DOI: 10.1093/brain/awh264
Brain (2004), 127, 2299–2306
The effects of ageing and Alzheimer’s disease
on semantic and gender priming
Rosa Manenti,1,2 Claudia Repetto,1 Simone Bentrovato,3 Alessandra Marcone,1 Elizabeth Bates,4,
and Stefano F. Cappa1
1
Cognitive Neuroscience Center, Vita Salute University
and San Raffaele Scientific Institute, Milan, 2IRCCS
S. Giovanni di Dio-Fatebenefratelli, Brescia, 3Istituto di
Psicologia, Università ‘La Sapienza’, Rome, Italy and
4
Center for Research in Language, University of California
at San Diego, San Diego, USA
y
Correspondence to: Stefano F. Cappa, MD, Vita Salute
University and San Raffaele Scientific Institute, DIBIT,
Via Olgettina 58, 20132 Milano, Italy
E-mail [email protected]
y
This paper is dedicated to the memory of Elizabeth Bates
Summary
Normal ageing as well as age-associated pathological conditions, such as Alzheimer’s disease, are associated with
modifications of language processing. In particular, an
impaired performance in semantic tasks, associated
with relatively spared syntactic processing, has been suggested to be the hallmark of the language disorder of
Alzheimer’s disease. The present experiment tests
semantic and syntactic aspects of language processing at
the same time, using an on-line paradigm, in patients with
Alzheimer’s disease, compared with elderly and young
controls. Normal ageing was associated with a profile of
performance, which was slowed but qualitatively comparable with that of young controls. Both gender agreement
and congruent sentential semantics resulted in facilitation
relative to baseline in young and elderly controls, with no
significant interference effects of incongruent grammatical and semantic information. In contrast, Alzheimer’s
disease patients presented both facilitation and interference effects. These findings suggest that interference
effects are amplified by dementia, and may result from
defective inhibitory processes due to Alzheimer’s disease
pathology.
Keywords: priming; ageing; Alzheimer’s disease; gender; semantics
Abbreviations: EC = elderly controls; RT = reaction time; YC = young controls.
Received October 14, 2003. Revised February 16, 2004. Second revision June 8, 2004. Accepted June 10, 2004.
Advanced Access publication August 11, 2004
Introduction
There is ample evidence of age-associated changes in language processing on a variety of tasks (MacKay and Abram,
1996). Language impairment is also one of the features of cognitive impairment in the most common form of degenerative
dementia, Alzheimer’s disease (Ramage and Holland, 2001).
According to one influential proposal, a common nonlanguage-specific mechanism, i.e. a disorder in inhibitory
processes, may account for the performance modifications
in language tasks observed both in normal ageing and in
the early stages of Alzheimer’s disease (Hasher and Zacks,
1988; Balota and Faust, 2001). At the neuropsychological
level, this could be due to the presence of executive dysfunction, associated with prefrontal cortex involvement, in ageing
and early Alzheimer’s disease. An alternative account is that
the language changes in dementia reflect a direct involvement
of language areas, resulting in loss or defective availability of
Brain Vol. 127 No. 10
#
linguistic knowledge. In particular, a lexical–semantic impairment, associated with preserved phonological and syntactic performance, has been considered to be specific for
Alzheimer’s disease, and reflecting the relative sparing of
prefrontal areas up to the advanced stages of disease (see,
for example, Ullman, 2001).
The results of imaging studies provide partial support for
the latter hypothesis, as they do not indicate a specific involvement of prefrontal areas in normal ageing and early
Alzheimer’s disease. Investigations of brain morphology in
ageing normal subjects have, in general, indicated the presence of a global linear decline in brain volume. Areas of
accelerated decline appear to be the hippocampal regions
(Scahill et al., 2003), as well as the insular and parietal cortex
(Good et al., 2002), rather than the prefrontal cortex. In
the case of early Alzheimer’s disease, both structural and
Guarantors of Brain 2004; all rights reserved
2300
R. Manenti et al.
functional imaging indicate a relative sparing of prefrontal
cortex in the early stages of disease (for a review see Perani
and Cappa, 2001; for recent evidence see Thompson et al.,
2003).
In order to assess independently the effects of ageing and
disease on a specific test, several methodological factors
should be taken into account. First, it is necessary to compare
the performance of young controls with samples of healthy
elderly subjects and of patients with a diagnosis of probable
Alzheimer’s disease using the same experimental procedure.
Secondly, the nature of the task should be taken into account.
On-line processing tasks, such as priming paradigms, are
ideally suited to assess the modifications of cognitive processing associated with normal ageing and early dementia. Priming is based largely on automatic processes, allowing the
evaluation of actual linguistic competence. Semantic priming
is of particular interest, given the general agreement that
performance in tasks requiring access to semantic knowledge
is affected from the early stages of Alzheimer’s disease.
Whether the impairment represents a loss of semantic knowledge or defective access to that knowledge is a matter of
debate (see, for example, Hodges et al., 1992; Ober and
Shenaut, 1999). Semantic priming studies of normal ageing
typically show that performance on lexical processing tasks is
slowed, but semantic priming effects appear to be preserved
(Myerson et al., 1997). On the other hand, semantic priming
has been investigated extensively in Alzheimer’s disease,
with contradictory results (for a review see Ober, 2002).
