Aphasiology

ISSN: 0268-7038 (Print) 1464-5041 (Online) Journal homepage: https://www.tandfonline.com/loi/paph20

Naming problems in dementia: Semantic or

lexical?

Arlene J. Astell & Trevor A. Harley

To cite this article: Arlene J. Astell & Trevor A. Harley (1998) Naming problems in dementia:
Semantic or lexical?, Aphasiology, 12:4-5, 357-374, DOI: 10.1080/02687039808249538

To link to this article: https://doi.org/10.1080/02687039808249538

Published online: 29 May 2007.

Submit your article to this journal

Article views: 215

View related articles

Citing articles: 3 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=paph20
APHASIOLOGY, 1998, VOL.12, NOS. 415,357-374

Naming problems in dementia : semantic

or lexical?

A R L E N E J. A S T E L L t and T R E V O R A. H A R L E Y $
t University of Abertay Dundee, Dundee, Scotland, UK
$ University of Dundee, Dundee, Scotland, UK

Abstract
The naming behaviour was explored of patients moderately deteriorated with
probable Alzheimer’s disease (PAD) by analysing responses made on a picture
naming and two word-picture matching tasks for the same items. Naming
responses were classified into target and non-target and the relationship
between non-target responses and their targets were explored. It was found
that semantic relatedness influenced these non-target responses but frequency
and imageability did not. The PAD participants performed significantly better
on the two word-picture matching tasks than the naming task, while there was
no dfference on these within- and between-category tasks. These findings are
explored with reference to a two-stage model of lexical access, which has
separate storage of semantic, lexical and phonological information. The
participants’ difficulties appear to lie largely in retrieving the labels for items
they comprehend.

Naming in dementia
Difficulty with name retrieval is considered a common early sign of probable
Alzheimer’s disease (PAD), an illness in which patients show progressive
impairment in ability across all cognitive domains. Reduced performance o n verbal
fluency tasks (Stuss and Benson 1986, Huff et a/. 1986, Hart eta/. 1988, Bayles eta/.
1989) and the increasing occurrence of circumlocutory responses (Kertesz e t a/.
1986, Hodges e t a/. 1991) are also consistently reported. Patients with PAD make
more tip-of-the-tongues than age-matched controls (Astell and Harley 1996a) and
their word definitions become more tangential, with decreasing acknowledgement
of conventional form (Astell and Harley 1996b). By contrast, spontaneous speech
remains fluent with appropriate syntactic structure (Appell e t a/. 1982, Kirshner
e t a/. 1984, Neary e t a/. 1986, Blanken et a/. 1987) and phonemic processes are
relatively preserved (Appell e t a/. 1982, Hodges e t a/. 1991). While utterances
characteristically become shorter than those of age-matched controls (Blanken eta/.
1987, Ripich eta/. 1991), PAD patients retain the structure of turn-taking and other
features of orderly conversation (Ripich e t a/. 1991). Overall communicative
function decreases due to the reduced information content (Constantinidis e t a/.
1978) as PAD spontaneous speech progressively contains fewer nouns and more
verbs and adverbs than that of controls (Blanken e t a/. 1987).

Address correspondence to: A. J . Astell, Division of Psychology, University of Abertay Dundee,

Dundee, D D l INJ, Scotland, UK.

0268-7038/98 $12.00 0 1998 Taylor & Francis Ltd

358 A . J. Astelland T. A . Harig

The progressive and accumulating language difficulties of PAD, highlight the

gradual but insidious loss of control over processing that occurs, including loss of
both metacognitive awareness and the ability to participate in formal assessment.
These factors interact progressively through the deteriorating course of the illness
with task difficulty (Nebes 1989, Bayles et al. 1991). Patients with PAD become less
aware of their performance on tasks and less and less able to carry out the tasks. The
influence of task demands have been shown in the performance of healthy ageing
too, for example in naming (Nicholas et al. 1985).
Naming has been widely investigated in PAD, in part because it is a prominent
early sign of dementia, evident before debilitating deterioration has occurred.
There is evidence that naming behaviour is a good predictor of rate of decline
(Boller et al. 1991) and correlates with severity (Skelton-Robinson and Jones 1984,
Neils et al. 1988). Early investigation of naming in PAD, and indeed healthy
participants, made use of the two-stage model of naming proposed by Oldfield and
Wingfield (1964,1965). Their first stage is recognition of the item, the second stage
is a search through the word store for the name of the recognized item. Visual
errors are explained by a failure at the item recognition stage (Barker and Lawson
1968, Rochford 1971). This is influenced by familiarity (Oldfield and Wingfield
1965) and by the quality of the stimuli (Kirshner et al. 1984), both of which interact
with frequency (Rochford 1971, Skelton-Robinson and Jones 1984). Problems
with perceptual processes have been shown to most notably influence naming
errors in the later stages of the illness (Huff et al. 1986, Hodges e t al. 1991).
In the early stages of PAD, semantic difficulties are thought to be more
influential on naming responses than perceptual responses and have attracted more
investigation. Semantically-relatederrors predominate in naming, suggesting that
there is an underlying semantic disorder (Bayles and Tomoeda 1983, Martin and
Fedio 1983, Nebes 1989, Hodges e t al. 1991). Explanations for this have followed
those proposed by Warrington and Shallice (1979) of either impaired access to
semantic information or a degraded semantic store.
Both the impaired access and impaired storage accounts have been applied to
naming behaviour in PAD. However, the assumptions underlying the criteria
proposed by Warrington and Shallice (1979) to distinguish between access and
storage problems, have been challenged (Caplan 1992, Rapp and Caramazza 1993).
Even the measurement of consistency in responding, the criterion considered by
Shallice (1988) to have most face validity in making this distinction, has been
questioned (Faglioni and Botti 1993). In addition, contradictory findings have been
reported, some supporting impaired storage and others impaired access. For
instance, Somners and Pierce (1990) found that their ten PAD participants, five
mildly and five moderately impaired, were worse at naming low-frequency
nouns than high, supporting an impaired storage explanation, according to the
Warrington and Shallice (1979) criteria. However, they also found that the PAD
participants were able to accurately identify semantic attributes, superordinates,
contextual and functional associates of the targets, suggesting naming problems
arise from impaired access, rather than loss of item-specific semantic knowledge.
This apparent contradiction may be resolved if frequency can alternatively reflect
an access disorder: for example low frequency items are likely to be less easy to
activate to threshold.
One contributing factor to the contradictory findings in PAD naming behaviour
may be the use made by many investigators of Tulving’s (1972) description of
Naming problems in dementia 359

G Semantic

0 Lexical

0 Phonological

+
Speech output
Figure 1. A basic two-stage model of lexicalization.

semantic memory (Chertkow and Bub 1990, Hodges e t al. 1992). Tulving
considered semantic memory to contain the lexicon, grammatical rules and the
knowledge required to manipulate incoming stimuli. With words and the concepts
they refer to as part of the same store, it is difficult to distinguish if problems lie
with concepts (semantic information) or labels for concepts (lexical items) or with
both. There is evidence that at least part of the naming difficulty in PAD results
from a problem accessing lexical items. For instance there are noted word-finding
problems and an increase in circumlocutory responses, where participants explain
the function of a to-be-named item, rather than actually naming it. Also, phonemic
cueing is beneficial to name production, at least in the earlier stages of PAD
(Nicholas e t al. 1985, Neils e t a/. 1988, Funnel1 and Hodges 1991). These findings
suggest that the naming problem in PAD reflects loss of access to the phonological
forms (Hier e t al. 1985, Neils et al. 1988, Nicholas et a/. 1996). This is in keeping
with studies of healthy older adults, where reduced target naming has been
attributed to impaired lexical access, rather than semantic impairment (Nicholas et
al. 1985).

