PreprintPDF Available
Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

Background. Previous studies have argued that two types of linguistic gender exist: grammatical gender, which is arbitrarily assigned to nouns, and semantic gender, which depends on the gender of the referent. Aim. We explore the hypothesis that these two types of gender entail distinct cognitive processes by investigating the performance of people with aphasia at the level of sentence comprehension. Methods and Procedure. Eleven people with aphasia and a control group of 13 age-matched healthy participants took part in a constrained completion choice task. The participants had to complete sentences in a way that made the last word gender congruent. The subjects of the sentences had either Semantic gender (“enfermera”, nurse; indicating the sex of the referent), Grammatical gender (“silla”, chair), or Opaque-Grammatical gender (“tomate”, tomato). Results. People with aphasia performed more poorly in all gender conditions than healthy controls. They also were less accurate in both the Grammatical and Opaque-Grammatical conditions than in the Semantic gender condition. Conclusion. We propose that semantic and grammatical gender entail two levels of gender processing and that semantic gender is processed faster because it provides more salient information.
Content may be subject to copyright.
Processing of semantic and grammatical gender in Spanish speakers with
G. Piazzaa,b*, M. Calabriac, C. Semenzad, C. Polettob,e
aBasque Center on Cognition, Brain and Language (BCBL), Donostia-San Sebastián, Spain;
b Department of Linguistic and Literary Studies (DiSLL), University of Padova, Padova,
c Faculty of Health Sciences, Universitat Oberta de Catalunya (UOC), Barcelona, Spain;
d Padova Neuroscience Center (PNC), University of Padova, Padova, Italy;
e Institute of Romance Languages and Literature, Goethe University Frankfurt, Frankfurt am
Main, Germany.
*corresponding author:
Author Note
This is a preprint of the paper published in Aphasiology. Please, find the published version at
Processing of semantic and grammatical gender in Spanish speakers with
Background. Previous studies have argued that two types of linguistic gender exist:
grammatical gender, which is arbitrarily assigned to nouns, and semantic gender, which
depends on the gender of the referent.
Aim. We explore the hypothesis that these two types of gender entail distinct cognitive
processes by investigating the performance of people with aphasia at the level of sentence
Methods and Procedure. Eleven people with aphasia and a control group of 13 age-
matched healthy participants took part in a constrained completion choice task. The
participants had to complete sentences in a way that made the last word gender congruent. The
subjects of the sentences had either Semantic gender (“enfermera”, nurse; indicating the sex of
the referent), Grammatical gender (“silla”, chair), or Opaque-Grammatical gender (“tomate”,
Results. People with aphasia performed more poorly in all gender conditions than healthy
controls. They also were less accurate in both the Grammatical and Opaque-Grammatical
conditions than in the Semantic gender condition.
Conclusion. We propose that semantic and grammatical gender entail two levels of gender
processing and that semantic gender is processed faster because it provides more salient
Keywords: aphasia, agreement, semantic gender, grammatical gender, sentence
1. Introduction
Gender processing is a critical part of both lexical and syntactic processing in many
languages. Transparent languages convey explicit grammatical gender information in the final
morpheme of nouns or through a noun affix. For instance, Spanish is considered
(morphologically) transparent because the (transparent) final morpheme of nouns provides
gender information: -a for words with feminine grammatical gender (e.g. “casa”, house.FEM),
-o for masculine words (e.g. “casco”, helmet.MASC). However, in some cases, the final
morpheme -e (opaque) does not permit gender identification because it is used for both
feminine and masculine nouns. Furthermore, there are two different types of gender:
grammatical gender is arbitrarily assigned to nouns, while semantic gender is dictated by the
gender of the referent1. In Spanish, both types of gender follow the same morphemic rules and
use the same features. For instance, changing the ending (-o/-a) creates a noun of the opposite
gender with some, but not all, animal names: mono (male monkey) or mona (female
monkey). Semantic gender is presumed to be strictly bound to the semantic meaning of nouns
and for this reason is hypothesized to involve a different retrieval path than grammatical gender
(Vigliocco and Franck, 1999).
In the present study, we investigate the hypothesis that processing grammatical gender and
semantic gender entail two levels of comprehension processes. We explore the assumption that
semantic gender carries salient semantic information making it easier to process agreement
and access meaning by comparing the performance of people with aphasia (henceforth PWA)
with healthy controls. In fact, studies on various populations of patients have highlighted that
gender can be preserved, lost, or difficult to retrieve in word recognition and sentence
comprehension (e.g., Manenti et al., 2009), suggesting possible dissociations between these
linguistic processes.
1.1. Semantic vs grammatical gender
The difference between semantic and grammatical gender is not merely a difference
between animate and inanimate nouns. Semantic gender is supplementary semantic
information about the gender of the noun referent that is reflected in the noun’s superficial
morphology. For instance, it is distinct from grammatical gender in words like “silla”
(chair.FEM), and “pájaro” (bird.MASC), where the morphology does not convey any
information about the sex of the referents; in fact, a chair is an object, while the word “pájaro”
does not have a feminine counterpart and is a general term that refers to all birds, both male
and female. It is worth underscoring that the gender or sex of an animate referent does not in
itself constitute semantic gender if it does not coincide with the morphological expression of
gender. For this reason, “pájaro”, which does not have a feminine counterpart, and “tigre”
(tiger.MASC), which ends with a morpheme that could be either masculine or feminine, do not
have semantic gender.
Comprehension and information retrieval mechanisms operate differently when nouns are
encountered in the context of sentences than they do in bare noun recognition. Accordingly,
while agreement is optional in word pairs, it is always active and involves a constant process
for selection of correct agreement candidates in sentences (see Roelofs et al., 1998; Cacciari
and Cubelli, 2003; Corbett, 2006). For this reason, our experimental design used an enriched
sentence comprehension context, which ensured gender was activated as participants accessed
word meaning and computed agreement at a distance.
There are two main accounts that aim to explain how semantic gender influences sentence
comprehension. The first account is known as the minimal input hypothesis and derives from
theories that claim semantic features have minimal impact on syntactic processing (Bock and
Levelt, 1994; Kempen and Hoenkamp, 1987; Levelt, 1989). The minimal input hypothesis
claims that grammatical gender is a purely syntactic feature, which cannot be influenced by
conceptual agreement
(Chomsky, 1965, 1981); “the committee have decided" is an example
of conceptual agreement. In this sentence, the subject-verb agreement is not correct from a
grammatical point of view, but it is conceptually acceptable because the meaning is plural.
When possible, syntactic agreement is preferred over conceptual agreement; children, for
example, autonomously create an agreement system between constituents before they learn
rules and irregularities. In this first account, there cannot be different levels (grammatical and
semantic) for processing gender, because gender is grammatical and therefore is only processed
at the syntactic level. The second account is called the maximal input hypothesis and claims
that semantic meaning and structures have some influence over agreement and syntactic
operations (Barlow, 1993; MacDonald et al., 1994; Pollard and Sag, 1994; Vigliocco, et al.,
1996; Vigliocco and Franck, 1999).
Since many linguistic phenomena demonstrate a dichotomy between semantic and
grammatical gender, Vigliocco and Franck (1999) postulate that they are, to some extent,
processed differently. According to this view, conceptual structures that carry semantic
information influence grammatical functions and syntactic processes (such as agreement).