Priming effects can also be induced by syntactic information. For example, pronominal modifiers, such as the article,
can also affect the processing of the nouns they modify. An
article matching the noun in grammatical gender facilitates
lexical access, while an incongruent article may interfere with
the same process. The effects of grammatical gender and of
semantic information (sentence context) on word reading
have been investigated recently by Bentrovato et al. (2003).
The results indicated the presence of a large, interacting facilitation effect of both context and gender in young individuals.
This was taken to indicate that normal subjects could take
advantage of both sources of information, with no significant
interference when they are both discordant.
In the present experiment, we have used the same sentential
priming paradigm in order to evaluate at the same time the
effect of semantic and grammatical information on word
reading in elderly normal subjects and in a sample of early
Alzheimer’s disease patients. The aim of the study was to
assess the impact of ageing and Alzheimer’s disease on the
on-line effects of congruent and incongruent semantic and
grammatical information.
Methods
Subjects
Three groups of subjects, all native Italian speakers, participated in
the experiment: 25 university students (12 females and 13 males;
mean age 22.8 years, range 19–27), 26 normal ageing subjects (nine
males and 17 females, mean age 73.58 years, range 63–87) and 26
patients with the diagnosis of Alzheimer’s disease (nine males and 17
females, mean age 76.23 years, range 67–84). The diagnoses were
made by a staff of neurologists of the San Raffaele Turro Hospital
(Milan), following the National Institute of Neurological and Communicative Disorders (NINCDS) and Alzheimer’s Disease and
Related Disorders Association (ADRDA; McKhann et al., 1984)
criteria. Patients with other neurological, metabolic and psychiatric
pathologies were excluded. All the patients recruited received a
complete neurological and neuropsychological evaluation. Informed
consent was required from all participants, according to the guidelines of the local ethics committee. Statistical analyses showed that
there was no difference between patients and normal ageing subjects
based on age and education [age, t(50) = 1.751; P = 0.087; instruction
t(50) = 0.459; P = 0.649]; the same two samples, however, had
significantly different Mini-Mental State Examination (MMSE)
scores [t(50) = 10.66; P < 0.0001]. Table 1 summarizes the descriptive data.
Materials and procedure
The main test is the Italian Word Reading Test (Bentrovato et al.,
2003). Every stimulus is a brief two-sentence narrative in Italian,
presented orally, with a target word presented visually in one of the
two sentences (the target position was varied within and between
sentences).
Word targets
The 110 target words for reading are the transcription of the names
produced by the majority of subjects for the corresponding picturenaming task used by Bentrovato et al. (1999): everyday objects
(Snodgrass and Vanderwart, 1980; Dunn and Dunn, 1981; Kaplan
et al., 1983; Abbate and LaChapelle, 1984) chosen from a list of 520
drawings from an International Picture-Norming Study (Bates et al.,
2000). The 110 target words have the following properties: 56 masculine (44 end with the phonologically transparent vowel o, 12 with
the phonologically ambiguous vowel e) and 54 feminine (46 end with
the phonologically transparent vowel a, eight with the phonologically
Table 1 Details of groups
YC
Mean
SD
EC
Alzheimer’s disease
Age (years)
Education (years)
Age (years)
MMSE
Education (years)
Age (years)
MMSE
Education (years)
22.56
2.27
16.88
2.07
73.58
6.17
27.04
1.34
5.35
1.70
76.23
4.65
19.69
3.25
5.65
2.97
MMSE = Mini-Mental State Examination score.
The effects of ageing and Alzheimer’s disease on priming
ambiguous vowel e). The words were written in black on a white
background, in lower case letters. The font was Times New Roman
with a font size of 48; the mean word length in letters was 6.5 (range
4–11), while the mean length in syllables was 2.7 (range 2–4). The
target names represent common objects: six of these correspond to
body parts or human activities [i.e. ‘dito’ (finger) or ‘marinaio’
(sailor)] while another 15 are names of animals, with fixed grammatical gender independent of biological sex (i.e. ‘cammello’—a masculine noun that means camel; ‘gallina’—a feminine noun that
means chicken).
Auditory sentence contexts
A total of 220 two-sentence contexts were constructed: 110 for the
2 2 factorial gender semantics design and another 110 as neutral
context in order to avoid semantic and gender cues. The contexts
were recorded in a soundproof room and transferred onto a
Macintosh computer as individual SoundEdit 1.0 files. They were
converted into individual Psyscope files (Cohen et al., 1993) for
experimental presentation. The average length of the sentences
was 11.24 s (range: 8.13–15.37).
The semantic cue was given by the meaning of the sentence, while
the gender cue was given by the article [for half the items indefinite
(‘uno–una’), and for remaining items definite (‘il–la’)]. The article
was the last word before the presentation of the target. These auditory
contexts represent the primes.
