, 1, 103–128. and stress in human judgment and decision making. New
Johnson, J. (1991). Developmental versus language-based York: Plenum Press.
factors in metaphor interpretation. Journal of Educational Swinney, D. & Cutler, A. (1979). The access and processing
Psychology, 83, 4, 470–483. of idiomatic expression. Journal of Verbal Learning and
Just, M. A. & Carpenter, P. A. (1992). A capacity theory of Verbal Behaviour, 18, 523-534.
comprehension: individual differences in working memory. Tabossi, P. & Cacciari, C. (1988). Context effects in the
Psychological Review, 99, 122-149. comprehension of idioms. Proceedings of the Tenth
Kello, C. T. & Plaut, D. C. (2003). Strategic control over rate Annual Conference of the Cognitive Science Society
of processing in word reading: a computational (pp.90-96). Hillsdale, NJ: Erlbaum.
investigation. Journal of Memory and Language, 48, 207- Tabossi, P., Fanari, R. & and Wolf, K. (in press). Spoken
232. idiom recognition: meaning retrieval and word
Long, D. L., Oppy, B. J. & Seely, M. R. (1994). Individual expectancy. Journal of Psycholinguistic Research.
differences in the time course of inferential processing. Tabossi, P., Zardon, F. (1993). The activation of idiomatic
Journal of Experimental Psychology: Learning, Memory, meaning in spoken language. In C. Cacciari & P. Tabossi
and Cognition, 20, 1456-1470. (Eds.). Idioms. Processing, Structure and Interpretation
Long, D. L., Seely, M. R. & Oppy, B. J. (1999). The (pp. 145-162). Hillsdale, NJ: Laurence Erlbaum.
strategic nature of less skilled readers’ suppression Tabossi, P., & Zardon, F. (1995). The activation of
problem. Discourse Processes, 27, 281-302. idiomatic meaning. In M. Everaert, E. van den Linden,
McCabe, J., & Hartman, M. (2003). Examining the locus of A. Schenk, & R. Schreuder (Eds.), Idioms: structural and
age effects on complex span tasks. Psychology and Aging, psychological perspectives (pp. 273-282). Hillsdale, NJ:
18, 3, 562-572. Erlbaum.
McNamara, D. (1997). Comprehension skills: a knowledge- Titone, D.A. & Connine, C.N. (1994). Comprehension of
based account. Proceedings of the Nineteenth Annual idiomatic expressions: Effect of predictability and
Conference of the Cognitive Science Society (pp. 508- literality. Journal of Experimental Psychology: Learning,
513). Hillsdale, NJ: Erlbaum. Memory, and Cognition, 20, 5, 1126-1138.
McNamara, D & McDaniel, M.A. (2004). Suppressing Vernon, P. A. & Jensen, A. R. (1984). Individual and group
irrelevant information: knowledge activation or differences in intelligence and speed of information
inhibition? Journal of Experimental Psychology: processing. Personality and Individual Differences, 5,
Learning, Memory, and Cognition, 30, 2, 465-482. 411-423.

377