Additional support for this hypothesis comes from work on argument hierarchy and feature-
sharing, which shows that agreement is driven by semantic as well as syntactic relations. This
has led to the assumption that gender has a double nature (Vigliocco and Franck, 1999), which
allows us to process semantic gender differently from grammatical gender (cf. Van Dyke, 2007;
for a different view see Bates et al., 1995). Linguistic elements create dependencies, even at
long distances, which comprehension processes resolve via cue-based retrieval mechanisms
(Lewis et al, 2006; Van Dyke and McElree, 2006). If a competing item shares some features
with the target, that can interfere with the retrieval process and induce errors (Parker et al.,
2017). Conversely, semantic cues would also explain why semantic gender nouns were
produced more accurately than grammatical gender nouns in Vigliocco and Franck’s
experiment (1999).
In the study by Vigliocco and Franck (1999), participants were first presented with
adjectives in both their masculine and feminine forms, then a preamble sentence. The gender
type of the subject (head noun of the sentence) had either semantic or grammatical gender.
The participants had to repeat the whole sentence, adding on the correct (gender-matched)
adjective at the end. The researchers found that agreement errors were more frequent when
the subjects had grammatical gender. They argued that their manipulation led participants to
distinguish between gender that carries both syntactic information and semantic properties
derived from the gender of the referent (semantic gender) from purely grammatical gender
with bare syntactic features (grammatical gender). Although we still lack evidence for this
division of labour in sentence comprehension processing, such results strongly suggest that
at least in speech production some kind of internal gender distinction is made, and that
different processes underpin the assignment of semantic and grammatical gender (see also
Vigliocco et al, 1995; for similar results in comprehension, see Franzon et al., 2013).
Vigliocco and Franck’s (1999) experiment also explored whether semantic gender is
actually assigned on the basis of animacy, given that all nouns endowed with semantic gender
are also animate nouns. They compared animate nouns with transparent gender morphology,
in which gender marking was the same for both sexes (as in the “pájaro” example mentioned
above) and inanimate nouns. This allowed them to manipulate animacy separately, isolating
potentially confounding factors. In this experiment, they did not find any difference between
animate nouns with fixed gender a single final morpheme (such as “pájaro”) and inanimate
nouns, indicating that semantic gender is not assigned on the base of animacy. Moreover,
Vigliocco and Franck (1999) investigated whether animate nouns with fixed grammatical
gender bear semantic gender but did not find an effect comparable to that seen for proper
semantic gender nouns. These results indicate that semantic gender is processed differently
than fixed gender.
This dissociation between semantic and grammatical gender has been demonstrated by a
number of studies using different types of paradigms and techniques. Schiller et al. (2003)
found that the frontal ERP N200 component was larger for nouns marked with semantic gender
than for animate nouns with opaque endings. Similarly, Franzon et al.’s (2013) preliminary
study on two people with aphasia confirmed Vigliocco and Franck’s (1999) results, finding
they had higher accuracy for semantic gender in noun-adjective pairs. In a more recent study,
Caffarra et al. (2016) reported an effect of semantic gender on bare words in Spanish-Basque
bilinguals. In this experiment, nouns were presented on a screen and the participants had to
decide which gender they belonged to. The authors manipulated the nouns in such a way that
half of the stimuli had opaque endings (i.e., the final -e in Spanish), which did not provide
information on the gender of the nouns, while the other half had transparent endings (i.e. final
-a/-o) expressing feminine/masculine gender. Within the transparent category, half of the nouns
had semantic gender and half had only grammatical gender. They observed facilitation in terms
of RTs and accuracy for nouns bearing semantic gender. This finding suggests that semantic
gender helps with the categorisation of bare nouns by means of explicit gender knowledge.
However, to date, no study has explored how people with aphasia (PWA) compute semantic
gender in comprehension and whether it facilitates agreement processing when no declarative
knowledge is required (i.e., when deciding whether a noun is masculine or feminine vs using
gender to match a noun and its adjective).
In sum, research has not provided conclusive evidence for the existence of two types of
gender processes nor found evidence for it in comprehension tasks. More importantly, it is not
clear whether semantic gender facilitates processing, or simply differs from grammatical
gender. The present study intends to contribute to clarifying these issues using a cognitive
neuropsychology method.
1.2. Aphasia and Gender Processing
Research has found evidence for and against the maintenance of gender knowledge in
PWA. Some research has demonstrated that gender knowledge is maintained in people with
aphasia. For instance, although Bates et al. (2001) tested both Broca’s and Wernicke PWA
and failed to observe gender priming effects, they found that PWA were able to correctly use
gender. In contrast, Akhutina et al. (2001) found that PWA, with various types of aphasia,
had comparable difficulties with recognizing grammatical gender during a grammaticality
judgement task. The apparent discrepancy between these results can be explained if we
assume that two types of gender knowledge exist: one procedural and the other declarative or
metalinguistic (Scarnà and Ellis, 2002). The former is unconscious syntactic gender
knowledge that allows for automatic computation of agreement between constituents without
the need for explicit recall of gender. The latter is the metalinguistic ability to explicitly
indicate the gender of an element without putting any syntactic operations into play
(Goodglass, 2000, see also Miceli et al, 2002). Scarnà and Ellis (2002) made this assumption
after they found that bilingual PWA had problems categorizing Italian nouns with explicit
grammatical gender, but were nevertheless able to modify adjectives to ensure agreement
with a Noun Phrase (NP). It appears that, in PWA, grammatical gender remains intact when
retrieved procedurally and unconsciously employed for agreement but is impaired when
consciously recalled.
Despite the evidence reported above, some research on PWA has shown varying
results for procedural knowledge. For instance, Mondini et al. (1999) tested an Italian with
aphasia on article-noun production and described serious deficits in producing gender
agreement even for nouns bearing gender information, (e.g., “il madre”, the.MASC mother).
It is worth noting that, because of the opaque final -e, cases like “madre” do not allow for
direct recall of gender information (“padre”, father ends with the same final -e), so it may not
be possible to observe the semantic gender facilitation, as Vigliocco and Franck (1999) also
pointed out. Friedmann and Biran (2003) studied picture naming, which does not require
declarative gender knowledge since gender is only used to produce agreement (by assigning
the correct gender morpheme), in Hebrew-speaking PWA. Their participants did not preserve
grammatical gender. The authors suggest that this result could reflect the special features of
Hebrew which allows bare nouns these nouns do not need a determiner to be linguistically
legal. Consistent with what we previously reported in Cacciari and Cubelli (2003),
grammatical gender might only be accessed if agreement is activated (see Roefols et al.,
1998). Again, for this reason we decided to use subject-adjective pairs in sentence
comprehension, to make sure that agreement comes into play. Moreover, congruent with
evidence on explicit and procedural gender knowledge distinction (Scarnà and Ellis, 2002),
we expected PWA to be able to unconsciously employ gender information in our experiment,
where adjective and noun gender is procedurally matched, without any need for
In sum, evidence from PWA is neither clear-cut nor definitive. In addition, to the best
of our knowledge, semantic gender comprehension has never been systematically investigated
in aphasic populations (see Semenza and Luzzatti, 2019). It is not known how semantic gender
is processed by PWA, that is, whether impairments in gender-explicit knowledge may
influence their performance in comprehension, or whether as we assume additional
semantic information on the sex of the referent will help them complete the task, as we expect
to find in control participants. As demonstrated by prior research, cognitive processes in
impaired and healthy populations diverge substantially; by pinpointing specific deficits, we can
learn more about such processes.
2. The Present Study
This study aimed to verify whether gender is processed differently during sentence
comprehension when its source (e.g., sex) matches morphology, and whether explicit
knowledge impairments differently influence the performance of PWA when restoring
grammatical and semantic gender. In a constrained completion choice task, participants read a
sentence and then saw in isolation the masculine and feminine forms of an adjective; they
had to decide which form matched the subject of the previous sentence. We employed three
conditions: Grammatical gender, Semantic gender (which shares the morphological
transparency of the first condition), and Opaque-Grammatical gender, a condition inserted to
prevent participants from basing their strategy solely on morphology and to check whether any
of the observed effects could be driven by morphology.