Prime–target pairs
As a first step, 110 base pairs were constructed: these were obtained
by pairing a prime context and a target word in order to satisfy both
semantic (S) and gender (G) congruence (with the target placed in
either the first or the second sentence). This condition is called
GþSþ . In order to create a semantically incongruent version, a
target word with congruent gender but incongruent meaning was
assigned to the auditory context (GþS). In the condition of
semantic congruence and gender incongruence (GSþ ), the pairs
were the same as the ones used in GþSþ , but the article before the
target was switched to the opposite gender. The last experimental
condition, called GS, was constructed by assigning to the context
a target that was a highly implausible completion on semantic
grounds, and by using an article that has an opposite gender to
both the actual (incongruent) target word and the word that would
2301
have served as the semantically congruent target. The neutral contexts (equivalent in length and complexity to the experimental contexts: see also Wicha et al., 1997) served as a neutral baseline, used to
assess whether any of the gender semantics conditions resulted in
facilitation or inhibition of the target reading. This condition was
called N/N (see Table 2 for examples).
List construction
Five lists were prepared, in order to insure that no single subject ever
heard the same sentence or saw the same word twice. In the first list,
the items were assigned in a quasi- random way to the five experimental conditions (GþSþ , GþS, GSþ , GS, N/N), until each
of the five conditions reached 22 items without repetitions. At the
end, the items were rotated across the five conditions: each word
target might occur in GþSþ in list 1, Gþ S in list 2, GSþ in list
3, GS in list 4, and N/N in list 5.
Procedure
The subjects were seated in front of a 15 inch computer screen in a
quiet testing room, wearing headphones with a microphone, that
were connected to the sound amplifier port of the computer.
Response times were collected in milliseconds through the Carnegie
Mellon Button Box, a measuring device including a voice key
with an external time crystal and 1 ms resolution, designed for
use with a Macintosh Computer, connected to the modem port of
the PowerBook.
The experimenter hand-recorded all naming errors on a score
sheet during testing. The subjects were assigned randomly to one
of the five lists based on their participant number.
At the beginning of the session, the participant saw the instructions on the screen; every experimental session was preceded by a
practice session, in which 10 context–target pairs were presented,
two for each experimental condition.
In the experimental session, the items were presented randomly, in
a continuous way. First, the participant saw the fixation point ( þþ )
in the centre of the screen and heard the first part of the auditory
context; at some point in the sequence, the sentence was halted and in
the place of the fixation point the target appeared. The participant was
instructed to read the word as soon as possible: when the voice was
detected by the voice key, the word disappeared; after 500 ms, the
second part of the context was heard. The time limit to read the word
was 5 s: after this period the word disappeared even if the subject had
Table 2 Examples of test sentences
Prime
Target
Quando vado a letto
prima di addormentarmi
leggo sempre
(When I go to bed before
falling asleep I always read )
Cond.
per questo mia mamma mi ha
regalato una collezione
di romanzi gialli
(for this reason, my mother gave
me a collection of murder mysteries)
Un (a MASC)
Un (a MASC)
Una (a FEM)
Una (a FEM)
LIBRO (BOOK)
TOPO (MOUSE)
LIBRO (BOOK)
TOPO (MOUSE)
Silvia ha fatto un test in
inglese, in cui doveva ripetere
(Sylvia took a test in English
in which she had to repeat)
Prime
LIBRO (BOOK)
GþSþ
GþS
GSþ
GS
per cinque volte. L’insegnante ha detto
che è l’unico modo per migliorare la
pronuncia
(five times. The teacher said that was
the only way to improve her pronunciation)
NN
2302
R. Manenti et al.
yet not read it. The interval between two subsequent trials was 3 s for
the young controls (YC) and 5 s for the elderly controls (EC) and for
Alzheimer’s disease patients. This decision was taken in order to
avoid fatigue effects. The entire experimental session lasted 45 min.
slowing. The corrected data are reported in Table 4. The 2 2
ANOVA on the corrected data replicated the results of the
analysis on the raw RTs. Using the corrected RTs, we then ran
three separate ANOVAs with one between-subjects factor
(group). In all three comparisons (YC–EC, EC– Alzheimer’s
disease, YC– Alzheimer’s disease), both factors showed a
Results
The reaction time (RT) analyses were based only on correct
responses. A response was discarded if any irrelevant noise
was recorded instead of the answer, if the microphone failed
to record the response or if the participant gave a wrong response. Following these criteria, we excluded 2.27% of the
trials for YC, 3.74% for EC (93% of them were due to recording
failures) and 10.98% for Alzheimer’s disease patients (85% of
them recording failures). The number of wrong responses was
not significantly different in the three groups.
The RT data in the five experimental conditions are summarized in Table 3.
In order to assess the effects of matching gender and
semantic congruence on the raw RT data, we performed a
2 2 within-subjects ANOVA over participants. The results
in YC have already been reported by Bentrovato et al. (2003),
and indicate significant effects of both gender [F(1,24) =
18.564; P = 0.0002] and semantic context [F(1,24) =
28.664; P < 0.0001], as well as a significant interaction
[F(1,24) = 4.49; P = 0.0446], due to the larger effect of gender
agreement in the semantically congruent condition (Fig. 1).
