NOTE
THE SEMANTIC ORGANISATION OF MASS NOUNS:
EVIDENCE FROM SEMANTIC REFRACTORY ACCESS DYSPHASIA
Sebastian J. Crutch and Elizabeth K. Warrington
(Dementia Research Centre, Department of Neurodegeneration, Institute of Neurology, University College, London, UK)

ABSTRACT
The conceptual properties and organisation of mass noun concepts are examined in a patient (A.Z.) with a semantic
refractory access disorder. A cardinal feature of this class of disorder is the build up of interference (refractoriness) between
semantically similar concepts. In a series of written word identification tasks, a gradient of semantic relatedness is
demonstrated within the broad domain of mass noun concepts but not between different word classes. More detailed
examinations of semantic similarity effects within subcategories of mass concepts reveal a variable degree of fine-grain
organisation, which may reflect the influence of premorbid personal knowledge and experience. Furthermore, it is
demonstrated that common mass or count noun status did not lead to an enhancement of the refractory effects of semantic
similarity.
Key words: mass/count nouns, comprehension, stroke, semantic refractory access dysphasia

INTRODUCTION
In recent years, the study of categorical aphasic
deficits arising from neurological damage has been
strongly biased towards trying to understand the
organisation of conceptual knowledge relating to
concrete objects such as animals, plants and manmade artefacts (for a review see Capitani et al.,
2003). However, other categorical distinctions that
are potentially of equal importance have received
far less investigative attention, such as the division
between mass nouns and count nouns.
Semantically, mass nouns are those items which
have no firm perceptual boundaries and which in
combination constitute a larger version of the same
substance (e.g., water, wood, coal), whilst count
nouns are those items which do have a perceptual
form but which in combination do not represent the
same entity (e.g., cat, sock, pear). Grammatically,
mass nouns cannot be pluralized, preceded by an
indefinite article or counted, whilst count nouns
can be pluralized, do take both the definite and
indefinite article and can be counted.
Under such definitions, it can be seen that the
majority of semantic category specific disorders
reported concern items falling within the domain of
count nouns. However, within this literature some
attention has been given to so-called sensoryquality items which include mass terms (e.g.,
materials, liquids, edible substances) as well as
count terms (e.g., animals). Proponents of the
traditional sensory/functional theory of category
specificity have used evidence of concomitant
deficits across sensory-quality items relative to
spared comprehension and naming of artefact items
to argue that sensory rather than functional
Cortex, (2007) 43, 1057-1067

information is central to the conceptual
representation of both mass and count-type
sensory-quality items (Borgo and Shallice, 2001,
2003). Irrespective of this particular theoretical
interpretation however, such fine-grain category
specific effects may suggest that either mass and
count nouns have separable neural bases or that the
mass/count distinction has little relevance at the
level of conceptual knowledge.
Evidence of dissociations between mass nouns
and count nouns has been reported at the level of
syntactic, and arguably lexical semantic, processing
(e.g., Semenza et al., 1997, 2000; Mondini et al.,
2004; Taler et al., 2004). For example, Semenza et
al. (1997) have reported a patient who, in the
context of otherwise intact grammar, showed a
selective deficit for the use of the grammatical
properties of mass nouns. This deficit was
demonstrated on a range of tasks including
sentence completion, sentence production and
judgements on the grammatical correctness of
sentences, but notably not on tasks such as naming
to definition, semantic judgement or semantic
association. Reports of this type could be taken as
supporting the view of Bunt (1976) that “the
mass/count distinction is not really a distinction
between words, but a distinction between ways of
using the words”. Nonetheless, there remains
debate over whether the information underlying the
differences between mass and count nouns is
primarily semantic (e.g., Jackendoff, 1991;
Wisniewski et al., 1996) or syntactic in nature
(e.g., Gillon, 1992; Vigliocco et al., 1999). It
should be noted that this debate is not exclusive,
with many linguists holding that the distinction has
both semantic and syntactic bases. However, the