The loctrs of the naming deficit

In contrast to Tulving’s (1972) approach, most current psycholinguistic models of
lexicalization have semantic and lexical information stored separately (see figure 1;
Kempen and Huijbers 1983, Martin e t al. 1989, Dell and O’Seaghdha 1992, Garrett
1992, Harley and MacAndrew 1992, 1995, see Levelt 1992, for a review, Harley
1993). Lexicalization is the process of moving from a semantic specification of what
is to be said to the phonological form of the words retrieved. The first stage is the
specification of a phonologically abstract lexical item or lemma, an abstract,
syntactically specified package, from the semantic specification (Levelt 1992). The
second stage is the accessing and retrieval of the appropriate phonological word
form, which also contains morphological information (Levelt 1992). Frequency
360 A . J . Astell and T. A . Harley

effects appear to arise post lexical access; that is after lemma retrieval (Jeschniak
and Levelt 1994, Harley 1995). With the two-stage model it is possible to
investigate whether difficulties lie with concepts or names or with both. On the
basis of the evidence reviewed above, this paper concentrates on the semantic-to-
lexical (lemma stage) of lexicalization.
Several possible explanations for difficulties arising at the lemma level suggest
themselves. First, at the semantic level, the semantic specification may be damaged
or ‘lost ’. Semantic memory is popularly represented as containing concepts,
defined by features, connected to related concepts. Ideas about the representation
of semantic information have been greatly influenced by the work of McClelland
and Rumelhart (1985) who proposed a distributed model of memory. Their
approach offered a synthesis of the main elements of the two previously most
popular approaches-networks (Collins and Quillian 1969, Collins and Loftus
1975, Glass and Holyoak 1975) and features (Meyer 1970, Rosch 1973, Smith et al.
1974, McCloskey and Glucksberg 1979). In models such as McClelland and
Rumelhart’s (1985), concepts are represented by patterns of activation across a
combination of nodes, each signifying an attribute. Should one or more of these
distinguishing features be lost, then semantically related items that share the
remaining features, could be produced. In dementia, increasing amounts of
semantic information may be lost over time, resulting in random output.
A second possibility is a problem at the lexical level of representation. Appell et
al. (1982) suggested that in PAD there may be ‘a loosening of links between words
and between words and the things that they represent’. It is possible, in the two-
stage model of lexicalization,that lemmas could be weakened or lost. Alternatively,
there may be a failure to inhibit plausible alternatives, as suggested by Miller
(1979). A third possible explanation lies in the vulnerability of the links between the
semantic and lexical levels, which may weaken and eventually become lost. In a
degenerative disorder it is possible that several, if not all, of these types of
disruption may occur over time.
To investigate these possible explanations, participants with moderate PAD
were asked to spontaneously name pictured items. In addition to examining overall
performance on this task, non-target responses were analysed. On naming tasks,
the nature of the erroneous responses can be more informative about the loci of
breakdowns in processing than overall scores (Nicholas et al. 1996). These
responses are distinct from ‘don’t knows ’ and non-responding. Non-target
responses include alternative names for the pictured item, naming parts of the
target item or other items in the picture, and descriptive responses rather than
single words. A descriptive classification of responses is used here based on that of
Hodges etal. (1991). Using this system, Hodges etal. found the PAD related errors
to be largely semantic superordinates or semantic associates of the targets. Over
time there was an increase in the number of semantic-associative errors and the
occurrence of increasingly more visually-related errors.
In this study, the naming task was followed by an assessment of their
comprehension for the named items using a word-picture matching task. This
comprehension task has both between- and within-category elements. Of these,
between-category choosing is easier, as identification of an item’s category is
sufficient for selection when all other items are from different semantic categories.
Within-category matching requires a more detailed match to select the correct
target from among category co-ordinates. This element thus probes the depth of
Naming probferns in dementia 361

knowledge available about target items in response to the name and visual
information. These data are not intended to examine the range of models that have
been proposed to capture the relationship between input and output processes.
These primarily divide between models with modality-specific semantic stores
(Warrington and Shallice 1979, Shallice 1988, 1993) and models with a single
amodal semantic store and different access for different modalities (Riddoch et af.
1988, Caramazza and Hillis 1990). This study is purely concerned with examining
the participants’ comprehension of the targets and makes no claims regarding
specific models. In one longitudinal study of a single patient, comprehension for
items was retained longer than the ability to name spontaneously the same items
(Funnell and Hodges 1991). These findings led Funnell and Hodges to propose that
in PAD the elements of the healthy naming system remain intact while the ability
to operate the system progressively deteriorates. This is supported by findings
from a word definition task, where PAD participants supplied target-relevant
information but lacked awareness of the conventional form adopted by older
children and healthy adults when asked to define a word (Astell and Harley 1996b).
It is expected that PAD participants will be worse at naming than at
comprehension, as naming is consistently reported as being impaired in PAD. As
word frequency is argued to be located post-lemma, it is not expected that
semantically-related incorrect naming responses will be higher in frequency than
the targets. There is evidence from analysis of spontaneous speech samples that low
frequency words are retained in PAD (Miller and Hague 1975). In addition, it was
found that intrusions were no more frequent than targets in substitutions made in
tip-of-the-tongue states in PAD (Astell and Harley 1996a). If, as predicted, the
naming problem in PAD arises at the lemma stage of lexicalization, there should be
no difference in performance on the within- and between-category parts of the
comprehension task. However, if semantic representations are compromised, then
PAD participants should do worse at within-category matching. If the problem
arises from impaired lexical representations, then one would expect comprehension
performance to be influenced. If there is a difliculty with within-level lexical
inhibition one expects semantic relatives of the target to receive activation.
However, feedback from the phonological level may also influence lexical selection
in this circumstance, where frequency and phonological similarity influence
output. This seems unlikely as the relative preservation of phonological processes
is well documented in PAD (Appell e t a/. 1982, Hodges e t al. 1991). Additionally,
Harley and MacAndrew (1992) found that varying within-level lexical inhibition is
unlikely to lead to the production of semantic substitutions.
On the between-category task there should be a low level of phonological
distractors selected as there is little evidence of phonological failure in PAD.
Alongside this one may expect selection of perceptual distractors in the between-
category comprehension task and visual errors in the naming task in more severely
impaired participants, as perceptual processes are reported to be disturbed in
advanced PAD.
362 A . I. Astell and T. A . Harley

