Content uploaded by Adriana Hanulíková
Author content
All content in this area was uploaded by Adriana Hanulíková on Dec 13, 2018
Content may be subject to copyright.
For Review Only
Processing gender stereotypes in dementia patients and
older healthy adults: a self-paced reading study
Journal:
Linguistics Vanguard
Manuscript ID
LingVan.2018.0029.R2
Manuscript Type:
research-article
Classifications:
Psycholinguistics & Neurolinguistics
Keywords:
aging, dementia, gender stereotypes, self-paced reading, sentence
processing
Abstract:
Effects of stereotypical knowledge on language processing have been
frequently observed in young adults, but little is known about age-
related changes in the activation and integration of stereotypical
information. We used self-paced reading to examine whether
stereotypical associations of verbs with women or men as prototypical
agents (e.g. the craftsman knits a sweater) are activated during
sentence processing in dementia patients and healthy older adults. While
syntactic processing may remain intact, semantic capacities are often
affected in dementia. Since inferences based on gender stereotypes
draw on social and world knowledge, access to stereotype information
may also be affected in dementia patients. Results from dementia
patients (n = 9, average age 86.6) and healthy older adults (n = 14,
average age 79.5) showed slower reading times and less accuracy in
comprehension scores for dementia patients compared to the control
group. While activation of stereotypical associations of verbs was visible
in both groups, they differed with respect to the time-course of
processing. The effect of stereotypes on comprehension accuracy was
visible for healthy adults only. The evidence from reading times suggests
that older adults with and without dementia engage stereotypical
inferences during reading, which is in line with research on young adults.
Linguistics Vanguard
Linguistics Vanguard
For Review Only
Running title: Gender stereotypes in dementia and older adults
1
Processing gender stereotypes in dementia patients and older
healthy adults:
a self-paced reading study
*Daniel Müller-Feldmeth, Katharina Ahnefeld, *Adriana Hanulíková
University of Freiburg, Freiburg, Germany
Corresponding author:
Adriana Hanulíková
Albert-Ludwigs-Universität Freiburg
Deutsches Seminar–Germanistische Linguistik
Platz der Universität 3, 79098 Freiburg
Tel.: 0049 761 2039341
Adriana.Hanulikova@germanistik.uni-freiburg.de
*Shared authorship
Acknowledgment: We would like to thank Maria Silveira for assistance with the rating
experiment. Author contribution statement: A.H. conceived, designed and planned the
experiments. K.A. collected the data under the supervision of A.H. D.M.F. carried out the
data analysis. D.M.F. and A.H. wrote the paper.
Page 1 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
Abstract
Effects of stereotypical knowledge on language processing have been frequently observed in
young adults, but little is known about age-related changes in the activation and integration
of stereotypical information. We used self-paced reading to examine whether stereotypical
associations of verbs with women or men as prototypical agents (e.g. the craftsman knits a
sweater) are activated during sentence processing in dementia patients and healthy older
adults. While syntactic processing may remain intact, semantic capacities are often affected
in dementia. Since inferences based on gender stereotypes draw on social and world
knowledge, access to stereotype information may also be affected in dementia patients.
Results from dementia patients (n = 9, average age 86.6) and healthy older adults (n = 14,
average age 79.5) showed slower reading times and less accuracy in comprehension scores
for dementia patients compared to the control group. While activation of stereotypical
associations of verbs was visible in both groups, they differed with respect to the time-
course of processing. The effect of stereotypes on comprehension accuracy was visible for
healthy adults only. The evidence from reading times suggests that older adults with and
without dementia engage stereotypical inferences during reading, which is in line with
research on young adults.