According to the maximal input hypothesis (Vigliocco & Franck, 1999), which
proposes that semantic processing exerts an influence on grammatical processing, we expected
the following results:
Healthy controls would respond faster when checking agreement between adjectives
and subjects bearing semantic compared to grammatical gender. As the task is easy for healthy
individuals, we did not expect gender conditions to influence their accuracy. In fact, their
performance might show reduced differences between conditions and hide the effect derived
from processing semantic gender.
We expected patients to be less accurate and slower in all conditions than healthy
controls. We were interested to understand if this impairment would vary with gender type; we
expected PWA to be faster and more accurate in the Semantic condition. Specifically, if
patients performed better in the Semantic than the other conditions, this would confirm the
maximal input hypothesis. Furthermore, unlike healthy controls, they might have issues in
computing agreement for grammatical and opaque-grammatical gender, which would widen
the gap between these two grammatical conditions and the Semantic condition. Alternatively,
if patients perform similarly in all gender conditions, we would have to conclude that gender
types are not dissociated, the maximal input hypothesis could not explain our results, and the
two gender types do not follow two distinct processing paths.
3. Method
3.1. Participants
A total of 11 Spanish-speaking people (7 Spanish-Catalan bilinguals) with aphasia were
recruited from the Speech Therapy Unit of the San Pau i la Santa Creu Hospital in Barcelona
(mean age 50 years ± 9.63, mean education 13.7 years ±3.6, aphasia severity range from mild
to moderate). Ten PWA had suffered a stroke and one had had cerebral cancer. All the PWA had
a brain lesion in the left hemisphere, exhibited adequate hearing, demonstrated stable health
status, and were in the chronic stage of language disorders (more than one year after the injury).
From the total number of PWA, two were discarded from the analysis due to their high error
rate (around 60%) leaving a final cohort of 9 PWA.
A control group, comprising 13 participants matched for age and education with the PWA
group, also completed both tasks. Eleven of these participants were native Catalan-Spanish
speakers, while two were Spanish speakers with some knowledge of Catalan (passive
bilinguals; mean age 50 ± 4.6, mean education 14.1 years, ± 2.3).
Before starting the experimental procedure, both patients and controls signed an informed
consent form approved by the ‘Parc de Salut MAR’ Research Ethics Committee.
Language assessment. To define the type and degree of language impairment in PWA
participants, a Spanish version of the Western Aphasia Battery test (WAB, Kertesz, 1982;
Kertesz & Pascual-Leone, 2000) was administered by a neuropsychologist with expertise in
aphasia. The WAB allows for the evaluation of the main clinical aspects of language: function,
content, fluency, auditory (sentence) comprehension, repetition, naming, reading, writing, and
According to the WAB assessment, and according to the subtest scores (see Table 1 for
individual subtest scores), only one PWA exhibited NonFluent aphasia along with scores
compatible with Broca’s aphasia (Pt. 5); the rest exhibited Fluent aphasia. The subtest scores
of three PWA were compatible with Anomic aphasia, two with Wernicke’s aphasia, and two
with Conduction aphasia. One PWA was not classified (Pt.9).
The degree of language impairment ranged from mild to moderate (WAB AQs of 54.1 to
89.2 out of 100) and the mean values for each subtest were: 13.12/20 (±4.02) for Fluency,
8.23/10 (±1.40) for Comprehension, 6.83/10 (±1.9) for Repetition, and 7.30/10 (±1.43) for
See the published version for more details about participants’ language profile.
3.2. Material
A set of 78 sentences, comprising 26 sentences for each of the three experimental
conditions (Gender Type), was selected (see Appendix A for examples). Sentence subjects (or
controllers, see Corbett, 1991, 2006) across the three gender conditions differed as follows:
a) Subjects with Semantic gender were nouns that had semantic gender reflecting the sex
of the referent (nouns referring to humans, animals, or occupations) and transparent endings (-
a/-o; e.g., “esposo”, spouse.MASC).
b) Subjects with Opaque-Grammatical (henceforth, Opaque) gender were inanimate
nouns with grammatical gender and opaque endings (-e ; e.g., “puente”, bridge.MASC).
c) Subjects with Transparent-Grammatical (henceforth, Grammatical) gender were
inanimate nouns with grammatical gender and transparent endings (-a/-o; e.g., “cuchara”,
Adjectives available for sentence completion were all phonologically transparent with -a
or -o endings that clearly indicated feminine or masculine gender. The sentences (e.g., “El
esposo en la iglesia está”, the spouse in the church is...) were formed with a subject that was a
full noun-phrase (NP), including an article and a noun, followed by a distractor that always had
the opposite gender as the subject (“en la iglesia”, in the church) and the third-person singular
of the verbs ser/estar (to be; hence, es or está). The final part of the sentence was the adjective,
which needed to agree with the subject. All adjectives had transparent endings (-a/-o) with
relevant features balanced across conditions (see below); all options provided plausible
sentence completions (e.g., “emocionado”, excited). We introduced distractors with the
opposite gender from subjects to induce a higher error rate, since research has shown that target
retrieval is slower when other items share some of its features (Van Dyke and McElree, 2006;
see also Parker et al., 2017).
We checked the values of the psycholinguistic variables of interest using Busca-Palabras
(B-Pal) software (Davis & Perea, 2006); B-Pal is a free software program that brings together
previous databases on various psycholinguistic properties (also subjective ratings) of
approximately 31,500 Spanish words. The final number results from the authors’ selection of
words from the various databases and the Real Academia Española (RAE) dictionary. We used
B-Pal scores to balance the subjects of the sentences for average log frequency means, which
did not significantly vary across conditions: 1.16 (SD = 0.53) for the Grammatical gender
condition, 1.15 (SD = 0.6) for the Semantic gender condition, and 1.11 (SD = 0.66) for the
Opaque gender condition. We did the same for the adjectives and distractors. Adjective average
log frequency did not significantly differ across conditions: 1.25 (SD = 0.57) for the
grammatical, 1.14 (SD = 0.5) for the semantic, and 1.15 (SD = 0.61) for the opaque gender
condition. The average frequency of distractors was, respectively, 1.37 (SD = 0.56), 1.6 (SD =
0.66), and 1.59 (SD = 0.52) for the three conditions.
Imageability was also balanced and did not significantly differ across conditions for either
adjectives or distractors. The average imageability score for adjectives was 4.01 (SD = 0.67)
for the Grammatical condition, 4.15 (SD = 0.51) for the Semantic condition, and 4.27 (SD =
0.62) for the Opaque condition. The average imageability score of the distractors was 5.59 (SD
= 0.91), 5.55 (SD = 0.74), and 5.83 (SD = 0.51), respectively.
Since objects do not have semantic gender, it was by definition impossible to balance the
animacy scores of subjects across conditions. To calculate the statistics, we simply assigned -
0,5 to inanimate and 0,5 to animate nouns. Subjects with Semantic gender had a significantly
higher animacy score than the other conditions (p < 0.05), while Opaque and Grammatical
gender did not differ. This is partially interesting in itself since, as already mentioned, Vigliocco
and Franck (1999) tested whether animacy carries the full effect of semantic gender. They
found instead that the Semantic gender effect distinguished between conditions with the same
level of animacy. Furthermore, they found no effect of animacy per se, indicating the results
from the gender type manipulation could not be attributed to animacy (see Vigliocco & Franck,
1999, for further discussion).