Method
Participants
The experimental group comprised four males and eight females aged between
70: 11 years and 88 :4 years, with a mean age of 81 :7 years. Their mean years of
formal education was 101 (standard deviation 1.2) with a range of 9-12 years. Each
had a diagnosis of probable Alzheimer’s disease (PAD), using the criteria developed
by the Work Group of the National Institute of Neurological and Communicative
Disorders and Stroke (NINCDS) and the Alzheimer’s Disease and Related
Disorders Association (ADRDA) (McKhann et al. 1984) and all had Hachinski
Ischaemic scores of four or less (Hachinski et al. 1975). Diagnoses were made by a
Consultant Psychiatrist based on history, toxicological and serology screenings and
reported course of deterioration. Each was tested on the Mini-Mental State
Examination (MMSE; Folstein e t al. 1975) producing a mean of 11.2, range 4-18.
These participants closely match the moderate group of PAD participants described
by Nicholas e t a/. (1996), who had a range of scores on the MMSE of 3-16, mean
age 77 years.
The naming task was also administered to two healthy elderly adults, one female
aged 80: 9 years and one male aged 76 :11 years. They both obtained MMSE scores
of 30 and their average number of years of education was 9. These two controls
were added to obtain some idea of healthy performance on this task as the drawings
were not originally intended for naming. Their responses were rated using the
criteria detailed below but were not subject to any further analysis.

Materials
All items were taken from the lexical comprehension task from the Philadelphia
Comprehension Battery for Aphasia (PCBA; Saffran et al. 1989). The lexical
comprehension task is designed to assess lexical-semantic knowledge and does this
by presenting the participant with a label (either verbal or written) which then has
to be matched with one of four pictures on a sheet. There were two sets of words
and pictures. The first set contained target items placed with semantically unrelated
distractors. Phonological and perceptual factors are sampled by having one of the
other items either look similar or sound similar to the target. In the second set a
subset of the above targets are presented with three other items from the same
semantic category.
In its original form there were 16 items in the within-category condition and 28
in the between-category one. For use with British subjects, four items from the
between-category section were dropped because their distractor value was based
on American pronunciations or names (cabbage, scale, lather, comb). This left 24
items in the between-categories condition, 16 with perceptual distractors and eight
with phonological distractors. These 24 target items were selected for the naming
task.

Procedure
Naming
In the naming task subjects were shown each picture in turn and asked to name the
item. There was no time constraint on how long subjects had to answer.
Naming problems in dementia 363

Table 1. Scoring criteria for naming responses

1.’ Target or synonym (e.g. ‘cabinet’ for chest-of-drawers).

2.* Augmented response-sentence or phrase containing target.
3.* Tip-of-the-tongue (TOT) resolved with either the target or a synonym. The participant
indicates that they know the name of the item but can not produce it straight away. After
some time the target name is produced.
4.* TOT unresolved. The participant indicates that they know the name of the item but they fail
to produce it. The participant may produce related words or the initial sound of the target
word.
5. Visual : responses visually similar to the target and from a different semantic category
(‘sweet’ for strawberry; ‘sole’ for chop). Also includes whole-part responses where subjects
name either a part of the target item (‘fence’ for crown: ‘plates’ for money) or something
incidentally present in the picture (‘hair’ for crown; ‘collar’ for tie).
6. Ambiguous visual/semantic category : responses from the same semantic category as the
target and visually similar such that the error could be either perceptually or semantically
based (‘stool’ for table; ‘bullock’ for goat; ‘saxophone’ for trombone).
7. Semantic-within-category : responses from the same semantic category as the target but
clearly not visually similar (‘lobster’ for octopus).
8. Semantic-superordinate: responses denoting the general class or category to which objects
belong (‘animal’ for goat; ‘bird’ for peacock).
9. Semantic associative : responses showing an obvious semantic association with the target
item including statements of action or function (‘something to eat’ for steak), physical
attributes (‘something that crawls’ for crab), contextual associates (‘in the kitchen’ for
steak), and specific subordinate or proper nouns examples of the target (‘Rolls Royce’ for
car).
10. Semantic circumlocutory : multiword responses showing accurate identification of the target
by physical attribute, function or action but not giving the name. Where the distinction
between 9. and 10. is unclear, apply the following criteria: does the response describe a
specific item? If so, score as 10.
11. Phonemic: mispronunciations or distortions of the target name sharing at least one syllable.
12. Perseverations : reutterances of a previous response (correct or incorrect) used to name one
of the previous five pictures.
13. Unrelated: no clear relationship between the target and response can be deduced (‘airy one’
for strawberry; ‘making a W. C.’ for road).
14. Nonresponse : includes ‘don’t know’ and no response.

* Additional categories to those of Hodges et af. (1991).

Comprehension
The comprehension task was preceded by two practice sheets to ensure that the
subjects understood the requirements. Each sheet was presented in turn and the
target item named by the interviewer. The subject was required to point to the
target item. All subjects completed the within-category part of the task first,
followed by the between-category section, Subjects were tested individually.

Scoring criteria
Naming
A scoring system for the naming task was developed based on that of Hodges et a/.
(1991 ; see table 1). Four additional categories were added to cover the full range
of responses made. All parts of the responses were considered in arriving at a
rating. Each set of responses was rated independently by three raters. Discrepancies
were discussed and agreement reached regarding classification.
364 A . 1.Astell and T . A . Harley

In addition to the above classification of responses, the pairs of target words and
substitutions made by the participants were included in a separate rating task. In
this the target-substitute pairs were rated for semantic relatedness by 14
independent raters. Raters assigned word pairs a value between zero and four using
the following criteria: (1) no link: the meanings of the words are completely
unrelated; (2) far-fetched link; (3) weak link; (4) strong link; and (0) where the
meaning of one or both of the words is unknown, after Jones (1989). Mean ratings
were calculated for each pair, excluding 0 ratings. A mean rating of 1-5or more was
taken as indicating that the pair of words are semantically related in some way.
The frequency of the targets to which errors were made were also examined,
using the ratings of Francis and KuEera (1982). Low frequency items were
designated as a frequency of occurrence of less than 20 times per million and high
as more than 20 times per million. In addition frequency, imageability and syntactic
relationship of the target-substitute pairs were also analysed, as these factors are
known to be important in word substitutions in normal speakers (Harley and
MacAndrew 1995). Frequency ratings for target-substitute pairs were also taken
from Francis and KuEera (1982). Imageability ratings for target-substitute pairs
were taken from the Oxford Psycholinguistic Database (Quinlan 1992) which is a
composite value arrived at by blending ratings from the Paivio et a/. (1968), Toglia
and Battig (1978) and Gilhooly and Logie (1980) imageability norms. The syntactic
relationship between target-substitute pairs was determined by comparing the
syntactic class of the target, all nouns, with the syntactic class of the substitute.