1058

Sebastian J. Crutch and Elizabeth K. Warrington

role of “semantic” information in this sense
corresponds to lexical semantic information rather
than purely conceptual information. In many
information processing models, this distinction
between lexical semantic and general semantic
information corresponds to the distinction between
the semantic component of a lemma level
representation and a more general knowledge base,
which lies prior to lemma level representations.
Despite the fact that the debate concerning the
count/mass nouns primarily has its roots in the
psycholinguistic rather than conceptual knowledge
literature, it remains of interest to examine whether
some essence of the mass/count distinction is
represented at the conceptual level, rather than
simply presume it unlikely. Such an examination
would be justified and of interest for several
reasons. First, a number of authors have postulated
that the mass/count distinction is represented not
only at the syntactic level but also at the conceptual
level; early in development, children are held to be
able to distinguish solid, shape-invariant items from
the materials from which they are constituted, and
moreover to know that different properties define
object and substance categories (Soja et al., 1991;
Soja, 1992). However, it is not known whether such
early conceptual differentiation may persist and
constitute a significant principle of conceptual
organisation in the adult brain. Secondly, not all
researchers accept the notion of separate conceptual
and lexical semantic representations (e.g.,
Caramazza, 1997; Cipolotti, 2000). Thus, any
evidence of a semantic basis to what many consider
to be a syntactic distinction would have
implications for theories of language production.
In this report, an attempt is made to address
some of these issues by examining the ability of a
patient (A.Z.) with a semantic refractory access
disorder to comprehend and identify mass and
count nouns. Semantic refractory access disorders
are characterised by a sensitivity to temporal
factors, with the conceptual system requiring time
to regain a resting level of activation following use
(Warrington and McCarthy, 1983, 1987). Semantic
distance effects are also commonly observed,
which are attributed to the effect of refractoriness
not only upon the subsequent re-activation of a
given representation but also other representations
that share semantic space (Crutch and Warrington,
2003, 2004). Importantly, semantic distance effects
can be observed both on word-picture matching
tasks and on tasks conducted entirely within the
verbal domain, namely spoken word-written word
matching. For example, the current patient, A.Z.,
has previously shown significantly higher error
rates identifying written brand names presented in
semantically related arrays (e.g., same product
type: Vauxhall, Rover, Honda, Ford) than when the
same items were presented in semantically
unrelated arrays (e.g., Vauxhall, Asda, Nestle,
Sony) (Crutch and Warrington, 2004). Thus, such

refractory access patients provide the opportunity
to investigate fine-grain aspects of conceptual
organisation and, in the current study, to examine
the spread of refractoriness within syntactic and
mass noun categories. Furthermore, we consider
whether refractoriness is enhanced by common
mass/count noun status.
CASE DESCRIPTION
The patient described, A.Z., is a 72-year-old
housewife who was admitted to Northwick Park
Hospital in January 2002 following a left middle
cerebral artery stroke. The prominent presenting
feature was a severe expressive and mild receptive
dysphasia. Her spoken output was characterised by
a fluent semantic and phonemic jargon. A CT scan
conducted at the time of admission revealed
damage to a large portion of the left middle
cerebral territory, sparing the basal ganglia. The
left parietal, temporal and posterior frontal regions
were all affected.
Following the vascular incident, A.Z. became
almost entirely anomic, naming only two items
from the Boston Naming Test. A.Z. also exhibited
a profound dyslexia and dysgraphia, reading
correctly only 1/99 items taken from the Patterson
and Hodges (1992) corpus. Word repetition skills
were somewhat better preserved, correctly
repeating 39/99 items from the same word pool.
Auditory lexical decision performance was also
relatively well preserved (71/80; PALPA 5) (Kay et
al., 1992). Comprehension was assessed on the
Pyramids and Palm Trees test for both word-word
matching and picture-picture matching, and A.Z.
showed a moderate impairment on both forms of
the task (38/52 and 38/52, respectively).
Furthermore, A.Z. gained comparable scores on the
spoken word-picture and written word-picture
matching versions of the British Picture Vocabulary
Scale (22/32 and 21/32, respectively).
A.Z.’s comprehension performance was also
observed to be both inconsistent and sensitive to
temporal factors. These aspects of her
comprehension deficit have been the focus of
several previous investigations (Crutch and
Warrington, 2003, 2004; Warrington and Crutch,
2004) in which it has been argued that A.Z.
exhibits a semantic refractory access disorder.
Central to our claims that A.Z.’s refractory access
deficit was localised to a semantic and not
presemantic level of processing was an experiment
reported in Warrington and Crutch (2004). Pictures
of 69 items were arranged into both 23
semantically related triplet arrays (e.g., goat, bear,
deer) and 23 semantically unrelated but
phonologically related (rhyming) triplet arrays
(e.g., goat, boat, coat). The tasks was administered
as a spoken word-picture matching task with items
in all arrays being probed three times in a