The results were the same in the case of EC subjects [gender,
F(1,25) = 17.861; P = 0.0003; semantics, F(1,25) = 26.199;
P < 0.0001; interaction, F(1,25) = 4.562; P = 0.0427] (Fig. 2).
In contrast, in Alzheimer’s disease patients, while the two
main effects were significant [gender, F(1,25) = 23.786;
P < 0.0001; semantics, F(1,25) = 23.066; P < 0.0001],
there was no significant gender semantics interaction
[F(1,25) = 0.226; P = 0.6385] (Fig. 3).
In order to compare the performance of the three groups
directly, the presence of a significant slowing in RT associated
with physiological ageing should be taken into consideration.
We thus correlated the RTs of YC and EC in each condition
with their ages. Since the equation calculated using the RTs in
the NN condition was quite similar to the one obtained using
all RTs (NN, coefficient = 6.766, t = 8.377, P < 0.001; all,
coefficient = 6.716, t = 10.123, P < 0.001), and the interaction
between conditions and age was not significant (F = 2.245, P >
0.05), we decided to use the equation calculated on the NN
condition to correct all the performance data for physiological
Fig. 1 Reaction times in YC, according to the presence of an
incongruent () or congruent (þ) semantic (S) of grammatical (G)
context. The NN line represents the neutral baseline condition.
Both gender agreement and a congruent semantic context
results in faster reading. The significant interaction indicated the
presence of a larger effect of gender agreement in the semantically
congruent condition.
Fig. 2 In the group of EC, the results were comparable to YC.
Significant effects of gender agreement and congruent
semantic context.
Table 3 Reaction time data (ms)
Alzheimer’s disease
EC
YC
Mean
SD
Mean
SD
Mean
SD
GþSþ
GþS
GSþ
GS
NN
915.74
361.95
767.46
218.49
456.44
82.16
1104.54
552.23
848.17
273.92
484.92
90.81
1018.53
419.97
825.64
269.61
480.43
92.83
1229.43
596.24
876.93
268.22
497.99
99.42
1069.92
471.99
861.45
244.68
501.68
119.29
The effects of ageing and Alzheimer’s disease on priming
2303
Table 4 ‘Corrected’ reaction time data (ms)
Alzheimer’s disease
EC
YC
Mean
SD
Mean
SD
Mean
SD
GþSþ
GþS
GSþ
GS
NN
717.64
196.97
602.39
142.94
632.78
53.45
885.47
354.72
683.81
170.52
661.99
62.33
816.95
218.48
665.01
164.09
661.13
65.84
1024.70
385.61
711.80
181.68
674.97
70.44
874.49
294.55
684.24
202.57
676.78
65.19
Fig. 3 In the AD group, only the two main effects were significant,
with no interaction.
significant interaction with group [YC–EC: gender, F(1,49) =
30.427, P < 0.001; semantics, F(1,49) = 40.480, P < 0.001;
group gender, F(1,49) = 4.282; P = 0.045; group semantics, F(1,49) = 9.987, P = 0.004; EC–AD: gender,
F(1,50) = 40.071, P < 0.001; semantics, F(1,50) = 36.483,
P < 0.001; group gender, F(1,50) = 8.055; P = 0.008; group
semantics, F(1,50) = 9.036, P < 0.005; YC–AD: gender,
F(1,49) = 32.365, P < 0.001; semantics, F(1,49) = 26.423,
P < 0.001; group gender, F(1,49) = 16.083; P < 0.001;
group semantics, F(1,49) = 16.796, P < 0.001].
In order to analyse these differences in further detail, and
to determine the direction of priming, we compared the raw
RTs of each condition with the neutral baseline, within each
group (Fig. 4). In the case of YC, a significant facilitation was
observed for the GþSþcondition [F(1,24) = 6.509; P = 0.0139],
while the GS condition failed to induce a significant interference [F(1,24) = 0.074; P > 0.05] (data reported in Bentrovato
et al., 2003). The two remaining conditions were also not significantly different from baseline [GþS, F(1,24) = 0.27; P >
0.05; GSþ, F(1,24) = 1.345; P > 0.05]. In the case of EC, the
raw RT data indicated a significant facilitation in GþSþ[F(1,25)
= 47.708; P = 0.0001], no significant interference in GS
[F(1,25) = 0.727; P > 0.05] and no significant differences for the
two remaining conditions [GþS, F(1,25) = 1.495; P > 0.05;
GSþ, F(1,25) = 4.231; P > 0.05]. Only for Alzheimer’s disease patients, there were both significant facilitation effects
with GþSþ [F(1,25) = 21.625; P < 0.0001] and interference
Fig. 4 The bar graphs reports the difference of each condition to
the neutral baseline, within each group, thus indicating the
presence of facilitation or inhibition of word reading. In the case
of YC and EC, there was only a significant facilitation effect
induced by the condition of combined grammatical and semantic
congruence (GþSþ). Only AD patients, had both a significant
facilitation effect for GþSþ, and a significant interference effect
in the case of the combined incongruence (GS).
effects with GS [F(1,25) = 10.731; P = 0.0031]. None of
the other facilitation or interference effects reached significance [GþS, F(1,25) = 1.146; P > 0.05; GSþ, F(1,25) =
2.328; P > 0.05].