Comprehension tasks
Unless otherwise stated, scores on the 16 item within-category comprehension task
are scaled up to the equivalent score for 24 items. This is for ease of comparison
with the 24-item between-category and naming tasks.
Distractor selection was analysed with regard to semantic relationship on the
within-category tasks and phonological, perceptual or unrelatedness, on the
between-category task. No attempt was made to compare the frequency of targets
and distractors as the selection of these items as materials in the PCBA did not
appear to have been based around these criteria. Participants performed the naming
task at least 1 week before the comprehension task.
The 24 pictures were presented to two elderly control participants to examine
their quality as naming stimuli. Of these one scored 23 correct with 1 ambiguous
visual/semantic response and the other scored 21 correct with 1 ambiguous
visual/semantic response, 1 semantic co-ordinate and 1 unresolved TOT (mean
correct 22).

Results
Naming
The highest number of correct items (response categories 1,2, and 3 - s e e figure 2)
by a PAD participant was 22 and the lowest 4, mean correct 143. To the total 288
items (12 patients x 24 pictures) the PAD participants made 172 (60%) responses
containing the correct target. Of the 116 items scored incorrect, 34 had some visual
relationship with the target and 38 some semantic relationship (both including
Naming problems in dementia 365
1 target- 1
2 augmented response
3 TOT nsolved
4 TOT umso~ved
e 5 visd I
&, 6 ambiguous visudsemantic
2
0
7 semantic within I
8 semantic superordinate 1
B 9 semantic associative I
10 semantic circumlocution I
11 phonological I
12perseverntion I
13 unrelated I
14don'tknow I
I I I I I I
0 20 40 60 80 100
Percentage
Figure 2. Distribution of naming responses across all response categories for PAD subjects.

ambiguous visual/semantic responses), five were perseverative 23 were unrelated,

and 18 were ' don't knows '. One phonological mispronunciation occurred
(banoodoo) followed immediately by the target (banana). Where frequency ratings
were available for the targets, 50 were made to low frequency items (mean
frequency rating of targets = 5.9) and 19 to high frequency (mean frequency rating
of targets = 178).
This classification of responses was supported by the semantic ratings of the
target-substitution pairs (see Appendix). Sometimes more than one word was
produced in a response and the total number of target-substitute pairs was 76.
These pairs were from three groups of responses (as outlined above)-categories
two to four inclusive (TOTS and target circumlocutions), categories six to 10
inclusive (some semantic relationship) and category five (visual errors). Of these
three groups, the pairs classified from categories six to 10 received the highest
semantic ratings (mean 3-32).The ratings for responses from categories two to four
were not much lower with a mean of 3.13. The responses judged to be visual errors
received much lower ratings of semantic relatedness (mean 1.54).
Comparisons of the frequency, imageability and syntactic category were also
made between the target words and substitutions produced in the visual and
semantic errors, the TOT states and circumlocutory responses. Perseverative and
unrelated responses were not included. Frequency ratings for both members of 59
of these pairs were obtained from Francis and Kui-era (1982) and there was no
significant difference between them (mean target frequency 48.39, mean substitute
frequency 40.18; t(58) = 0.55, p > 0.5). Imageability ratings for 30 target-
substitute pairs were available from the Oxford Psycholinguistic Database (Quinlan
1992). As with frequency there was no significant difference in imageability (mean
target imageability 6.01, mean substitute imageability 5.94; (20) = 0.65,p > 0.5).
All of the target words and all of the substitute words were from the same syntactic
class, that is nouns.
Analysis of frequency and imageability was also carried out separately on the
three largest groups of substitutions, the semantic (response categories seven to
lo), the visual (category five) and the ambiguous visual/semantic (category six).
366 A . J. Astelland T. A . Harley

within correct

semanticrelative

l5- within don’t know

3
c.,
between comct
cp
0
I
I
I
between don’t know
t 1 1 I 1 J
0 20 40 60 80 100

percentage
Figure 3. Distribution of responses on between-category comprehension and within-category
comprehension.

Among these comparisons the only significant difference found was that targets of
visual errors were more imageable than the substitutions (target mean 6.10, visual
error mean 5.59; t(7) = 2.26, p < 0.05).

Comprehension
Subjects were scored for how many items they correctly identified. Performance on
the naming and two comprehension tasks was significantly different (means of 143
for naming, 19.75 for within-category comprehension and 21 for between-category
comprehension; F(2,22) = 33.85, p < 0.0005). Post-hoc Tukey tests revealed that
the number of targets named was lower than the number of items comprehended
on both between and within category tasks (both at p < 0001). There was no
significant difference in performance of the two elements of the comprehension
task (p > 0 3 ) . On both within- and between-category comprehension over 80 %
of responses were correct (see figure 3).
On the within-category part of the task, twice as many of the non-target
responses were semantic distractors (11.45 %) than were ‘don’t knows’ (5.72 % ;
see figure 3). On the between-category part almost two thirds of incorrect
responses were ‘don’t know’ (7.9%) or no responses, rather than selection of a
phonological (0.7 %) or perceptual relative of the target (0.7 % ; see figure 3).
To explore patterns in responding, responses to the 16 items that occurred on
both parts of the comprehension task and the naming task were looked at. Initial
analysis of the conditional probability of any response on the within-category
comprehension given that same response on the between-category suggested that
participants were most likely to respond in the same way on the within- and
between-category parts (t(l5) = 2695, p < 0 0 5 ) . In addition, comparisons across
items revealed that the probability of both naming and comprehending an item was
Naming problems in dementia 367

Table 2. Comparison of correct and incorrect responses made to each of the 16 items on each
task. Tasks compared in order of appearance

Succeed-succeed Fail-fail Succeed-fail Fail-succeed

Between & naming 164 29 88 7

Within & naming 117 24 41 10
Between & within 158 10 19 5

Key: between = between-category comprehension; within = within-category comprehension.

Table 3. Patterns of correct and incorrect responses made to the 16 items across all tasks.
Tasks compared in order of appearance

SSS SSF SFS FSS SFF FSF FFS FFF

Between-, within-, naming 114 39 7 1 14 6 1 10

Key: S = succeed; F = fail; between = between-category comprehension; within = within-

category comprehension.

greater than the probability of naming alone on both comprehension tasks (within-
category: t(l5) = 3.460, p < 0005; between-category : t(23) = 3.175, p < 0005).
However, individual item-by-item comparison across tasks revealed inconsis-
tencies in performance (see tables 2 and 3). So while participants were significantly
more likely to correctly name an item if they recognized it, there were instances
where items were named correctly but not recognized, or were recognized on one
comprehension task and not the other.