Semantic organisation of mass nouns

pseudorandom order. Despite the phonological
similarity of items in the semantically unrelated
condition, A.Z.’s performance was significantly
more accurate than in the semantically related
condition. This result was interpreted as
demonstrating that at least a significant proportion
of A.Z.’s refractory deficit could be localised to a
semantic stage of information processing. This
interpretation is particularly relevant in the present
paper as the purpose of Experiment 1 was to
constrain further our hypothesising about the
information
processing
stage
at
which
refractoriness is seen to act in this patient.
EXPERIMENTAL INVESTIGATIONS
Experiment 1
The Locus of the Refractory Access Deficit:
A Parts of Speech Effect?
As noted in the Case Description, previous
reports of the current patient have provided
empirical evidence that the refractory access deficit
has a semantic and not a pre-semantic locus. The
aim of this experiment was to localise further this
deficit, and in particular to attempt to establish
whether syntactic information or only conceptual
information influences the build up of
refractoriness when this patient attempts spoken
word-written word matching tasks.
Procedure
The stimuli consisted of 32 words: 8 nouns,
8 adjectives, 8 adverbs and 8 verbs (two of
which could unfortunately also be classified
as nouns). These stimuli were arranged into 8
arrays each containing four words from the same
part of speech (e.g., all nouns). The same stimuli
were also arranged into 8 arrays each containing
four words which took a different part of speech
(i.e., 1 nouns, 1 adjective, 1 adverb and 1 verb)
(see Appendix I). With only very minor exceptions,
the four words within each array had little or
no semantic relationship between them. The task
was presented as a spoken word-written word
matching procedure, with the identity of every
word in each array being probed four times in a
pseudorandom order. A natural presentation rate of
a 1 sec response-stimulus interval (RSI) was
adopted. The same and different parts of speech
arrays were administered in a counterbalanced
design.
Results
A.Z. showed no difference in response accuracy
between the same part of speech and different parts
of speech conditions (same = 97/128, different =
94/128; χ2 = .19, p > .6, df = 1).

1059

Comment
The absence of an effect of parts of speech
upon A.Z.’s accuracy in identifying stimulus words
in this experiment provides empirical evidence
which is useful with regard to establishing the
locus of the refractory disorder in information
processing terms. If the refractory disorder was
affecting a level of processing at which information
about the syntactic status of words was
represented, then one would predict an effect of
syntactic similarity akin to the semantic similarity
effects which are observed so readily in this
patient. The absence of such a syntactic similarity
effect suggests that this type of information is not
represented sufficiently strongly at the level at
which refractoriness is occurring for a difference in
the size of refractory effect to be observed.
Although this negative result does not rule out the
possibility of some refractoriness occurring among
either pure syntactic or lemma level
representations, the evidence is most compatible
with a predominant localisation of the refractory
deficit to the central conceptual system.
Experiment 2
Multiple Levels of Semantic Distance within
the Mass Noun Domain
The influence of semantic relatedness upon the
comprehension abilities of semantic refractory
access patient A.Z. has previously been explored in
the broad domains of common count nouns and
proper nouns. However, semantic distance effects
contingent upon refractoriness have not been
examined or observed in the broad domain of mass
nouns. In this experiment, the semantic organisation
of a range of mass noun concepts was probed.
Procedure
The test stimuli consisted of 27 written mass
nouns which were arranged into 9 semantically
close arrays each containing three items drawn
from the same category (e.g., gases, liquids, solids,
meats, carbohydrates, alcoholic drinks, metals,
stones, jewels). The same items were also arranged
into 9 semantically intermediate triplet arrays
containing either 3 chemicals (i.e., 1 gas, 1 liquid,
1 solid), 3 foodstuffs (i.e., 1 meat, 1 carbohydrate,
1 alcoholic drink) or 3 minerals (i.e., 1 metal, 1
stone, 1 jewel). Finally, the items were arranged
into 9 semantically distant arrays each consisting of
three items drawn from a different broad semantic
category (e.g., 1 chemical, 1 foodstuff, 1 mineral).
A full list of stimuli is provided in Appendix II.
The experiment was conducted as a spoken
word-written word matching task, with the patient
being requested to point to target written words in
response to the examiner’s presentation of
corresponding spoken words. For each array, a

1060

Sebastian J. Crutch and Elizabeth K. Warrington

Fig. 1 – Performance on the variable semantic distance of mass nouns task (Experiment 2). (A) The percentage correct scores for the
semantically close, intermediate and distant arrays. (B) The number of errors made on consecutive probes of the test stimuli, both for
each individual semantic distance condition and the combined total error score.

repetitive testing procedure was adopted such that
every item was probed four times in a
pseudorandom order. Thus there were 12 responses
per 3-item array. A natural presentation rate was
used with a 1 sec pause between each pointing
response and the presentation of the subsequent
spoken word target (RSI). The arrays for each of
the three conditions were presented in a serial ABC
ABC design, and were ordered so that no
individual word appeared in consecutive arrays.

calculated and is shown in Figure 1B. Cochran Q
tests demonstrated a significant difference in
response accuracy between probes for the
semantically intermediate arrays (Q = 13.0, p <
.01, df = 3) which reflects the gradual increase in
the error rate across successive probes. However,
such an effect was not observed with either
semantically close or semantically distant arrays (Q
= 3.36, p > .3 and q = 2.0, p > .5 respectively; df
= 3 both).