We created 4 different lists to avoid order effects or effects due to random noun-adjective
associations. The sentence-adjective pairs were not randomized in order to avoid nonsense
sentence-adjective pairs but were varied across the 4 lists. We picked one list of stimuli for
every participant in consecutive order. The order of stimulus presentation was randomized.
3.3. Procedure
The experimental software used for the administration of all tasks was DMDX (Forster &
Forster, 2003). The experiment was displayed on a 15-inch Toshiba screen with the participant
seated in front of the screen at a distance of approximately 40cm. A fixation cross was presented
at the centre of the screen for 700ms, then a sentence appeared for 3500ms. The sentences, for
example, “El esposo en la iglesia está” (The spouse in the church is) appeared in the mid-upper
area of the monitor. In all stimuli, the sentential subject was followed by a filler phrase (e.g.,
“en la iglesia”) and the verb es or está.
. The filler phrase (also called the distractor) was
included to increase demands on cognitive control. After the initial sentence presentation
(3500ms), two options for the final adjective were presented below (in the mid-lower area of
the screen) to the left and right, for example, the masculine (“emocionado ”, excited) on one
side and the feminine (“emocionada”, excited) on the other. Right and left positions were
In Spanish the verb to be can be expressed using two forms, deriving from different roots, depending on the
balanced across participants and trials. The sentence remained on the screen for up to 10
seconds or until a response was provided. The sentences were black (RGB 0,0,0) in size 16
(Font Corbel) presented in the centre of the screen with a grey background (RGB 192,192,192).
The adjectives were written in all caps and presented in the centre mid-lower area of the screen.
Using a standard keyboard (see Damian, 2010, for a discussion on its experimental use),
participants had to choose between the two options in order to complete the sentence according
to the gender of the subject: pressing ‘Z’ for the left-sided and ‘M’ for the right-sided adjective.
We substituted keys ‘Z’ and ‘C’ for PWA who could only use their left hand. We
counterbalanced the side of the correct answer for all participants in order to reduce any bias
on RTs due to hemiplegia. Moreover, participants were asked to initiate responses starting from
a neutral position on the keyboard.
3.4. Statistical analysis
Repeated-measure ANOVAs were performed for the analysis of the experimental data,
including within-subject and Group factors (PWA vs. control group). The analyses were
conducted separately on RTs and accuracy. All missed and overtime responses (exceeding
timeout), software malfunctions, and inaccurate responses were classified as errors and
discarded from the analysis. In addition, we removed all responses exceeding 2 standard
deviations above or below the individual RT mean.
Given that Broca’s aphasia (NonFluent aphasia) can entail issues with grammar, and
therefore also issues with processing grammatical gender, we conducted an analysis on the
PWA excluding the participant with Broca’s aphasia. In addition, we took into consideration
the behaviour of Pt.9 who was not classified, and we checked whether their behavioural pattern
was in line with that of participants with Fluent aphasia. Since the pattern of results and the
analysis without the participant with Broca’s aphasia did not change the overall results, here
we report only the main analysis conducted. However, individual data from the PWA were
analysed and are presented in Figures 2 and 4.
4. Results
Given the relatively small sample size, all data were also analysed using non-parametric
analyses. As the pattern of results did not differ, here we report only results from the parametric
statistics. When significant effects were found in the main analysis, we employed Bonferroni
correction for multiple comparisons in post-hoc analyses.
Repeated-measure ANOVAs were performed, including Gender Type (Grammatical vs.
Opaque vs. Semantic) as within-subject factors and Group as the between-subjects factor.
Reaction times (RTs). The main effect of Gender Type was significant [F (2, 40) = 6.256,
p = 0.004, ηp² = 0.238]. The post-hoc analysis showed that participants were faster in the
Semantic (2006ms) than the Grammatical (2063ms, p = 0.006) and Opaque (2043ms, p = 0.01)
conditions (Fig. 2). The interaction between Gender Type and Group did not reach significance
[F (2, 40) = 2.590, p = 0.08], while the Group effect did [F (1, 20) = 16.546, p = 0.001, ηp² =
0.453], revealing that PWA overall were slower (2943ms) than the control group (1084ms).
In order to control for bilingualism as a possible confounding factor, we conducted a
further analysis including only bilingual participants from both groups. This revealed that the
Gender Type effect was still significant [F (2, 34) = 5.090, p = 0.01, ηp² = 0.230], as was a
Gender Type*Group interaction [F (2, 34) = 3.287, p = 0.050, ηp² = 0.162], possibly suggesting
that the two groups carried out the task differently. To explain this interaction, we performed
further analyses by comparing the magnitude of the effects of the conditions in the two groups
of participants. The magnitude of the effects was calculated as follows: for the Semantic
condition, we calculated the difference in RTs between the Semantic and the Grammatical
conditions, divided by the RTs of the Semantic condition and multiplied by 100 (as in Calabria
et al., 2012). For the Opaque condition, we calculated the difference in RTs between the Opaque
and the Grammatical conditions, divided by the RTs of the Grammatical condition and
multiplied by 100. Despite the interaction, the magnitude of the Semantic gender effect
(Semantic-Grammatical) did not differ between the two groups [F (1, 17) = 0.009, p = 0.92].
The interaction effect was more likely driven by the Opaque condition. In the control group,
the Opaque condition resulted in slower RTs (1076ms) than the Grammatical (1044ms) and the
Semantic (989ms) conditions, whereas in the PWA group the Grammatical condition was the
slowest (3233ms), followed by the Opaque (3039ms) and Semantic (2986ms) conditions.
Nevertheless, the magnitude of the Opaque-Grammatical effect did not differ for the two
groups [F (1, 17) = 1.133, p = 0.3].
Individual level analysis. Given that variability of performance in the PWA group might reflect
different types of aphasia, we ran an individual level analysis to assess whether there was a
substantial difference in the magnitude of the effect in PWA compared to the control group.
We used the magnitude of the Semantic gender effect (vs. the Grammatical condition) as
explained above and applied the modified t-test described by Crawford and Howell (1998)
for independent samples. The t values were calculated as follows:
where X1 represents individual performance, X2 represents the mean of the control group,
s2 is the standard deviation of the control group, and N2 is the sample size of the control group.
The analysis showed that 2 PWA had proportionally shorter RTs in the Semantic condition (vs
the Grammatical condition) in comparison to the mean score of the control group, Pt.4 -13,7%
[t = 2.5, df = 12, p = 0.03], Pt.5 -13,3% [t = 2.45, df = 12, p = 0.03], suggesting that this group
benefitted most from the Semantic condition. In contrast, the Semantic condition slowed down
Pt.1’s responses, as shown by the proportional difference between the Semantic and
Grammatical conditions [t = 2.25, df = 12, p = 0.04].
Fig. 1. RTs broken down by Gender Type and participant group (control group and PWA). The
error bars indicate SE. The effect of Group is not reported, even though it is significant.
Fig. 2. Magnitude of the Semantic gender effect based on RT differences between the Semantic
condition and the Grammatical condition. Individual PWA scores (stars indicate where individual
performance significantly differed from the controls). More negative values indicate (proportionally)
faster responses in the Semantic condition.