Discussion
As expected the PAD subjects performed better on the comprehension task than on
the naming task. O n the naming component, targets and erroneous responses were
all drawn from the same syntactic category and over a third of the non-target
responses were either unrelated or ‘don’t knows’. Errors related to the target were
distributed between visual and semantic relatives, with similar amounts of each.
Among the semantic substitutions, there was no evidence of a tendency to produce
superordinate semantic category responses as reported in other studies (Martin and
Fedio 1983, Hodges et al. 1991). Semantically related errors were as likely to be
category co-ordinates of the target or even more likely to have some associative
relationship with the target, such as its function. However, there was also quite a
high number of visual errors. Kirshner e t al. (1984) demonstrated the importance
of the quality of the stimuli to both PAD and healthy naming performance.
Alongside this, some of the PAD participants were approaching the severely
deteriorated stage, as indicated by their MMSE scores, where perceptual processes
have been shown to influence naming performance (Kirshner e t al. 1984, Hodges
etal. 1991).
Frequency has also been shown to influence naming performance in both healthy
participants and PAD. As predicted, in this naming task, there were no differences
in frequency between targets and substitutes in the nontarget responses, which
mainly comprised semantically- or visually-related errors. However, more semantic
and visual errors were made in naming the low frequency items. Naming low
368 A . J. Astell and T. A . H a d 9

frequency items of any sort is hard because their lexemes are less likely to be
available (as frequency effects, ex hypothesi, are located at the lexeme level). Hence
errors are more likely to be made to low frequency items. Damage earlier in the
system exacerbates this (Hinton and Shallice 1991, Tippett and Farah 1994, Harley
1998).
Most words in the language are low frequency (Francis and KuEera 1982).
Therefore, in a random sample, the substitutes are also likely to be low frequency.
Indeed, this supports the hypothesis because low frequency items will only be
randomly affected by semantic-lemma damage. It is not the case that all low
frequency items are equally dected by early damage. This pattern of results would
not be obtained if it were the lexemes themselves that were affected. In this case,
one would expect higher frequency lexemes to substitute for low, which is not
observed. In addition, some errors were made to high frequency items. This
supports findings from analysis of spontaneous speech samples where low
frequency words are retained (Miller and Hague 1975). In addition, no frequency
differencesbetween targets and substitutes were found in PAD tip-of-tongue states
(Astell and Harley 1996a).
The substitutions were also compared with their targets on rated imageability,
which influences semantic substitutions in normal speech errors (Harley and
MacAndrew 1995). In the present study only one group of target-substitute pairs
differed significantly in imageability. These were the visual errors, where the
targets were rated more imageable than their substitutes. This may reflect the
quality of the stimuli, which the findings from the two controls suggest was not
completely satisfactory for the naming task.
However, on the comprehension task the majority of items were identified
correctly. That there were so few perceptual errors on the between-category
picture-word matching suggests there was no great problem with perceptual
processing. It may be that the drawings were adequate within the constraints of the
word-picture matching task, where they were supported by provision of the target
name, but were inadequate to meet the demands of the naming task. This suggests
there was not a great difficulty moving from the drawings to the semantic
specification. There was also a low level of phonological distractors selected, as
predicted, alongside the occurrence of only a single phonological mispronunciation
in the naming task.
The finding that there was no difference in performance on the within- and
between-category parts of the comprehension task suggests that the naming
problem does not result from a loss of semantic knowledge for the concepts. This
is because the PAD patient’s knowledge of the items must be sufficiently detailed
to perform within-category comprehension at the same level as between-category.
The words for each item were understood, suggesting at least that the lexical
referents were in storage. As significantly more items were comprehended than
named, failure in naming cannot be interpreted as loss of the item, as the items must
have been present to enable comprehension. These data do not speak directly to
competing models of the relationship between input and output processes
connected to semantic representations. However, the finding that successfully
named items were more likely to be comprehended than not, requires explanation
by any model hoping to explain these processes. The results from the naming and
comprehension tasks together suggest that evidence of the integrity of semantic
concept knowledge may not always be gained from naming tasks alone.
Naming problems in dementia 369

Returning to the hypotheses generated from the two-stage model of lexical-

ization, one can examine each in the light of the data presented here. It is suggested
that there could be problems with the representations at semantic or lexical levels.
The findings of good comprehension and recognition of the items suggest that
there is no loss of items, either at the semantic or lexical level of representation.
Distinguishing between the other two hypotheses, those of loss of within-level
inhibition at the lexical level of representation and of impaired links between the
semantic and lexical levels, is more difficult. However, there is evidence which
suggests that disinhibition of within-level lexical connections cannot account for
semantic word substitutions (Harley and MacAndrew 1992). In their simulations
with an interactive activation two-stage model of lexicalization, decreasing within-
level lexical inhibition did not result in output of semantic relatives.
This leaves one hypothesis, that of impaired links between semantic and lexical
representations. This was also considered the most likely explanation of word
substitutions in aphasic speech (Harley and MacAndrew 1992) and of tip-of-the-
tongue states in PAD (Astell and Harley 1996a). With this explanation, there would
be no prediction of frequency difference between target and substitute in semantic
substitutions. This is because if activation does not pass to the target lexical item
it will pass to a close semantic relative, which will be an item with a very similar
semantic specification, and which may or may not be higher in frequency. The
semantic similarity alone will guide substitute selection, with frequency affecting
only post-lemma processing. This hypothesis is the equivalent, with a two-stage
model of lexicalization, of the Warrington and Shallice (1979) impaired access
account of semantic impairment. Thus the data suggest that semantic and lexical
representations are intact, but there is a problem translating between them.
The PAD participants in this study made similar responses on the naming task
to those of healthy elderly people reported by Nicholas e t al. (1985). They too
concluded that the problem lay in lexical retrieval rather than with the loss of
semantic knowledge. Taken together with findings from other tasks, it appears that
in PAD, the progressive breakdown that occurs during the illness, affects processes
(such as naming) rather than components (for instance words and concepts;
Funnel1 and Hodges 1991, Astell and Harley 1996b). As the illness progresses, the
loss of control over processing interacts with task difficulty. Patients with PAD
become less and less able to recognize visually presented items and stay within the
demands of the task, such as providing single-word answers in a naming task. They
appear to retain the concepts and labels in storage, but are decreasingly able to
marshal1 and retrieve stored information as required by formal tasks and then, as
the illness progresses, in conversation. The finding that comprehension outlasts
production, perhaps no surprise given that it develops earlier in the developmental
process, is important for guiding interactions with people with PAD. Although
their output may lessen and finally lose communicative function, comprehension of
input lingers on.

Acknowledgements
This work was undertaken as part of a PhD dissertation by the first author, funded
by a research studentship from the UK ESRC/MRC/SERC Initiative in Cognitive
Science/HCI. The second author’s research was supported by Grant No. SPG
9018232 from the ESRC/MRC/SERC Initiative on Cognitive Science/HCI. We
370 A . J. Astell and T. A . Harig

would like to thank the staff and patients of the South Warwickshire Health
Authority for their help.