Results

Comment

The percentage correct responses for the three
conditions are shown in Figure 1A. A Friedman
test showed a significant difference in the number
of errors made in the three conditions (Xr2 = 7.18,
p < .05, df = 2). Referring to Figure 1A, it can be
observed that this significant between-conditions
effect reflects a deterioration in A.Z.’s performance
level the closer the semantic relationship between
stimuli within an array. The number of errors
committed on each of the four serial probes was

These results indicate that there is a gradient of
semantic relatedness within the broad domain of
mass nouns. This gradient can be attributed to a
greater degree of refractoriness among mass nouns
which have a strong semantic relationship than
among those which do not. Extraneous variables
such as word frequency and age of acquisition
cannot account for the semantic effect because
exactly the same words were used in each condition,
merely arranged into different combinations.

Semantic organisation of mass nouns

Experiment 3
Semantic Relatedness Effects among Categories
of Mass Nouns
The semantic relatedness effect observed in the
previous experiment provides evidence of a broad
level of categorical organisation among the
semantic representations of mass noun concepts.
However, this observation was partly based on a
comparison of A.Z.’s ability to identify items from
the same narrow category with items drawn from
across very different categories (i.e., the
‘semantically distant’ arrays were in fact ‘very
distant’). In this experiment, an attempt was made
to identify finer-grain principles of organisation
within particular semantic categories.
Procedure
The experiment was conducted in three separate
parts using three different sets of experimental
stimuli:
Set 1: 16 food names, 4 drawn from each of
four classes of foods, namely meats (e.g., pork,
lamb, beef, veal), cheeses, desserts and sauces (see
Appendix III).
Set 2: 16 drink names, 4 drawn from each of
four types of drink, namely long alcoholic drinks
(e.g., lager, beer, cider, shandy), spirits, hot drinks
and mixers (see Appendix IV).
Set 3: 16 names of non-edible solids, 4 drawn
from each of four classes of solids, namely metals
(e.g., iron, copper, steel, lead), stones, heavy
fabrics and light fabrics (see Appendix V).
Each set of words was arranged into four
semantically close written word arrays. The same
stimulus words were also arranged into four
semantically distant arrays containing one item from
each of the four semantically close arrays within
that set (e.g., distant foods: pork, stilton, custard,
vinegar) (see Appendices III-V). A spoken wordwritten word matching paradigm was employed for
each set of words. For each array, all the items were
probed four times in a pseudorandom order. A 1 sec
RSI was adopted, and within each set, close and
distant arrays were presented in an ABAB BABA
design. The three parts of the experiment were
conducted separately, with the food and drink sets
being administered on the same day and the nonedible solids four weeks later.
Results
The percentage correct responses in the
semantically close and semantically distant
conditions of the food, drink and non-edible solid
experiments are shown in the leftmost portions of
Figures 2A-2C. Wilcoxon matched-pairs rankedsums tests revealed A.Z. was significantly less
accurate in identifying foods when their names
were presented in semantically close rather than

1061

semantically distant arrays (z = 2.80, p < .01).
However, no such effects of semantic similarity
were observed in the identification of the names of
drinks or non-edible solids (z = .40, p > .6 and z =
1.08, p > .2). Serial position curves showing the
percentage error rate on each consecutive probe of
the stimulus items in each set are also presented in
the right-hand portion of Figures 2A-2C. Cochran
Q tests were employed to demonstrate a significant
difference in the number of errors committed on
each probe of the food names in the semantically
close array condition (Q = 12.24, p < .01, df = 3).
Reference to Figure 2A reveals that this difference
corresponded to an increase in the error rate
following initially accurate performance on the first
probe. No such between-probe difference was
observed when the same food names were probed
under distant array conditions (Q = 3.00, p > .3, df
= 3). By contrast, in the drink and non-edible solid
experiments, significant differences between the
number of errors committed on each probe were
observed both with semantically close arrays (Q =
8.23, p < .05, df = 3 and Q = 12.00, p < .01, df =
3, respectively) and with semantically distant
arrays (Q = 13.15, p < .01, df = 3 and Q = 10.86,
p < .02, df = 3, respectively). In all cases, these
differences reflected a gradual increase in error rate
from initially accurate performance on the first
probe to more inaccurate performance on
subsequent probes.
Comment
The observation of a semantic distance effect in
A.Z.’s identification of food names provides
empirical support for the role of semantic similarity
in the fine-grain organisation of mass nouns.
However, the absence of equivalent effects in the
categories of drinks and non-edible solids despite
overall error rates which are comparable to that
observed in the foods experiment suggests that, for
this individual patient, the level of organisation is
not identical in every semantic domain. More
specifically, greater personal knowledge of domain
as compared with another may have motivated
differential patterns of semantic organisation. Thus
the ability of the semantic distance paradigm to
demonstrate fine-grain category specific effects
may be limited to areas of semantic cognition in
which similarity is particularly strongly
represented. To some extent, the strength of this
representation will be individually determined,
depending on the depth of an individual’s
knowledge base.
Experiment 4
Effect of Shared Mass/Noun Status upon
Refractoriness Effects
It has been demonstrated previously that at least
a major component of the refractory access deficit