Discussion
The present findings indicate that grammatical and semantic
cues provide parallel information, which can be used quickly
to predict the target word. Not surprisingly, the combination
of matching gender and semantic congruence yields the
largest facilitation effects in both control groups, as well as
in Alzheimer’s disease patients. The significant interaction of
gender with the semantic context clearly indicates a role for
the integration of all sorts of available information.
Healthy ageing does not appear to affect this general pattern
of results. While the RTs were slower for ECs, the profile of
effects remained the same, including the presence of a significant gender semantics interaction.
In contrast to these findings for both ECs and YCs, performance of the Alzheimer’s disease subjects was associated
with two significant modifications. First, the interaction
2304
R. Manenti et al.
between gender and context was not significant. A second,
more striking difference from the findings in both young and
elderly normal subjects was the appearance of an interference
effect of the combined incongruence.
This pattern of results highlights the contribution of on-line
tasks to the assessment of language processing in the case
of both normal and neurologically impaired populations. In
contrast to traditional metalinguistic methods, such as, for
example, picture naming or sentence verification, these methods do not require explicit judgements about the structure or
meaning of the test material. The latter feature is particularly
attractive in the case of conditions associated with modifications of multiple cognitive abilities, such as dementia.
The preserved facilitation effect of sentence context is in
line with previous findings. Semantic priming has been
applied extensively to the investigation of lexical–semantic
processing. The most widely used paradigms assess the effect
of a preceding related or unrelated single word on tasks such
as lexical decision or word reading. In general, studies of
healthy ageing have reported an overall slowing of RTs,
with preserved priming effects (Myerson et al., 1992).
Semantic priming has been applied extensively to the
investigation of the status of semantic memory in Alzheimer’s
disease. The results may prima facie appear highly inconsistent, as priming effects have been reported to be reduced
(Ober and Shenaut, 1988), normal (Nebes et al., 1984) or
increased (Chertkow et al., 1989). While some of the discrepancies are related to differences in methodology, there is
recent evidence that the heterogeneity of the effects may
be due in part to the stage of disease, and to the severity of
semantic impairment as measured by off-line tasks (Giffard
et al., 2002). Recently, Giffard et al. (2002) have argued on
the basis of longitudinal data that the presence of decreased or
increased priming depends on relatedness conditions. An
early increase is observed only in the case of coordinate priming (tiger–lion), while attribute priming (tiger–stripe) progressively declines from normal levels. The latter finding was
considered to reflect the dynamics of semantic memory deterioration. When specific attributes begin to be lost, the loss of
distinctions between close concepts results in increased coordinate priming. On the other hand, the same mechanism
reduces attribute priming from the start. With the progression
of semantic deterioration, priming progressively reduces in
both conditions. Sentence context effects have been investigated less extensively in ageing and dementia. A facilitation of
word processing has been observed when a word is a plausible
completion of a sentence (Fischler and Bloom, 1985). Conversely, an incongruous word is processed more slowly, in
particular in the case of young readers (West and Stanovich,
1978). In the case of lexical decision, the nature of the context
modulates the facilitation effect. With high-constraint sentences, only expected congruous words are facilitated. With
low-constraint contexts, both expected and unexpected congruous words enjoy facilitation effects (Schwanenflugel and
LaCount, 1988). A small increase in the interference effect on
picture naming has been reported in healthy elderly subjects
(Roe et al., 2000). In the same study, facilitation was unaffected
by age. With a similar task, a reduction of the neurophysiological N400 effect has been reported in normal elderly subjects
(Cameli and Phillips, 2000). This has been attributed to defective inhibition of irrelevant semantic information in working
memory, or ‘diffuse semantic activation’ (Miyamoto et al.,
1998). In the case of Alzheimer’s disease, Nebes and Brady
(1991), using a task in which the patient had to judge whether a
word was a sensible completion of a sentence, reported a relative preservation of the facilitation effect of appropriate
context on word processing. In a recent study with eventrelated potentials, Schwartz et al. (2003) found delayed but
preserved lexical priming effects, which were greatly facilitated by the presence of a congruent sentential context, in a
group of Alzheimer’s disease patients. The authors interpret
this finding as evidence of preserved sensitivity to lexical and
sentential factors, which actually show an increased additivity
in comparison with normal elderly controls. An over-reliance
on context has been suggested to be a feature of reduced comprehension skills. For example, there is evidence for increased
benefit on reading times from predictable context in the case of
less skilled readers (Perfetti and Roth, 1981).
The gender priming effect was also preserved. The evidence about priming effects induced by non-semantic information is more limited in both normal subjects and
pathological conditions. At the sentence level, syntactic priming has been assessed in term of the probabilities of re-using
particular syntactic structures in sentence production, when a
prime sentence containing the specific structure has been presented (Bock, 1986).