References
APPELL,J,, KERTESZ,A. and FISMAN,M. 1982, A study of language functioning in Alzheimer
patients. Brain and Language, 17, 7391.
ASTELL,A. J. and HARLEY,T. A. 1996a, Tip-of-the-tongue states and lexical access in dementia.
Brain and Language, 54, 196-215.
ASTELL,A. J. and HARLEY, T. A. 1996b, The structure of semantic memory in dementia: Evidence
from a word definition task. (Submitted).
BARKER, M. G. and LAWSON, J. S. 1968, Nominal aphasia in dementia. British Journal of Pychiatry,
114, 1351-1356.
BAYLES,K. A. and TOMOEDA, C. K. 1983, Confrontation naming impairment in dementia. Brain and
Language, 19, 98-1 14.
BAYLES,K. A., SALMON, D. P., TOMOEDA, C. K., JACOBS, D., CAFPREY, J. T., KASZNIAK, A. W. and
TROSTER,A. I. 1989, Semantic and letter category naming in Alzheimer’s patients: A
predictable difference. Developmental Neuropsycbology, 5, 335-347.
BAYLES,K. A., TOMOEDA, C. K., KASNIAK, A. W. and TROSSET, M. W. 1991, Alzheimer’s disease
effects on semantic memory: loss of structure or impaired processing? Journal of Cognitive
Neuroscience, 3, 165-1 82.
BLANKEN, G., DITTMANN, J., HAAS,J-C. and WALLESCH, C-W. 1987, Spontaneous speech in senile
dementia and aphasia: Implications for a neurolinguistic model of language production.
Cognition, 27, 247-274.
BOLLER,F., BECKER, J. T., HOLLAND, A. L., FORBES, M. M., HOOD,P. C. and MCGONIGLE-GIBSON,
K. L. 1991, Predictors of decline in Alzheimer’s disease. Cortex, 27, S 1 7 .
CAPLAN, D. 1992, Language: Structure, Processing and Disorders (Cambridge, MA: MIT Press).
CARAMAZZA, A. and HILLIS,A. E. 1990, Where do semantic errors come from? Cortex, 26,95122.
CHERTKOW, H. and BUB,D. 1990, Semantic memory loss in dementia of Alzheimer’s type: What do
various measures measure? Brain, 113, 397-417.
COLLINS,A. M. and LOFTUS,E. F. 1975, A spreading-activation theory of semantic processing.
Psychological Review, 02, 407428.
COLLINS,A. M. and QUILLIAN, M. R. 1969, Retrieval time from semantic memory. Journal of Verbal
Learning and Verbal Behaviour, 8,24&247.
CONSTANTINIDIS, J., RICHARD,J. and DE AJURIAGUERRA, J. 1978, Dementias with senile plaques and
neurofibrillary changes. In A. D. Isaacs and F. Post (Eds), Studies in Geriatric Psythiatry
(Chichester: John Wiley and Sons).
DELL,G. S. and O’SWGHDHA,P. G. 1992, Stages of lexical access in language production. Cognition,
42, 287-314.
FAGLIONI,P. and BOTTI,C. 1993, How to differentiate retrieval from storage deficit: A stochastic
approach to semantic memory modelling. Cortex, 29,501-518.
FOLSTEIN,M. F., FOLSTEIN, S. E. and MCHUGH.P. R. 1975, “Mini-Mental State”. A practical guide
for grading the cognitive state of patients for the clinician. Journal o j Psychiatrjc Researti, 12,
189-198.
FRANCIS,W. N. and K U ~ E R A H., 1982, Frequency Ana& of English Usage: Lexicon and Grammar
(Boston: Houghton MifRin Co.).
FUNNELL, E. and HODGES, J. R. 1991, Progressive loss of access to spoken word forms in a case of
Alzheimer’s disease. Proceedings of the Royal Society, series B, 243, 173-179.
GARRETT, M.1992, Disorders of lexical selection. Cognition, 42, 143-180.
GILHOOLY, K. J. and LOGIE,R. H. 1980, Age of acquisition, imagery, concreteness, familiarity and
ambiguity measures for 1944 words. Behavioural Research Methods and Instrumentation, 12,
395-427.
GLASS,A. L. and HOLYOAK, K. J. 1975, Alternative conceptions of semantic theory. Cognition, 3,
313-339.
HACHINSKI, V. C., ILIFF,L. D., ZILHKA,E., Du BOULAY, G. H., MCALLISTER, V. L., MARSHALL, J.,
RUSSELL, R. W. R. and SYMON, L. 1975, Cerebral blood flow in dementia. ArchivesofNeurology,
32, 632-637.
Naming problems in dementia 371
HARLEY, T . A. 1933, Phonological activation of semantic competitors during lexical access in speech
production. Language and Cognitive Processes, 8, 291-309.
HARLEY, T. A. 1995, Connectionist models of anomia: A comment on Nickels. Language and Cognitive
Processes, 10, 47-58.
HARLEY, T. A. 1998, The semantic deficit in dementia: connectionist approaches to what goes wrong
in picture naming. Aphasiology, 12, 299-318.
HARLEY,T. A. and MACANDREW, S. B. G. 1992, Modelling paraphasias in normal and aphasic
speech. Proceedings of the 14th Annual Conference of the Cognitive Science Societ_y (Hillsdale, NJ :
Erlbaum), pp. 378-383.
HARLEY,T. A. and MACANDREW,S. B. G. 1995, Interactive models of lexicalisation: Some
constraints from speech error, picture naming, and neuropsychological data. In J. P. Levy, D.
Rairaktaris, J. A. Bullinaria and P. Cairns (Eds) Connectionist Models o j Memar), and Language
(London: UCL Press), pp. 311--331.
HART,S., SMITH,C. M. and SWASH,M. 1988, Word fluency in patients with early dementia of
Alzheimer type. British Journal of Clinical P:ychology, 27, 115-124.
HIER, D. B., HAGENLOCKER, K. and SHINDLER,A. G. 1985, Language disintegration in dementia:
Effects of etiology and severity. Brain and Language, 25, 117-133.
HINTON,G. E. and SHALLICE, T. 1991, Lesioning an attractor network: Investigations of acquired
dyslexia. Psychological Review, 98, 74-95.
HODGES, J . R., SALMON, D. P. and BUTTERS, N. 1991, The nature of the naming deficit in Alzheimer’s
and Huntington’s disease. Brain, 114, 1547-1 558.
HODGES,J . R., SALMON,D. P. and BUTTERS,N. 1992, Semantic memory and impairment in
Alzheimer’s disease : Failure of access or degraded knowledge? Neuropsychologia, 30, 301-314.
HUFF,F. J., CORKIN,S. and GROWDON, J. H. 1986, Semantic impairment and anomia in Alzheimer’s
disease. Brain and Language, 28, 235-249.
JESCHNIAK, J . D. and LEVELT,W. J. M. 1994, Word frequency effects in speech production:
Retrieval of syntactic information and phonological form. Journal of Experimental Psychology:
Learning, Memory and Cognition, 20, 824-843.
JONES,G. V. 1989, Back to Woodworth: Role of interlopers in the tip-of-the-tongue phenomenon.
hfemor), & Cognition, 17, 69-76.
KEMPEN,G . and HUIJBERS, P. 1983, The lexicalization process in sentence production and naming:
Indirect election of words. Cognition, 14, 185-209.
KERTESZ, A., APPELL,J. and FISMAN, M. 1986, The dissolution of language in Alzheimer’s disease.
Canadian Journal of Neurological Sciences, 13, 41 5-418.
KIRSHNER,H. S., WEBB, W. G. and KELLY,M. P. 1984, The naming disorder of dementia.
Neuropsychologia, 22, 23-30.
LEVELT, W. J. M. 1992, Accessing words in speech production: Stages, processes and representations.
Cognition, 42, 1-22.
MARTIN,A. and FEDIO,P. 1983, Word production and comprehension in Alzheimer’s disease: The
breakdown of semantic knowledge. Brain and Language, 19, 124-141.
MARTIN,N., WEISBERG, R. W. and SAFFRAN, E. M. 1989, Variables influencing the occurrence of
naming errors: Implications for models of lexical retrieval. Journalof Memar),and Language, 28,
462 -485.
MCCLELLAND, J. L. and RUMELHART, D. E. 1985, Distributed memory and the representation of
general and specific information. Journal of Experimental Psychology :General, 114, 159-1 88.
MCCLOSKEY, M. E. and GLUCKSBERG, S. 1979, Decision processes in verifying category membership
statements: Implications for models of semantic memory. Cognitive Psychology, 11, 1-37.
MCKHANN, G., DRACHMAN, D., FOLSTEIN, M., KATZMAN, R., PRICE,D. and STADLAN, E. M. 1984,
Clinical diagnosis of Alzheimer’s disease : Report of the NINCDS-ADRDA Work Group
under the auspices of Department of Health and Human Services Task Force on Alzheimer’s
Disease. Neurology, 34, 939-944.
MEYER,D. E. 1970, O n the representation and retrieval of stored semantic information. Cognitive
P y c h o l ~ 1,~ ,242-300.
MILLER,E. 1979, Memory and ageing. In M. M. Gruneberg and P. E. Morris (Eds) Applied Problems
in MemorJ, (London: Academic Press).
MILLER,E. and HAGUE,F. 1975, Some characteristics of verbal behaviour in senile dementia.
I’<ychological Medicine, 5, 255-259.
NEARY, D., SNOWDEN, J. S., BOWEN,D. M., SIMS,N. R., MANN,D. M. A., BENTON,J. S., NORTHEN,
372 A . J. Astelland T. A . Harley