1062

Sebastian J. Crutch and Elizabeth K. Warrington

Fig. 2 – The percentage correct responses (histogram) and number of errors on successive probes (line graph) are shown for
semantically close and distant arrays of (A) food mass nouns, (B) drink mass nouns, and (C) non-edible solid nouns (Experiment 3).

exhibited by A.Z. lies at the level of conceptual
processing (Experiment 1 and Crutch and
Warrington, 2004). The current observations of
semantic distance effects add further weight to this
position. It is in this context that A.Z. offers the
opportunity to examine whether the distinction

between mass nouns and count nouns is
represented at the conceptual level, syntactic level
or partly at both stages of processing. More
specifically, we examine among semantically
similar items, is refractoriness enhanced if
mass/count noun status is shared?

1063

Semantic organisation of mass nouns

Procedure
The stimuli consisted of 8 count nouns and 8
mass nouns. These items were arranged into stimulus
pairs under three conditions (see Appendix VI):
(i) Semantically close arrays – Same noun type:
this condition yielded four 4 two-item arrays
containing two conceptually similar count nouns
(e.g., cloud, storm) and 4 arrays containing two
conceptually similar mass nouns (e.g., rain, snow).
(ii) Semantically close arrays – Different noun
type: 8 arrays containing two items which are
conceptually similar but which differ in their
mass/count status (e.g., cloud, snow).
(iii) Semantically distant arrays – Mixed noun
type: 8 arrays containing two semantically dissimilar
items. Half the arrays contained items with the same
mass/count status (e.g., snow, pork) and half items
with a different status (e.g., branch, thunder).
The task was administered using a spoken
word-written word matching paradigm. For every
pair of items, the identity of each word was probed
five times in a pseudo random order (i.e., a total of
10 responses per item pair). In order to ensure that
the subject did not use the simplicity of the twoitem written word array to associate a particular
response simply with pointing to the left or right,
the word array was altered throughout the
experiment so that for half of the responses a
particular written word appeared on the left whilst
on the remaining half of responses the same word
appeared on the right. The location of the target
was designed such that the correct response was to
point to the left equally as often as to the right. A
1 sec RSI was used, and arrays from the three
conditions were administered in a counterbalanced
design, ensuring that no stimulus item appeared in
consecutive arrays.
Results
The percentage correct responses under the
three experimental conditions are shown in Figure
3A. A series of pairwise Wilcoxon matched-pairs
signed-ranks tests revealed that A.Z. made
significantly more errors on both the semantically
close/same noun type and semantically
close/different noun type arrays than on the
semantically distant arrays (z = 2.08, p < .05 and z
= 2.40, p < .02, respectively). However, looking at
the two types of semantically close arrays, there
was no difference in response accuracy whether or
not the stimuli were drawn from the same or
different class of nouns (z = .29, p > .7). The serial
position curves describing the percentage error
rates for each probe of the test stimuli are shown
in Figure 3B. As with Experiments 2 and 3, A.Z.’s
performance is generally characterised by a
deterioration in accuracy with repeated probing of
an item’s identity. It is of note that in this case,
serial position effects are observed with both the