Mean control
Semantic gender effect (RTs)
Individual PWA scores
Semantic-Grammatical (%)
Accuracy. The main effect of Group was significant [F (1, 20) = 33.893, p < 0.001, ηp² =
0.629] suggesting that PWA (74.92%) performed more poorly than controls (98.32%). The
main effect of Gender Type was significant [F (2, 40) = 6.950, p = 0.003, ηp² = 0.258] and the
post-hoc analysis showed that performance on Semantic (92,31%) was significantly more
accurate than Opaque (85,67%, p = 0.01), and marginally better than Grammatical (88,29%, p
= 0.07) gender, while the difference between the latter conditions was not significant (p = 0.14)
(Fig. 3). Also, the interaction between Gender Type*Group was significant [F (2, 40) = 4.272,
p = 0.02, ηp² = 0.176]. To explore this interaction, we conducted further post-hoc analyses and
found that the magnitude of the Semantic-Grammatical effect (Semantic gender effect) was
significantly greater in PWA [F (1, 20 = 4.542, p = 0.04, ηp² = 0.185], whereas the Opaque-
Grammatical effect was not significantly different between groups [F (1, 20) = 0.905, p = 0.3]
indicating that they had more difficulty processing agreement in the Opaque and Grammatical
conditions than controls, whereas the way in which they processed semantic gender was more
similar to the control group, therefore more preserved in the Semantic condition. The Analysis
on bilingual participants confirmed the above reported results: the main effect of Gender Type
was significant [F (2, 34) = 4.940, p = 0.01, ηp² = 0.225], and there was a Gender Type*Group
interaction [F (2, 34) = 4.097, p = 0.02, ηp² = 0.194]. Post-hoc analysis indicates that the
Semantic condition significantly differed from both the Grammatical (p = 0.009) and Opaque
(p = 0.04) conditions, which did not differ from each other (p = 1).
Individual level analysis. We ran an individual level analysis on the magnitude of the
Semantic gender effect (Semantic-Grammatical) found in PWA to compare each individual’s
performance to the control group mean. The analysis showed that 6 PWA had a larger
magnitude Semantic gender effect compared to the control group mean (Fig. 4): Pt.1 +21% [t
= 5.75, df = 12, p < 0.0001], Pt.2 +60% [t = 16.75, df = 12, p < 0.0001], Pt.7 + 13% [t = 3.57,
df = 12, p = 0.004], Pt.8 +21% [t = 5.75, df = 12, p < 0.0001], Pt.9 +18% [t = 4.94, df = 12, p
= 0.0003]. By contrast, Pt.5 showed the reverse pattern of effects with a significantly lower
magnitude Semantic gender effect [t = 6.15, df = 12, p < 0.0001]. Pt.3 and Pt.4, whose
magnitude did not significantly differ from the controls’, nevertheless showed clear
directionality in the effect (a Semantic gender advantage).
Fig. 3. Accuracy broken down by Gender Type and participant group (control group and PWA). The
error bars indicate SE. The effect of Group is not reported, even though it is significant.
Fig. 4. Magnitude of the Semantic gender effect based on accuracy difference between the Semantic
condition and the Grammatical condition as described in the main text. Individual PWA scores (stars
where the individual performance significantly differed from the controls). More positive values
indicate (proportionally) more accurate responses.
Mean control
Semantic gender effect (accuracy)
Individual PWA scores
Semantic-Grammatical (%)
✱✱✱ ✱✱
5. Discussion
We investigated whether grammatical and semantic gender are processed in different
ways in sentence comprehension as proposed by the maximal input hypothesis (Barlow,
1993; Pollard and Sag, 1994; Vigliocco and Franck, 1999). We used experimental material
similar to that in Vigliocco and Franck (1999), including a distractor with the opposite gender
to that of the sentence subject, so as to induce a relatively high number of errors and avoid
any possible ceiling effects. However, our experiment differed in two key respects from their
study: we used a comprehension task and we tested PWA, a group that is more likely to show
sharper differences in processing the two types of gender.
Our main finding was that semantic gender was processed faster and more accurately than
grammatical gender (in both the Grammatical and Opaque conditions) by both the control
group as well as PWA. This supports the hypothesis that gender type influences semantic
comprehension. Crucially, we found a substantial difference in the magnitude of the Semantic
gender effect in the PWA compared to the control group. The individual analysis highlighted
that this effect was significantly larger for the majority of PWA than for controls. This finding
further confirms our assumption that semantic gender provides salient information that
supports both lexical processing and the computation of agreement. By contrast, although
accuracy was above chance, grammatical gender agreement processing was impaired in PWA
compared to controls (consistently with the cue-based retrieval model for sentence
comprehension, Lewis and Van Dyke, 2003; Van Dyke and McElree, 2006; see also Parker et
al.’s, 2017). The fact that the effect was of larger magnitude in PWA than controls supports our
hypothesis that semantic gender helps PWA by making comprehension (and syntactic)
processes easier; in contrast, computing grammatical gender requires greater cognitive
resources. Below, we discuss this PWA Semantic gender effect in the context of our hypothesis
and consider the role of cognitive control in sentence comprehension.
Since semantic gender reflects the gender of its referent, it facilitates gender agreement
tasks. This is because semantic information is retrieved during lexical access (see Friedmann
et al., 2013) and strictly bound to meaning, two factors that facilitate lexical processing. We
assume that semantic gender originates at the level of the conceptual representations that
underpin lexical concepts, in line with studies that have demonstrated the existence of semantic
priming (see Masson, 1995; Ortells et al., 2006). Importantly, in our task we found the two
gender types were processed differently, suggesting that semantic gender could play an active
role in lexical access. Compatible with cue-based retrieval models, which assume that cue
weights (or salience) determine dependency retrieval (Lewis et al., 2006; Van Dyke and
McElree, 2006; Parker et al., 2017), our results reveal that semantic gender is a higher-level
feature than grammatical gender. The interference we observed was limited to (both) the
grammatical gender conditions, which shared the same gender feature as sentence distractors.
Conversely, semantic gender facilitated computation of agreement because more salient
information was accessible during the retrieval process. For this reason, semantic gender was
preserved in the memory of PWA, reducing the influence of grammatical gender candidates;
by contrast, PWA’s processing of both transparent and opaque grammatical gender was
impaired. This cue-based retrieval model could also explain the lack of statistical difference
between the Opaque and Grammatical conditions, since both have equally low salient features.
This result also confirms that the effect we observed was not driven by the degree of
morphological transparency. If this had been the case, the Opaque condition would have been
slower and less accurate. Caffarra et al., (2016) showed a difference between nouns with
opaque and transparent grammatical gender, which we failed to observe in both groups. We
assume this discrepancy is due to task differences: Caffarra et al.’s task entails explicit use of
gender (i.e. gender categorization of bare nouns), while ours elicits procedural knowledge of
gender in sentence comprehension. In fact, Gollan and Frost (2001) found that Hebrew bare
nouns with irregular gender morphemes (which do not indicate noun gender) are processed less
accurately and more slowly than bare transparent nouns, but this difference was eliminated
these same nouns were computed together with the article to form a full noun phrase
(procedural knowledge).
Impairment of inhibition and cognitive control (Faroqi-Shah et al., 2016) may play a role
in the larger Semantic effect observed in PWA. In the context of sentence comprehension, PWA
were overall able to process agreement at distance. However, their adjective choice mostly
echoed the distractor’s grammatical gender in the Grammatical and Opaque gender conditions,
presumably because they were unable to inhibit this irrelevant but recent gender cue. This
indicates that in PWA non-relevant information (the grammatical gender of the distractor) tends
to reverberate and becomes more difficult to inhibit. Such findings are also in line with Lewis
et al.’s (2006) computational principles of sentence comprehension, which describe how rapid
retrieval of word is hindered by interference from similar items (‘similarity-based interference’,
see also Van Dyke and Johns, 2012).