Grammatical priming effects have also been induced in
normal adults in inflected languages, such as Russian
(Akhutina et al., 1999) and Italian (Bates et al., 1996). In
both languages, the gender of a noun modifier served both
to facilitate (if congruent) and inhibit (if incongruent) recognition of the following noun. The same priming effects were
shown to be present in ECs, but either missing altogether (in
Italian; Bates et al., 2001) or severely attenuated (in Russian;
Akhutina et al., 2001) in both fluent and non-fluent aphasic
patients in the same languages. The observation that gender
priming is vulnerable in aphasic patients provides an interesting contrast to our present findings with Alzheimer’s
disease patients.
It has been shown recently that the presentation of a mismatching article before a contextually expected noun elicits an
event-related potential (ERP) negativity at the time of article
presentation, similar to the N400 (Gunter et al., 2000; Wicha
et al., 2003). As in the studies of young Italian subjects
(Bentrovato et al., 1999, for picture naming; Bentrovato
et al., 2003, for word reading), the findings of Wicha et al.
(2003) suggest that both grammatical and semantic information
is accessed early, and in parallel, during auditory language
comprehension (see also Wicha et al., 1997). It is noteworthy
that in off-line tasks, such as an off-line (untimed) variant of
picture naming, there is evidence that when the response is a
semantic paraphasia, in Alzheimer’s disease patients the gender
The effects of ageing and Alzheimer’s disease on priming
of the target word fails to exert the biasing effect which is
observed in normal controls (Paganelli et al., 2003).
In the present study, the facilitation role of semantic and
gender information on word naming appears to be additive,
and to be preserved in both normal ageing and Alzheimer’s
disease. While the lack of an interaction effect in the latter
group may be taken to indicate the presence of subtle
qualitative differences in the integration of different sources
of information, the presence of the combined priming effects
is compatible with a substantial preservation of semantic and
grammatical knowledge.
The finding of a significant interference effect induced only
by the combined grammatical and semantic violation in
Alzheimer’s disease is in agreement with the notion of a combined effect of both types of information. Hasher and Zacks
(1988) suggested that a basic feature of primary (normal)
ageing is a loss of inhibition of irrelevant information. This
process has been suggested to be defective in Alzheimer’s
disease patients, as shown by their performance in a number of
non-linguistic tasks (Balota and Duchek, 1991; Sullivan et al.,
1995; Spieler et al., 1996; Faust et al., 1997; Amieva et al.,
2002). In the case of language processing, this dysfunction may
result in a situation in which the inappropriate information
drives the response. For example, it may account for the
frequency–regularity interaction observed in reading tasks in
Alzheimer’s disease patients (Balota and Ferraro, 1993). In
terms of the dual-route reading model (Morton and Patterson,
1980), there are two reading mechanisms, an addressed (lexical) and an assembled (sublexical) pathway. These mechanisms lead to different phonological outcomes, which enter into
competition with each other. The lexical route always produces
the correct output, and the output of the assembly route must be
inhibited for the correct pronunciation of irregular words. In the
case of on-line tasks, Faust and Gernsbacher (1996) have shown
that inappropriate information from semantically ambiguous
words had a larger interference effect on sentence comprehension in Alzheimer’s disease patients than in healthy elderly
subjects. The subjects had to judge whether a word matched
the overall meaning of a sentence. In some cases, the last word of
a sentence was ambiguous, and the test word was related to one
of its meanings, which, however, was not appropriate for the
sentence (e.g. sentence, he dug with the spade; test word, ace). In
these trials, the Alzheimer’s disease patients were significantly
slowed (and made more errors) in comparison with matched
elderly subjects, suggesting an inability to suppress the irrelevant meaning. The combined violation of expectancy induced
by the G–S– condition in our study (but not the single gender
mismatch or semantic incongruence) may represent a serious
challenge to the reduced inhibitory abilities of Alzheimer’s
disease patients. It has been suggested that the impairment of
inhibitory mechanisms in Alzheimer’s disease is limited to
tasks requiring at least some degree of controlled inhibition
(Amieva et al., 2004), while completely automatic tasks,
such as inhibition of return, appear to be unaffected. These
authors propose that the automatic/controlled distinction,
rather than the nature of the task, is the crucial determinant
2305
of the effect of Alzheimer’s disease. According to this view,
the interference effect observed in the present study may be
considered to reflect a failure of controlled suppression of the
expectancy induced by the combined gender mismatch and
semantic incongruence.
In conclusion, at least some of the modifications in language processing manifested in Alzheimer’s disease may be
the result of dysfunction in non-linguistic domains, such as
attention and executive control, rather than the consequence
of loss of linguistic knowledge. This issue may be relevant for
both behavioural and pharmacological interventions in early
Alzheimer’s disease, and deserves further investigations,
extending to other dementing conditions associated with
prominent language dysfunction, such as frontotemporal
dementia.
Acknowledgements
We wish to thank Dr Patrizio Pasqualetti for his help with
statistical analysis. Partial support was provided by NIDCD
R01 DC00216, ‘Cross-linguistic studies of aphasia’, directed
by Elizabeth Bates.
References
Abbate MS, LaChapelle NB. Pictures, please! Tucson (AZ): Communication
Skill Builders; 1984.
Akhutina T, Kurgansky A, Polinsky M, Bates E. Processing of grammatical
gender in a three-gender system: experimental evidence from Russian.
J Psycholinguist Res 1999; 28: 695–713.