semantically close arrays and semantically distant
arrays; we would attribute the latter effect to the
fact that in a two-item array, the same stimuli are
probed five times in rapid succession.
Comment
A.Z.’s accuracy in identifying both mass and
count nouns appears to be significantly affected by
the semantic similarity of the target item to the
distractor item. However, the degree of mass/count
status similarity between target and distractor seems
to have no effect upon the probability of an accurate
response. In A.Z. the refractoriness observed
between semantically similar items does not appear
to be enhanced by shared mass/count noun status.
DISCUSSION
In this paper, we report a series of experiments
conducted with a patient with semantic refractory
access dysphasia which provide information about
the fine-grain organisation of conceptual
knowledge relating to mass nouns and parts of
speech. In particular, it has been shown that there
is no gradient of semantic relatedness across
different parts of speech whereas there is a gradient
of semantic relatedness across the broad mass noun
domain. This gradient is considered to reflect the
build up of refractoriness leading to a higher error
rate when identifying targets under conditions of
semantically similar than semantically dissimilar
distractor items. Moreover, a more fine-grain level
of organisation has been detected within some
mass noun categories (e.g., foods) but not others
(e.g., drinks and non-edible solids). Finally,
identification of written mass and count noun terms
was also found to be affected by semantic
similarity between the target and distractor but not
by whether the target and distractor were drawn
from the same noun class. These results are
discussed in terms of potential principles of
semantic organisation within the mass noun
domain, and also with reference to existing
theoretical models of the mass/count distinction.
Although several neuropsychological reports in
the literature detail patients who show a selective
deficit for either mass or count nouns, these
impairments correspond to an inability to use
specific grammatical rules corresponding to a
particular class of noun rather than a selective
inability to comprehend a class of words per se (e.g.,
Semenza et al., 1997). The current study does not
counteract this trend; we do not provide evidence of
a dissociation between mass and count nouns at the
level of comprehension. However, to the best of our
knowledge, we do provide the first empirical
evidence of a fine-grain level of organisation
between such concepts. These results extend our
previous observations of gradients of semantic

1064

Sebastian J. Crutch and Elizabeth K. Warrington

Fig. 3 – Performance on the noun class and semantic similarity task (Experiment 4). (A) The percentage correct scores for the
semantically close (same noun type), close (different noun type) and distant arrays. (B) The number of errors made on consecutive probes
of the test stimuli under the three experimental conditions.

relatedness in the count noun domain (Crutch and
Warrington, 2005). With regard to the absence of a
semantic distance effect with drinks and non-edible
solids, it should be noted that premorbid personal
knowledge may affect significantly the quality and
quantity of organisation within the conceptual
system. Thus we would predict contrasting semantic
distance patterns to those observed in the current
patient if an equivalent refractory access deficit were
to be present in an individual with an extensive
understanding of the drinks or textiles market. A
similar account was given for the observation that
A.Z., an English lady, showed semantic distance

effects in her identification of the names of English
towns and cities but not US states (Crutch and
Warrington, 2003).
It has been hypothesised that neurological
deficits with a refractory access quality may affect
any cognitive system (McCarthy and Kartsounis,
2000; Crutch and Warrington, 2001). With the
present patient A.Z., it had been demonstrated
previously that identification of stimuli was affected
significantly more by the semantic relatedness of
stimuli than by their phonological relatedness
(Warrington and Crutch, 2004). This result was held
to indicate that a significant proportion of A.Z.’s

Semantic organisation of mass nouns

refractory deficit could not be attributed to presemantic aspects of information processing. In
Experiment 1 of this study, this information
processing analysis was extended by testing for
evidence of a post-conceptual locus for the refractory
disorder. This was achieved by examining the
influence of syntactic similarity (stimulus items
drawn from the same parts of speech category or
different parts of speech categories) upon the build
up of refractoriness in a written word identification
paradigm. No effect of syntactic similarity was
observed. We consider the result of Experiment 1 to
indicate that syntactic information is not strongly
represented at the information processing stage
which is primarily affected by refractoriness. As this
result constitutes a negative finding, its interpretative
power is relatively weak; however, we would claim
that the failure to find a syntactic similarity effect
suggests that the refractory disorder in A.Z. is
localised to a conceptual level of processing. In other
words, Experiment 1 fails to provide any clear
evidence that refractoriness affects the level at which
syntactic information is represented (either syntactic
or lemma level representations, depending upon
whether one adheres to the single or multiple stages
of semantic processing viewpoint).
Having provided empiricial evidence which is
at least highly consistent with a refractory disorder
which is restricted largely to pure conceptual
processing, we attempted to examine directly
whether mass/count status information is
represented at the conceptual level of processing.
The rationale for this hypothesis was that shared
mass/count status (above and beyond semantic
similarity) might yield some enhancement of the
refractory effect in our patient. The findings of
Experiment 4 are essentially negative, particularly
bearing in mind that the refractory effects were
generated in each condition using small stimulus
sets. As such this null result cannot contribute very
much to the debate as to whether the distinction
between mass and count nouns is fundamentally
semantic or syntactic in nature (e.g., Jackendoff,
1991; Vigliocco et al., 1999).
Acknowledgements. We would like to thank Dr David
Cohen and the therapists on the Harrow Stroke Unit,
Northwick Park Hospital, who allowed us to study a
patient under their care.
REFERENCES
BORGO F and SHALLICE T. When living things and other “sensoryquality” categories behave in the same fashion: A novel
category-specificity effect. Neurocase, 7: 201-220, 2001.
BORGO F and SHALLICE T. Category specificity and feature
knowledge: Evidence from new sensory-quality categories.
Cognitive Neuropsychology, 20: 327-353, 2003.
BUNT HC. The formal semantics of mass terms. In Kailsson F