Alongside deficits in cognitive control, previous research showed that short-term and
working memory explained poor performances in sentence comprehension depending on
syntactic complexity and on the amount of semantic information available (Martin and Romani,
1994; Martin et al., 1994; Martin and He, 2004; Pettigrew and Hillis, 2014). Semantic and
grammatical gender processing may entail two distinct paths one of which may be more taxing
for WM and cognitive control. Processing agreement for grammatical gender relies more
heavily on working memory (WM) – as Sagarra and Herschensohn’s (2010) results suggest –
while the salient information in semantic gender facilitates lexical processing, probably
requiring a smaller contribution from WM. PWA have weaker working memory than controls
when processing sentences, and this may lead to the difficulties they display in the Grammatical
and Opaque conditions relative to controls (see Minkina et al., 2018; Varkanitsa and Caplan,
2018 for a discussion on short-term/working memory in aphasia). In fact, PWA were deceived
by the distractor noun presented in the sentence and were apparently unable to disregard this
distracting feature (i.e., the opposite gender). This irrelevant cue influenced their adjective
choice for gender agreement. Despite the fact that this task put some demands on PWAs’ WM
that could have influenced their performance, we do not believe that WM overload was so
complete as to have completely invalidated their results. If this had been the case, WM overload
would have been reflected by drastic decreases in PWAs accuracy scores. They would have
performed the task with such low accuracy that there would have been no differences observed
between gender types.
In sum, we confirmed our hypothesis that semantic gender and grammatical gender are
processed but incur different levels of difficulty for PWA and healthy participants. We also
found the Semantic gender effect was of greater magnitude in most PWA than controls. This
suggests that semantic gender is retained in this impaired population (regardless of aphasia
type), whereas agreement for grammatical gender of both transparent and opaque
morphology – is partially disrupted. PWA’s procedural knowledge of gender, although worse
than that of controls, is generally spared. This may be the skill they employ during post-lexical
processes when agreement is required (Cacciari and Cubelli, 2003, see also Roelofs et al.,
1998). If so, it was the additional semantic feature exhibited by semantic gender that allowed
PWA to perform similarly to the control group only in the semantic condition.
We had a heterogeneous group of PWA, and we did not measure agrammatism in
production and comprehension. Nevertheless, we analysed individual PWA performances and
the results demonstrated that the responses are unlikely to depend on aphasia type. In fact,
Akhutina et al.’s (2001) work, which reported a gender priming effect in PWA (facilitation
when prime and target were matched for gender), acknowledged that the gender priming effect
did not correlate with any aphasia factor score, symptom, or classification. In addition, some
research has been conducted on heterogeneous groups of PWA. Kulke and Blanken (2001), for
example, grouped patients with various types of aphasia to study the preservation of
grammatical gender. They discovered that in participants with Broca’s and Wernicke’s aphasia,
the level of gender preservation was significantly above chance, with no significant difference
between the two groups.
Moreover, the tasks we employed did not involve speaking or naming, which could have
been more impaired in some participants than in others. Despite possible impairments, the
PWA we analysed were able to complete the experiments and obtain a good accuracy score. It
is known that a given brain lesion may produce a recognizable pattern of errors across different
tasks (Whitworth et al., 2006). Together with neurological evidence, this supports the
classification of aphasia types. Nevertheless, there is considerable variability across patients
with similar lesions or types of aphasia. Two patients with the same lesion or diagnosis may
not perform similarly, even on the same language tasks (Whitworth et al. 2006). Assigning
PWA to particular types of aphasia cannot precisely predict linguistic performance.
Consistently, Grunden et al. (2020) explored voluntary language switching in bilingual aphasia
and demonstrated that PWA with the same aphasia diagnosis have varied and sometimes even
opposite types of language control issues. We compared PWA’s performance to non-brain-
damaged controls and showed that, in most cases, PWA performance was in line with that of
the control participants (see Fig.2 and Fig.5). In addition, Pt.5, who uniquely produced the
opposite pattern of accuracy to the other PWA, also showed a significant Semantic gender
advantage in RT results. Conversely, the Semantic gender effect produced significantly slower
RTs for Pt.1, who nevertheless clearly benefitted from higher accuracy in the same Semantic
condition. We conclude that the presence of different aphasia types did not compromise our
Moreover, we do not consider bilingualism to be a concern for our experiment because we
found no substantial difference between bilinguals and monolinguals in the analysis of the sub-
set of bilinguals. Note that both groups of participants had monolinguals and bilinguals and the
gender of the stimuli was the same in both Spanish and Catalan. Future investigations should
probe effects in bilinguals whose languages differ in the ascription of gender to check whether
bilingualism is a confounding factor.
6. Conclusion
Semantic gender is assumed to be the first type of gender retrieved because this retrieval
happens concurrently with lexical access at an early stage of processing. Semantic gender may
also enhance word recognition in a sentence and – since it is bound to noun meaning – may
induce longer-lasting effects than grammatical gender, which is retrieved later. This viewpoint
is also supported by evidence from PWA, who were slower to inhibit irrelevant stimuli when
processing agreement at distance. Further research using declarative knowledge of semantic
gender in PWA will shed light on the nature of semantic gender processing.
MC was supported by the postdoctoral Ramón y Cajal fellowship (RYC-2013-14013),
Agencia Estatal de Investigación (AEI, National Research Agency), and Fondo Europeo de
Desarrollo Regional (FEDER, European Regional Development Fund) under project PSI2017-
Disclosure of interest
The authors report no conflict of interest.
Akhutina, T., Kurgansky, A., Kurganskaya, M., Polinsky, M., Polonskaya, N., Larina, O., Bates,
E., & Appelbaum, M. (2001). Processing of Grammatical Gender in Normal and Aphasic
Speakers of Russian. Cortex, 37(3), 295326.
Balota, D. A. (1992). Visual word recognition: The journey from feature to meaning. In M.
Gernsbacher (ed.), Handbook of psycholinguistics, 303-357. New York: Academic Press.
Barlow, D. H. (Ed.). (1993). Clinical handbook of psychological disorders: A step-by-step
treatment manual, 2nd ed (pp. x, 534). The Guilford Press.
Bates, E. (1999). Language and the infant brain. Journal of Communication Disorders, 32(4),
Bates, E., Devescovi, A., Pizzamiglio, L., D'Amico, S., & Hernandez, A. (1995). Gender and
lexical access in Italian. Perception and Psychophysics, 57(6), 847-862.
Bates, E., Marangolo, P., Pizzamiglio, L., & Dick, F. (2001). Linguistic and Nonlinguistic Priming
in Aphasia. Brain and Language, 76(1), 6269.
Berg, T., & Levelt, W. J. M. (1990). Speaking: From Intention to Articulation. The American
Journal of Psychology, 103(3), 409.
Bock, K., & Levelt, W. J. M. (1994). Language Production: Grammatical Encoding. In M. A.
Gernsbacher (Ed.), Handbook of Psycholinguistics. pp. 945-984. San Diego, CA: Academic
Cacciari, C., & Cubelli, R. (2003). The Neuropsychology of Grammatical Gender: An
Introduction. Cortex, 39(3), 377382.
Caffarra, S., Zimnukhova, S., & Mancini, S. (2016). What usage can do: The effect of language
dominance on simultaneous bilinguals’ morphosyntactic processing. Linguistics Vanguard,
Chomsky, N. (1965). Aspects of the Theory of Syntax. Cambridge: MIT press.
Chomsky, N. (1981). Lectures on government and binding: the Pisa lectures. Holland: Foris
Corbett, G. G. (1991). Gender. Cambridge University Press.
Corbett, G. G. (2006). Agreement. Cambridge University press.
Crawford, J. R., & Howell, D. C. (1998). Comparing an Individual’s Test Score Against Norms
Derived from Small Samples. The Clinical Neuropsychologist, 12(4), 482486.
Damian, M. F. (2010). Does variability in human performance outweigh imprecision in response
devices such as computer keyboards? Behavior Research Methods, 42(1), 205211.