B., YATES,P. 0. and DAVISON,A. N. 1986, Neuropsychological syndromes in presenile

dementia due to cerebral atrophy. Journalof Neurology, Neurosurgery and Psycbiaty, 49,163-1 74.
NEBES,R. D. 1989, Semantic memory in hlzhcimcr’s discasc. Psychological Bulletin, 106, 377-394.
NEILS,J., BRENNAN, M. M., COLE,M.,, BOLLER,F. and GERDEMAN, B. 1988, The use of phonemic
cueing with Alzheimer’s disease patients. Neuropsychologia, 26, 351-354.
NICHOLAS, M., OBLER,L. K., ALBERT,M. L. and GOODGLASS, H. 1985, Lexical retrieval in healthy
aging. Cortex, 21, 595-606.
NICHOLAS, M., OBLER,L. K., Au, R. and ALBERT,M. L. 1996, O n the nature of naming errors in
aging and dementia: A study of semantic relatedness. Brain and Language, 54, 184-195.
OLDFIELD,R. C. and WINGFIELD,A. 1964, The time it takes to name an object. Nature, 202,
1031-1032.
OLDFIELD, R. C. and WINGFIELD,A. 1965, Response latencies in naming objects. Quarterb Journalof
Experimental Psychology, 17, 273-281.
PAIVIO,A., YUILLE,J. C. and MADIGAN,S. A. 1968, Concreteness, imagery and meaningfulness
values for 925 words. Journal of Experimental Psycbology, Monograph Supplement, 76 (Part 2),
1-25.
QUINLAN, P. T. 1992, Oxford Psycholinguistic Database (Oxford : Oxford University Press).
RAW, B. and CARAMAZZA, A. 1993, O n the distinction between deficits of access and deficits of
storage: A question of theory. Cognitive Neuropycbology, 10, 113-141.
RIDDOCH,M. J., HUMPHREYS, G. W., COLTHEART, M. and FUNNELL, E. 1988, Semantic systems or
system? Neuropsychological evidence re-examined. Cognitive Neuropychology, 5, 3-25.
RIPICH,D. N., VERTES,D., WHITEHOUSE, P., FULTON, S. and EKELMAN, B. 1991, Turn-taking and
speech act patterns in the discourse of senile dementia of the Alzheimer’s type patients. Brain
and Language, 40, 330-343.
ROCHFORD, G. 1971, A study of naming errors in dysphasic and in demented patients. Neuropsychologia,
9, 437-443.
ROSCH,E. 1973, Natural categories. Cognitive Psycbology, 4, 677-688.
SAFFRAN, E., SCHWARTZ, M. F., LINNEBARGER, M., MARTIN,N. and BOCHETTO, P. 1989, Philadelphia
Comprehension Battery for Aphasia (unpublished).
SHALLICE, T. 1988, From Neuropsycbology t o Mental Structure (Cambridge: Cambridge University
Press).
SHALLICE, T. 1993, Multiple semantics : Whose confusion? Cognitive Neuropsychology, 10, 251-261.
SKELTON-ROBINSON, M. and JONES,S. 1984, Nominal dysphasia and the severity of senile dementia.
British Journal of Psychiatry, 145, 168-171,
SMITH,E. E., SHOBEN,E. J. and RIPS, L. J. 1974, Structure and process in semantic memory: A
featural model for semantic decisions. Psychological Review, 81, 214-241.
SOMNERS, L. M. and PIERCE,R. S. 1990, Naming and semantic judgements in dementia of the
Alzheimer’s type. Apbasiology, 4, 573-586.
STUSS,D. T. and BENSON,D. F. 1986, The Frontal Lobes (New York: Raven Press).
TIPPETT,L. J. and FARAH, M. J. 1994, A computational model of naming in Alzheimer’s disease:
Unitary o r multiple impairments? Neuropsycbology, 8, 1-11.
TOGLIA,M. P. and BATTIG,W. R. 1978, Handbook ofSemantic Word Norms (New York: Erlbaum).
TULVING, E. 1972, Episodic and Semantic Memory. In E. Tulving and W. Donaldson (Eds) The
Organization of Memory (New York: Academic Press).
WARRINGTON, E. K. and SHALLICE, T. 1979, Semantic access dyslexia. Brain, 102, 4 3 4 3 .
Naming problems in dementia 373