1065

(Ed), Papers from the Third Scandinavian Conference of
Linguistics. Turku: Academy of Finland, 1976.
CAPITANI E, LAIACONA M, MAHON B and CARAMAZZA A. What are
the facts of semantic category-specific deficits? A critical
review of the clinical evidence. Cognitive Neuropsychology,
20: 213-261, 2003.
CARAMAZZA A. How many levels of processing are there in lexical
access? Cognitive Neuropsychology, 14: 177-208, 1997.
CIPOLOTTI L. Sparing of country and nationality names in a case of
modality-specific oral output impairment: Implications for
theories of speech production. Cognitive Neuropsychology,
17: 709-729, 2000.
CRUTCH SJ and WARRINGTON EK. Refractory dyslexia: Evidence
of multiple task-specific phonological output stores. Brain,
124: 1533-1543, 2001.
CRUTCH SJ and WARRINGTON EK. Spatial coding of semantic
information: Knowledge of country and city names depends
upon their geographical proximity. Brain, 126: 1821-1829,
2003.
CRUTCH SJ and WARRINGTON EK. The semantic organisation of
proper nouns: The case of people and brand names.
Neuropsychologia, 42: 584-596, 2004.
CRUTCH SJ and WARRINGTON EK. Gradients of semantic relatedness
and their contrasting explanations in refractory access and
storage semantic impairments. Cognitive Neuropsychology,
22: 851-876, 2005.
GILLON B. Towards a common semantics for English count
and mass nouns. Linguistics and Philosophy, 15: 597-640,
1992.
JACKENDOFF R. Parts and boundaries. In Levin B and Pinker S
(Eds), Lexical and Conceptual Semantics. Oxford: Blackwell,
1991.
KAY J, LESSER R and COLTHEART M. PALPA: Psycholinguistic
Assessments of Language Processing in Aphasia. Hove:
Lawrence Erlbaum Associates, 1992.
MCCARTHY RA and KARTSOUNIS LD. Wobbly words: Refractory
anomia with preserved semantics. Neurocase, 6: 487-497,
2000.
MONDINI S, JAREMA G and LIGUORI F. Semantics and syntax of
mass and count nouns: Data from aphasia and dementia. Brain
and Language, 91: 138-139, 2004.
PATTERSON K and HODGES JR. Deterioration of word meaning:
Implications for reading. Neuropsychologia, 30: 1025-1040,
1992.
SEMENZA C, MONDINI S and CAPPELLETTI M. The grammatical
properties of mass nouns: An aphasia case study.
Neuropsychologia, 35: 669-675, 1997.
SEMENZA C, MONDINI S and MARINELLI K. Count and mass nouns:
Semantics and syntax in aphasia and Alzheimer’s disease.
Brain and Language, 74: 428-431, 2000.
SOJA NN. Inferences about the meanings of nouns: The
relationship between perception and syntax. Cognitive
Development, 7: 29-45, 1992.
SOJA NN, CAREY S and SPELKE ES. Ontological categories guide
young children’s inductions of word meaning: Object terms
and substance terms. Cognition, 38: 179-211, 1991.
TALER V, JAREMA G and SAUMIER D. Processing mass/count
information in semantic dementia and agrammatic aphasia.
Brain and Language, 91: 160-161, 2004.
VIGLIOCCO G, VINSON DP, MARTIN RC and GARRETT M. Is “count”
and “mass” information available when the noun is not? An
investigation of tip of the tongue states and anomia. Journal of
Memory and Language, 40: 534-558, 1999.
WARRINGTON EK and CRUTCH SJ. A circumscribed refractory
access disorder: A verbal semantic impairment sparing visual
semantics. Cognitive Neuropsychology, 21: 299-315, 2004.
WARRINGTON EK and MCCARTHY R. Category specific access
dysphasia. Brain, 106: 859-878, 1983.
WARRINGTON EK and MCCARTHY R. Categories of knowledge:
Further fractionations and an attempted integration. Brain,
110: 1273-1296, 1987.
WISNIEWSKI EJ, IMAI M and CASEY L. On the equivalence of
superordinate concepts. Cognition, 60: 269-298, 1996.
Sebastian Crutch, Dementia Research Centre, Box 16, National Hospital for Neurology
and Neurosurgery, 8-11 Queen Square, London, WC1N 3BG, UK.
e-mail: s.crutch@dementia.ion.ucl.ac.uk