Akhutina T, Kurgansk A, Kurganskaya M, Polinsky M, Polonskaya N,
Larina O, et al. Processing of grammatical gender in normal and aphasic
speakers of Russian. Cortex 2001; 37: 295–326.
Amieva H, Lafont S, Auriacombe S, Le Carret N, Dartigues JF, Orgogozo JM,
et al. Inhibitory breakdown and dementia of the Alzheimer type: a general
phenomenon? J Clin Exp Neuropsychol 2002; 24: 503–16.
Amieva H, Phillips LH, Della Sala S, Henry JD. Inhibitory functioning in
Alzheimer’s disease. Brain 2004; 127: 949–64.
Balota DA, Duchek JM. Semantic priming effects, lexical repetition effects,
and contextual disambiguation effects in healthy aged individuals and
individuals with senile dementia of the Alzheimer type. Brain Lang
1991; 40: 181–201.
Balota DA, Faust M. Attention in dementia of the Alzheimer’s type. In:
Boller F, Cappa S, editors. Handbook of neuropsychology, Vol. 6.
Aging and dementia. Amsterdam: Elsevier; 2001. p. 51–80.
Balota DA, Ferraro FR. A dissociation of frequency and regularity effects in
pronunciation performance across young adults, older adults, and individuals with senile dementia of the Alzheimer’s type. J Mem Lang 1993; 32:
573–92.
Bates E, Devescovi A, Hernandez A, Pizzamiglio L. Gender priming in
Italian. Percept Psychophys 1996; 58: 992–1004.
Bates E, Federmeier K, Herron D, Iyer G, Jacobsen T, Pechmann T, et al.
Introducing the CRL International Picture-Naming Project (CRL-IPNP).
Center for Research in Language Newsletter, Vol. 12, No. 1, 2000. San
Diego: University of California; 2000.
Bates E, Marangolo P, Pizzamiglio L, Dick F. Linguistic and nonlinguistic
priming in aphasia. Brain Lang 2001; 76: 62–9.
Bentrovato S, Devescovi A, D’Amico S, Bates E. Effect of grammatical
gender and semantic context on lexical access in Italian. J Psycholinguist
Res 1999; 28: 677–93.
Bentrovato S, Devescovi A, D’Amico S, Wicha N, Bates E. The effect of
grammatical gender and semantic context on lexical access in Italian
2306
R. Manenti et al.
using a timed word-naming paradigm. J Psycholinguist Res 2003; 32: 417–
30.
Bock JK. Syntactic persistence in language production. Cogn Psychol 1986;
18: 355–87.
Cameli L, Phillips NA. Age-related differences in semantic priming: evidence
from event related brain potentials. Brain Cogn 2000; 43: 69–73.
Chertkow H, Bub D, Seidenberg M. Priming and semantic memory loss in
Alzheimer’s disease. Brain Lang 1989; 36: 420–46.
Cohen J, MacWhinney B, Flatt M, Provost J. Psyscope: an interactive graphic
system for designing and controlling experiments in the psychology laboratory using Macintosh computers. Behav Res Methods Instrument Comput
1993; 25: 257–271.
Dunn LM, Dunn LM. Peabody Picture Vocabulary Test-Revised. Circle Pines
(MN): American Guidance Service; 1981.
Faust ME, Gernsbacher MA. Cerebral mechanisms for suppression of inappropriate information during sentence comprehension. Brain Lang 1996;
53: 234–59.
Faust ME, Balota DA, Duchek JM, Gernsbacher MA, Smith S. Inhibitory
control during sentence comprehension in individuals with dementia of the
Alzheimer type. Brain Lang 1997; 57: 225–53.
Fischler I, Bloom PA. Effects of constraints and validity of sentence contexts
on lexical decisions. Mem Cognit 1985; 13: 128–39.
Giffard B, Desgranges B, Nore-Mary F, Lalevee C, Beaunieux H, de la
Sayette V, et al. The dynamic time course of semantic memory impairment
in Alzheimer’s disease: clues from hyperpriming and hypopriming effects.
Brain 2002; 125: 2044–57.
Good CD, Scahill RI, Fox NC, Ashburner J, Friston KJ, Chan D, et al.
Automatic differentiation of anatomical patterns in the human brain:
validation with studies of degenerative dementias. Neuroimage 2002;
17: 29–46.
Gunter TC, Friederici AD, Schriefers H. Syntactic gender and semantic
expectancy: ERPs reveal early autonomy and late interaction. J Cogn
Neurosci 2000; 12: 556–68.
Hasher L, Zacks RT. Working memory, comprehension and aging. A review
and a new view. In: Bower G, editor. The psychology of learning and
motivation. Advances in research and theory. San Diego: Academic
Press; 1988. p. 193–225.
Hodges JR, Salmon DP, Butters N. Semantic memory impairment in Alzheimer’s disease: failure of access or degraded knowledge? Neuropsychologia
1992; 30: 301–14.
Kaplan E, Goodglass H, Weintraub S. Boston Naming Test. 2nd edn.
Philadelphia: Lea & Febiger; 1983.