Appendix. Naming target and substitute pairs-semantic,

visual and TOTS-with frequency and semantic relatedness
ratings. Unrelated responses and perseverations are not included

P T TW Freq Im Sub Freq Im Sem-re1

P12 vs Banjo 1 N.A. Guitar 2 N.A. 3.571

P1 vs Bear 24 5.72 Wolf 9 610 2.92
P11 sc Bus 42 N.A. Van 2 5.72 3.142
P3 vs Crab 2 N.A. Elephant 18 616 207 1
P11 v Crown 19 6.02 Fence 46 611 1.142
P3 v Crown 19 6.02 Hair 160 5.80 3.142
P11 vs Goat 8 5.85 Bullock N.A. N.A. 2714
P10 vs Goat 8 5.85 cow 46 6.32 3.5
P11 vs Goat 8 5.85 cow 46 6.32 3.5
P8 VS Goat 8 585 Deer 13 6.24 2.785
P7 TR Goat 8 5.85 Dog 147 636 2928
P7 TR Goat 8 5.85 Donkey 1 N.A. 3.214
P7 TR Goat 8 5.85 Horse 203 6.24 3.142
P9 v Leash/lead 4 N.A. Horseshoe N.A. N.A. 1.214
P2 v Leash/lead 4 N.A. Rope 19 5.96 3142
PI v Leash/lead 4 N.A. Pair of clippers N.A. 1.ooo
P12 v Leash/lead 4 N.A. Saddle 26 5.78 1.5
P7 Au Money 275 6.04 Crockery N.A. N.A. 1 el42
P7 v Money 275 6.04 Plates 44 5.27 1.357
P8 V Money 275 6.04 Plates 44 527 1.357
P10 v Money 275 6.04 Saucers 2 N.A. 1.214
P5 SA Money 27 5 604 Sovereigns 16 N.A. 3857
P8 V Money 275 604 Trays 21 5.50 1.000
P4 vs octopus 1 N.A. Crab 2 N.A. 3.428
P7 vs octopus 1 N.A. Crab 2 N.A. 3.428
P1 TR octopus 1 N.A. Crab 2 N.A. 3.428
P12 v octopus 1 N.A. Face 379 581 1 .ooo
P4 TR octopus 1 N.A. Lobster 1 6.30 3214
P5 vs octopus 1 N.A. Man-o-war N.A. N.A. 2714
P4 TR octopus 1 N.A. Oyster 16 5.21 2642
P3 vs octopus 1 N.A. Spider 2 5.97 2714
P7 vs octopus 1 N.A. Toad 4 5.91 2214
P7 vs Owl 6 5.95 Parrot 2 N.A. 3.428
P1 Au Owl 6 595 Parrot 2 N.A. 3.428
P2 vs Owl 6 5.95 Pigeon 5 610 3.142
P10 Au Parrot 2 N.A. Bird 83 614 4000
P11 su Parrot 2 N.A. Bird 83 6.14 4000
P9 su Parrot 2 N.A. Bird 83 6.14 4000
P5 vs Parrot 2 N.A. Bird of paradise N.A. N.A. 3.285
P8 SC Parrot 2 N.A. Owl 6 5.95 3.214
P5 vs Parrot 2 N.A. Parakeet 1 N.A. 3571
P6 TR Parrot 2 N.A. Polly N.A. N.A. 2214
P6 TR Parrot 2 N.A. Polyanthus N.A. N.A. 1.071
P8 TU Peacock 2 N.A. Bird 83 6.14 4000
P6 TU Peacock 2 N.A. Bird 83 614 4000
PO v Peacock 2 N.A. Flower 78 618 1.285
P1 vs Peacock 2 N.A. Pheasant 3 N.A. 3.571
P9 v Road 262 609 Dragonfly N.A. N.A. 1 *ooo
P1 v Road 262 609 Gnat N.A. N.A. 1 a000
P5 v Road 262 6.09 Hill 119 607 2214
P5 v Road 262 6.09 Leaf 33 6.08 1 .ooo
374 A. J . Astelland T. A. Had9

Appendix. (cont.)
P T TW Freq Im Sub Freq Im Sem-re1

P7 Au Steak/chop 14 6.47 Beef 32 6.25 3.785

P8 SC Steak/chop 14 6.47 Beef 32 6.25 3.785
P8 Au Steak/chop 14 6.47 Lamb 14 6.14 3412
P7 VS Steak/chop 14 647 Meat 57 6.18 3857
P1 SU Steak/chop 14 647 Meat 57 618 3.857
P10 SU Steak/chop 14 647 Meat 57 618 3857
P12 SA Steak/chop 14 647 Mutton 8 484 2.928
P12 SA Steak/chop 14 647 Pig 14 6.35 2142
P9 V Steak/chop 14 6-47 Sole 7 462 2142
P10 V Strawberry 2 N.A. Rose 18 6.23 1.214
P10 V Strawberry 2 N.A. Sweet 2 493 2428
P3 Au Table 242 5.82 Stool 8 5.84 3.285
P7 VS Table 242 5.82 Stool 8 5.84 3.285
P1 SA Tie 27 5.51 Collar 14 5.82 3857
P11 SA Tie 27 5.51 Collar 14 582 3.857
P7 Au Tie 27 5.51 Collar 14 5.82 3.857
P1 SA Tie 27 5.51 Shirt 29 6.12 3.214
P11 SA Tie 27 5.51 Shirt 29 6.12 3.214
P12 SC Trombone N.A. N.A. Banjo 1 N.A. 2214
P2 TU Trombone N.A. N.A. Bugle N.A. N.A. 3.571
P7 TR Trombone N.A. N.A. Horn 33 5.66 3.214
P4 VS Trombone N.A. N.A. Saxophone 4 6.02 3357
P1 SC Trombone N.A. N.A. Trumpet 6 628 3571
P7 SC Trombone N.A. N.A. Trumpet 6 628 3.571
P8 SC Trombone N.A. N.A. Trumpet 6 628 3.571
P10 SC Trombone N.A. N.A. Trumpet 6 6.28 3.571
P5 SC Wheelbarrow N.A. N.A. Weedcart N.A. N.A. 3-357

Key: P = participant; T = type oferror; TW = target word; Freq = frequency; Im = imageability;

Sub = substitute; Sem-re1 = semantic relatedness rating; N.A. = no rating available.
Error types: Au = Augmented response; T R = resolved T O T ; TU = unresolved TOT; V =
visual; VS = ambiguous visual/semantic; SC = semantic co-ordinate; SU = superordinate; SA =
semantic associate.