(Received 4 October 2004; reviewed 4 February 2005; revised 5 September 2005; accepted 11 October 2005; Action
Editor Ria De Bleser)

1066

Sebastian J. Crutch and Elizabeth K. Warrington

APPENDIX I
The stimulus arrays for the same and different parts of speech conditions (Experiment 1)
Parts of speech arrays
Same parts of speech
Nouns
emission
calculus
stipend
bard

Adjectives

tendency
virtue
mood
devotion

Verbs

Adverbs

pliable
arid
frugal
grisly

foolish
usual
humble
social

kindle
simper
usurp
abhor

seem
harm
guess
been

divinely
slyly
mortally
wantonly

simply
quickly
softly
easily

simper
frugal
slyly
calculus

been
foolish
quickly
mood

arid
mortally
stipend
usurp

humble
easily
virtue
seem

wantonly
emission
abhor
grisly

softly
devotion
harm
social

Different parts of speech
bard
kindle
pliable
divinely

tendency
guess
usual
simply

APPENDIX II
The semantically close, medium and distant arrays used in Experiment 2
Variable distance arrays
Close arrays
Solids

Liquids

Gases

Meats

Carbohydrates

Drinks

Jewels

Metals

Stones

etin
copper
zinc

petrol
oil
acid

oxygen
helium
air

lamb
beef
pork

rice
pasta
bread

wine
beer
sherry

ruby
emerald
diamond

silver
bronze
gold

marble
granite
quartz

Middle arrays
Chemicals
tin
oil
air

Consumables

Precious minerals/metals

petrol
helium
Zinc

oxygen
copper
acid

lamb
pasta
sherry

rice
beer
pork

wine
beef
bread

ruby
bronze
quartz

silver
granite
diamond

marble
emerald
gold

petrol
beer
diamond

oxygen
beef
gold

lamb
bronze
air

rice
granite
zinc

wine
emerald
acid

ruby
oil
sherry

silver
helium
pork

marble
copper
bread

Distant arrays
tin
pasta
quartz

APPENDIX III
The semantically close and distant arrays of food items used in Experiment 3
Food mass noun arrays
Close arrays
Meats

Cheeses

Desserts

Sauces

beef
lamb
pork
veal

cheddar
brie
stilton
edam

trifle
custard
pavlova
semolina

ketchup
mustard
vinegar
mayonnaise

mustard
brie
trifle
lamb

pork
stilton
custard
vinegar

edam
mayonnaise
semolina
beef

Distant arrays
veal
cheddar
pavlova
ketchup

1067

Semantic organisation of mass nouns

APPENDIX IV
The semantically close and distant arrays of drink items used in Experiment 3
Drink mass noun arrays
Close arrays
Long drinks

Spirits

Hot drinks

Mixers

lager
beer
cider
shandy

rum
brandy
vodka
whisky

coffee
tea
ovaltine
cocoa

water
lemonade
soda
tonic

lager
lemonade
tea
brandy

rum
shandy
water
ovaltine

cider
whisky
cocoa
tonic

Distant arrays
coffee
vodka
beer
soda

APPENDIX V
The semantically close and distant arrays of non-edible solid items used in Experiment 3
Non-edible solid mass noun arrays
Close arrays
Metals

Stones

Heavy fabrics

Light fabrics

iron
copper
steel
lead

marble
granite
concrete
slate

corduroy
tweed
denim
flannel

linen
silk
nylon
cotton

nylon
denim
concrete
steel

granite
tweed
copper
silk

flannel
lead
slate
cotton

Distant arrays
iron
corduroy
marble
linen

APPENDIX VI
The semantically close same noun type arrays, semantically close different noun type arrays and semantically distant noun
arrays used in Experiment 4
Mass vs. count noun arrays
Semantically close pairs

Semantically distant pairs

Same noun type

Different noun type

cloud – storm1
branch – twig1
blizzard – hurricane1
steak – chop1
snow – rain2
wood – bark2
thunder – lightning2
pork – mince2

snow – cloud
rain – storm
branch – bark
twig – wood
pork – steak
mince – chop
blizzard – lightning
hurricane – thunder

snow – pork
rain – mince
steak – cloud
storm – chop
branch – thunder
twig – hurricane
lightning – bark
wood – blizzard

1Same noun type pairs consisting of two count nouns. 2Same noun type pairs consisting of two mass nouns.