Davis, C. J., & Perea, M. (2005). BuscaPalabras: A program for deriving orthographic and
phonological neighborhood statistics and other psycholinguistic indices in Spanish. Behavior
Research Methods, 37(4), 665671.
Faroqi-Shah, Y., Sampson, M., Pranger, M., & Baughman, S. (2016). Cognitive control, word
retrieval and bilingual aphasia: Is there a relationship? Journal of Neurolinguistics.
Forster, K. I., & Forster, J. C. (2003). DMDX: A Windows display program with millisecond
accuracy. Behavior Research Methods, Instruments, & Computers, 35(1), 116124.
Franzon, F., Peressotti F., Arcara G., & Semenza C. (2013). Gender Agreement: a comparate
psycholinguistic and aphasia case study. Academy of Aphasia-52nd Annual Meeting.
Frontiers in Psychology, 5-7.
Friedmann N., Biran, M., & Dotan, D. (2013). Lexical retrieval and its breakdown in aphasia and
developmental language impairment. In Boeckx, C., & Grohmann, K.K. (Eds.), The
Cambridge Handbook of Biolinguistics. Cambridge: Cambridge University Press.
Friedmann, N., & Biran, M. (2003). When is Gender Accessed? A Study of Paraphasias in
Hebrew Anomia. Cortex, 39(3), 441463.
Friedmann, N., Biran, M., & Dotan, D. (2013). Lexical retrieval and its breakdown in aphasia and
developmental language impairment. In C. Boeckx & K. K. Grohmann (Eds.), The
Cambridge Handbook of Biolinguistics (pp. 350374). Cambridge University Press.
Gollan, T. H., & Frost, R. (2001). Two routes to grammatical gender: Evidence from Hebrew.
Journal of Psycholinguistic Research, 30(6), 627651.
Goodglass, H. (2000). Grammatical gender is not always syntactic. In Grodzinsky, Y., Shapiro,
L.P., & Swinney, D. (Eds), Language and the brain: representation and processing. San
Diego: Academic Press.
Grunden, N., Piazza, G., García-Sánchez, C., & Calabria, M. (2020). Voluntary Language
Switching in the Context of Bilingual Aphasia. Behavioral Sciences, 10(9), 141.
Kempen, G., & Hoenkamp, E. (1987). An Incremental Procedural Grammar for Sentence
Formulation. Cognitive Science, 11(2), 201258.
Kertesz, A. (1982). The Western Aphasia Battery: Test Manual, Stimulus Cards, Test Booklets
(Test Kit). New York: Grune and Stratton.
Kertesz, A., Pascual, Á. P.-L., & García, A. P.-L. (1990). Batería de afasías “Western”: : (The
western aphasia battery en versión y adaptación castellana). Manual del test. Nau Llibres.
Kulke, F., & Blanken, G. (2001). Phonological and syntactic influences on semantic misnamings
in aphasia. Aphasiology, 15(1), 315.
Levelt, W. J. M. (1989). Speaking: From Intention to Articulation. Cambridge, MA: MIT Press.
Lewis, R. L., Vasishth, S., & Van Dyke, J. A. (2006). Computational principles of working
memory in sentence comprehension. Trends in Cognitive Sciences, 10(10), 447454.
MacDonald, M. C., Pearlmutter, N. J., & Seidenberg, M. S. (1994). The lexical nature of syntactic
ambiguity resolution. Psychological Review, 101(4), 676703.
Manenti, R., Repetto, C., Bentrovato, S., Marcone, A., Bates, E., & Cappa, S.F. (2004). The
effects of ageing and Alzheimer’s disease on semantic and gender priming. Brain, 127(10),
Martin, R.C., & He, T. (2004). Semantic short-term memory and its role in sentence processing: A
replication. Brain and Language, 89(1), 7682.
Martin, R.C., & Romani, C. (1994). Verbal working memory and sentence comprehension: A
multiple-components view. Neuropsychology, 8(4), 506523.
Martin, R.C., Shelton, J. R., & Yaffee, L. S. (1994). Language Processing and Working Memory:
Neuropsychological Evidence for Separate Phonological and Semantic Capacities. Journal of
Memory and Language, 33(1), 83111.
Masson, M. E. J. (1995). A distributed memory model of semantic priming. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 21(1), 323.
Miceli, G., Turriziani, P., Caltagirone, C., Cappaso, R., Tomaiuolo, F., Caramazza, A. (2002). The
neural correlates of grammatical gender: An fMRI investigation. Journal of Cognitive
Neuroscience, 14: 618-628, 2002.
Minkina, I., Salis, C., & Martin, N. (2018). Short-term and working memory deficits in aphasia:
Current issues in theory, evidence, and treatment. Journal of Neurolinguistics, 48, 13.
Mondini, S., Luzzatti, C., & Semenza, C. (1999). Grammatical gender knowledge in an Italian
agrammatic patient. Brain and Language, 69(3), 278281.
Ortells, J. J., Vellido, C., Daza, M. T., & Noguera, C. (2006). Efectos de priming semántico con y
sin conciencia perceptiva. [Semantic priming effects with and without perceptual
awareness.]. Psicológica, 27(2), 225242.
Parker, D., Shvartsman, M., & Van Dyke, J. A. (2017). The cue-based retrieval theory of sentence
comprehension: New findings and new challenges. In L. Escobar, V. Torrens, & T. Parodi
(Eds.) Language Processing and Disorders (pp. 121-144). Newcastle: Cambridge Scholars
Pettigrew, C. & Hillis, A. E. (2014) Role for memory capacity in sentence comprehension:
Evidence from acute stroke. Aphasiology, 28:10, 1258-1280,
Pollard, C., & Sag, I. A. (1994). Head-Driven Phrase Structure Grammar. Chicago: The
University of Chicago Press.
Repetto, C., Manenti R., Cappa S.F., Miniussi C., & Riva G. (2009). Semantic and gender priming
in frontotemporal dementia. Annual Review of Cybertherapy and Telemedicine, 237-239.
Roelofs, A., Meyer, A.S., & Levelt, W. J. M. (1998). A case for the lemma/lexeme distinction in
models of speaking: Comment on Caramazza and Miozzo (1997). Cognition, 69, 219-230.
Sagarra, N., & Herschensohn, J. (2010) The role of proficiency and working memory in gender
and number agreement processing in L1 and L2 Spanish. Lingua, 120(8), 2022-2039.
Scarnà, A., & Ellis, A. W. (2002). On the assessment of grammatical gender knowledge in
aphasia: The danger of relying on explicit, metalinguistic tasks. Language and Cognitive
Processes, 17(2), 185201.
Schiller, N. O., Munte, T. F., Horemans, I., & Jansma, B. M. (2003). The influence of semantic
and phonological factors on syntactic decisions: An event-related brain potential study.
Psychophysiology, 40, 869869.
Schiller, N. O., Schuhmann, T., Neyndorff, A. C., & Jansma, B. M. (2006). The influence of
semantic category membership on syntactic decisions: A study using event-related brain
potentials. Brain Research, 1082(1), 153164.
Semenza, C., & Luzzatti, C. (2019) Disturbi lessicali nell'afasia. In Denes et al. (Eds), Manuale di
Neuropsicologia, Zanichelli, Bologna.
Van Dyke, J. A. (2007). Interference effects from grammatically unavailable constituents during
sentence processing. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 33(2), 407430.
Van Dyke, J. A., & Johns, C. L. (2012). Memory Interference as a Determinant of Language
Comprehension: Interference in Comprehension. Language and Linguistics Compass, 6(4),
Van Dyke, J. A., & McElree, B. (2006). Retrieval interference in sentence comprehension.