MacKay DG, Abram L. Language, memory and aging: distributed deficits and
the structure of new-versus-old connections. In: Birren JE, Schaie KW,
editors. Handbook of psychology of aging. San Diego: Academic Press;
1997: p. 251–65.
McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM.
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 1984; 34: 939–44.
Miyamoto T, Katayama J, Koyama T. ERPs, semantic processing and age. Int
J Psychophysiol 1998; 29: 43–51.
Morton J, Patterson KE. A new attempt at an interpretation, or an attempt at a
new interpretation. In: Coltheart M, Patterson KE, Marshall JC, Kegan P,
editors. Deep dyslexia. London: Routledge; 1980.
Myerson J, Ferraro FR, Hale S, Lima SD. General slowing in semantic
priming and word recognition. Psychol Aging 1992; 7: 257–70.
Myerson J, Hale S, Chen J, Lawrence B. General lexical slowing and the
semantic priming effect: the roles of age and ability. Acta Psychol (Amst)
1997; 96: 83–101.
Nebes RD, Brady CD. The effect of contextual constraint on semantic judgments by Alzheimer patients. Cortex 1991; 27: 237–46.
Nebes RD, Martin DC, Horn LC. Sparing of semantic memory in Alzheimer’s
disease. J Abnorm Psychol 1984; 93: 321–30.
Ober BA. RT and non-RT methodology for semantic priming research with
Alzheimer’s disease patients: a critical review. J Clin Exp Neuropsychol
2002; 24: 883–911.
Ober BA, Shenaut GK. Lexical decision and priming in Alzheimer’s disease.
Neuropsychologia 1988; 26: 273–86.
Ober BA, Shenaut GK. Well-organized conceptual domains in Alzheimer’s
disease. J Int Neuropsychol Soc 1999; 5: 676–84.
Paganelli F, Vigliocco G, Vinson D, Siri S, Cappa S. An investigation of
semantic errors in unimpaired and Alzheimer’s speakers of Italian. Cortex
2003; 39: 419–39.
Perani D, Cappa SF. Brain imaging in normal aging and dementia. In:
Boller F, Cappa S, editors. Handbook of neuropsychology, Vol. 6.
Aging and dementia. Amsterdam: Elsevier; 2001. p. 429–52.
Perfetti CA, Roth S. Some of the interactive processes in reading and their role
in reading skill. In: Lesgold AM, Perfetti CA, editors. Interactive processes
in reading. Hillsdale (NJ): Lawrence Erlbaum; 1981. p. 269–97.
Ramage AE, Holland A. Language in normal aging and age-related neurological disorders. In: Boller F, Cappa SF, editors. Handbook of neuropsychology, Vol. 6. Aging and dementia. Amsterdam: Elsevier; 2001.
p. 81–95.
Roe K, Jahn-Samilo J, Juarez L, Mickel N, Royer I, Bates E. Contextual
effects on word production: a lifespan study. Mem Cognit 2000; 28:
756–65.
Scahill RI, Frost C, Jenkins R, Whitwell JL, Rossor MN, Fox NC.
A longitudinal study of brain volume changes in normal aging using
serial registered magnetic resonance imaging. Arch Neurol 2003; 60:
989–94.
Schwanenflugel PJ, LaCount KL. Semantic relatedeness and the scope of
facilitation for upcoming words in sentences. J Exp Psychol Learn Mem
Cognit 1988; 14: 344–54.
Schwartz TJ, Federmeier KD, Van Petten C, Salmon DP, Kutas M. Electrophysiological analysis of context effects in Alzheimer’s disease. Neuropsychology 2003; 17: 187–201.
Snodgrass JG, Vanderwart M. A standardized set of 260 pictures: norms for
name agreement, image agreement, familiarity, and visual complexity.
J Exp Psychol [Hum Learn] 1980; 6: 174–215.
Spieler DH, Balota DA, Faust ME. Stroop performance in healthy younger
and older adults and in individuals with dementia of the Alzheimer’s type.
J Exp Psychol Hum Percept Perform 1996; 22: 461–79.
Sullivan MP, Faust ME, Balota DA. Identity negative priming in older adults
and individuals with dementia of the Alzheimer’s type. Neuropsychology
1995; 9: 537–55.
Thompson PM, Hayashi KM, de Zubicaray G, Janke AL, Rose SE, Semple J,
et al. Dynamics of gray matter loss in Alzheimer’s disease. J Neurosci
2003; 23: 994–1005.
Ullman MT. A neurocognitive perspective on language: the declarative/
procedural model. Nat Rev Neurosci 2001; 2: 717–26.
West RF, Stanovich KE. Automatic contextual facilitation in readers of three
ages. Child Dev 1978; 40: 717–27.
Wicha NY, Bates A, Orozco-Figueroa A, Reyes I, Hernandez A, Gavaldòn de
Barreto LG. When zebras become painted donkeys: the interplay between
gender and semantic priming in a Spanish sentence context. Center for
Research in Language. San Diego: University of California; 1997.
Wicha NY, Bates EA, Moreno EM, Kutas M. Potato not Pope: human brain
potentials to gender expectation and agreement in Spanish spoken sentences. Neurosci Lett 2003; 346: 165–8.