Journal of Memory and Language, 55(2), 157166.
Varkanitsa, M., & Caplan, D. (2018). On the association between memory capacity and sentence
comprehension: Insights from a systematic review and meta-analysis of the aphasia literature.
Journal of Neurolinguistics, 48, 425.
Vigliocco, G., Butterworth, B., & Semenza, C. (1995). Constructing Subject-Verb Agreement in
Speech: The Role of Semantic and Morphological Factors. Journal of Memory and
Language, 34(2), 186215.
Vigliocco, Gabriella, & Franck, J. (1999). When Sex and Syntax Go Hand in Hand: Gender
Agreement in Language Production. Journal of Memory and Language, 40(4), 455478.
Vigliocco, Gabriella, Butterworth, B., & Garrett, M. F. (1996). Subject-verb agreement in Spanish
and English: Differences in the role of conceptual constraints. Cognition, 61(3), 261298.
Whitworth, A, Webster J, & Howard D. (2006). A Cognitive neuropsychological approach to
assessment and intervention in aphasia. Hove and New York: Psychology Press.
1 Throughout this paper we will refer to the gender of the referent and biological sex as the
source for semantic gender. Without doubt, gender and sex have two substantially different meanings.
Nevertheless, for the purpose of this paper, it is not always easy to determine which is the more
appropriate term, because semantic gender pertains to both animals and humans. When referring to
the semantic gender of human nouns, it is probably more correct to say it originates from gender,
although (but this is not the aim of the paper) we do not know how semantic gender would be
conceptualized in the case of non-correspondence between gender identity and sex.
Table 1. Individual scores for the Western Aphasia Battery
Type of aphasia
(max 20)
(max. 10)
(max. 10)
APPENDIX A. Examples of sentences predicative and adjectives.
El esposo en la iglesia está
Semantic condition
The spouse in the church is
El mono en la clínica está
Semantic condition
The monkey in the clinic is
La cuchara en el suelo está
Grammatical condition
The spoon on the ground is
La comida en el plato está
Grammatical condition
The food on the dish is
La fuente del suelo está
Opaque condition
The fount on the ground is
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
As studies of bilingual language control (BLC) seek to explore the underpinnings of bilinguals' abilities to juggle two languages, different types of language switching tasks have been used to uncover switching and mixing effects and thereby reveal what proactive and reactive control mechanisms are involved in language switching. Voluntary language switching tasks, where a bilingual participant can switch freely between their languages while naming, are being utilized more often due to their greater ecological validity compared to cued switching paradigms. Because this type of task had not yet been applied to language switching in bilingual patients, our study sought to explore voluntary switching in bilinguals with aphasia (BWAs) as well as in healthy bilinguals. In Experiment 1, we replicated previously reported results of switch costs and mixing benefits within our own bilingual population of Catalan-Spanish bilinguals. With Experiment 2, we compared both the performances of BWAs as a group and as individuals against control group performance. Results illustrated a complex picture of language control abilities, indicating varying degrees of association and dissociation between factors of BLC. Given the diversity of impairments in BWAs' language control mechanisms, we highlight the need to examine BLC at the individual level and through the lens of theoretical cognitive control frameworks in order to further parse out how bilinguals regulate their language switching.
Full-text available
Even when bilinguals learn both languages from birth and achieve high levels of proficiency, they rarely use their languages to the same degree. Recent findings suggest that individual differences in bilingual profile such as the usage of the bilingual's different languages could affect the way they retrieve and analyse linguistic information, with greater use of linguistic mechanisms from the dominant language. One of the linguistic areas where a wide variety of bilingual performance has been reported is morphosyntax. The present study tests whether language usage can account for a certain amount of the individual variability in morpho-syntactic feature extraction. Basque-Spanish simultaneous bilinguals with a range of language dominance profiles were asked to judge the grammatical gender of Spanish nouns the ending of which could provide a reliable cue to gender (i. e., transparent) or not (i. e., opaque). Results showed that the more bilinguals used Basque (i. e., an agglutinative language) on a daily basis, the faster they were at detecting the presence of transparent morphemes relative to opaque nouns. These findings suggest that simultaneous bilinguals have different ways of retrieving grammatical gender depending on their language profile. Language usage can contribute to explaining the presence of individual differences in morphosyntactic feature retrieval.
In this article, we introduce a special issue of the Journal of Neurolinguistics that focuses on current research in short-term and working memory impairment in aphasia. The papers are summarized briefly and discussed with respect to their individual and collective contributions to our understanding of the role of short-term and working memory processes in language and its impairment in aphasia. The timeliness of the special issue is noted, as it sets the stage for future work in this area.
An active question in psycholinguistics concerns how speakers mentally encode and retrieve linguistic representations in memory during sentence comprehension. In this paper, we briefly review contemporary perspectives on memory retrieval in sentence comprehension, focusing on accounts of dependency resolution, and present new puzzles and challenges for existing models. We propose that the current data support the conclusion that linguistic dependencies are resolved using a direct-access, cue-based retrieval mechanism that gives preferential weighting to syntactic information when navigating linguistic representations in memory. We then discuss current empirical and theoretical gaps, highlighting areas in need of further research.
It is proposed that successful word retrieval involves lateral inhibition of lexical competitors, and suppression of the non-target language in bilingual speakers. Thus cognitive control is crucial for word production. Given that word retrieval difficulty is a hallmark feature of aphasia, the relationship between word retrieval and cognitive control in aphasia has not been sufficiently explored. This study examined whether persons with aphasia show 1) evidence of a cognitive control deficit, 2) bilingual status interacts with cognitive control deficit in persons with aphasia, and (3) a relationship between measures of word naming and cognitive control. Thirty-eight persons with aphasia were administered a task of cognitive control (Stroop color-word task) and two word production tasks (picture naming and category fluency). We found weakened cognitive control in aphasia relative to age-matched neurologically healthy adults. A bilingual advantage in cognitive control was found in neurologically healthy adults and in one group of bilingual speakers with aphasia, but not the other group. Word retrieval in persons with aphasia was not correlated with Stroop task performance. These findings show that cognitive control performance (as measured by the Stroop task) is compromised in persons with aphasia, irrespective of bilingual status. There was a bilingual advantage in two out of three groups, showing a general support for the bilingual inhibitory control advantage (BICA) hypothesis.
Background: Previous research has suggested that short-term and working memory resources play a critical role in sentence comprehension, especially when comprehension mechanisms cannot rely on semantics alone. However, few studies have examined this association in participants in acute stroke, before the opportunity for therapy and reorganization of cognitive functions. Aims: The present study examined the hypothesis that severity of short-term memory deficit due to acute stroke predicts the severity of impairment in the comprehension of syntactically complex sentences. Furthermore, we examined the association between damage to the short-term and working memory network and impaired sentence comprehension, as an association would be predicted by the previous hypothesis. Methods & procedures: 47 participants with acute stroke and 14 participants with a transient ischemic attack (TIA; the control group) were included in the present study. Participants received a language battery and clinical or research scans within 48 hours of hospital admittance. The present study focused on the behavioral data from the short-term memory and working memory span tasks and a sentence-picture matching comprehension task included in this battery. Using regression analyses, we examined whether short-term and working memory measures explained significant variance in sentence comprehension performance. Outcomes & results: Consistent with prior research, short-term memory explained significant variance in sentence comprehension performance in acute stroke; in contrast, working memory accounted for little variance beyond that which was already explained by short-term memory. Furthermore, ischemia that included the short-term/working memory network was sufficient to cause sentence comprehension impairments for syntactically complex sentences. Conclusions: The present study suggests that short-term memory resources are an important source of sentence comprehension impairments.