Keywords: aging, dementia, gender stereotypes, self-paced reading, sentence processing
Page 2 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
1 Introduction
Syntactic and semantic processes during language comprehension are intimately
intertwined, and comprehenders make extensive use of world knowledge acquired
throughout their life span. This knowledge includes information about typical or probable
situations, e.g. what kind of action particular agents frequently or plausibly perform. Upon
reading a sentence such as the cat knits, readers do not only establish the morphosyntactic
relationship between the subject cat and the predicate knits; they also consider their world
knowledge in sentence meaning construction. Unless there is a preceding discourse about
imaginative cats, readers will perceive a mismatch between the sentence meaning and their
world knowledge (i.e. cats do not knit). In contrast, the sentence the man knits, is not
entirely inconsistent with our world knowledge of what humans are able to do (men have
the capability to knit), but it may be evaluated less probable compared to the sentence
containing a female agent. Since people gain knowledge about how often events co-occur,
their expectations about who is likely to knit may be violated. Such expectations can lead to
overgeneralizations of gender-based characteristics, resulting in gender stereotyping.
Evidence from numerous language-processing studies and a wide range of methods has
shown that stereotypical knowledge is immediately and automatically activated and very
difficult to suppress (e.g. Oakhill et al. 2005). Effects of gender stereotypes occur in many
contexts, for example when grammatical gender of the anaphora (he/she) mismatches the
gender stereotype of a preceding noun (e.g. doctor/nurse) (e.g. Duffy and Keir 2004;
Esaulova et al. 2014; Kennison and Trofe 2003; Kreiner et al. 2008; Reali et al. 2015;
Siyanova-Chanturia et al. 2012) as well as in gender stereotypical adjectives (White et al.
2009) and verbs (Müller-Feldmeth and Hanulíková submitted). Despite a large number of
Page 3 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
4
empirical studies, theoretical debates continue about how gender stereotypes are stored
and accessed during processing. While the minimalist approach (McKoon and Ratcliff 1992)
argues that inferences from world knowledge are drawn upon only if necessary to establish
discourse coherence, a number of studies provide evidence that gender stereotypes are
activated automatically and immediately upon encountering role nouns (Garnham et al.
2002; Oakhill et al. 2005), indicating that stereotypical gender is stored and retrieved as part
of the lexical representation of a role noun, and inferences are immediately used in
attempting to construct a mental model of sentence meaning (Carreiras et al. 1996).
However, comparing nouns with definitional gender (e.g. king) to nouns with stereotypical
gender (e.g. minister), eye-tracking (e.g. Esaulova et al. 2014; Kreiner et al. 2008) and ERP
studies (Siyanova-Chanturia et al. 2012) indicate that using definitional and stereotypical
gender cues employ different processes. It has also been shown that, although stereotype
information is activated in neutral contexts, it can be overridden if an agent’s gender is
explicitly provided in the previous context (e.g. Duffy and Keir 2004; Kreiner et al. 2008).
Thus, researchers largely agree that activation of gender stereotypes is immediate and
automatic, but it is still under debate how stereotypical gender is represented, and which
mechanisms are involved in making stereotypical inferences.
The formation of probabilistic stereotypical inferences may result from world knowledge as
well as from repeated exposure to lexemes in gender-specific contexts. While it is well
known that world knowledge increases with age (e.g. Clark-Cotton et al. 2007; Park et al.
2002; Park and Gutchess 2006), it remains poorly understood whether and how
stereotypical inferences during language processing change throughout the adult life span.
Most research on gender stereotyping has been conducted with young adults, and only a
few studies consider age-related changes in activation or inhibition of stereotypical
Page 4 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
5
inferences. It has been suggested that older adults are more likely than younger adults to
rely on stereotypes and prejudice in general (von Hippel et al. 2000), and that their skills to
inhibit stereotypical inferences during reading narrative texts are reduced (Radvansky et al.
2010). Using self-paced reading, however, Radvansky et al. (2009) demonstrated that effects
of unwanted stereotypes can be reduced based on counter-stereotypical information in both
young and older adults. Siyanova-Chanturia et al. (2015) found no age-related changes
(comparing children, young adults and older adults) with respect to sensitivity to gender
congruency between a target word and a stereotypical gender use of a preceding prime.
However, there is still little research into on-line processing of gender stereotypes during
sentence comprehension in older adults and clinical populations. Using self-paced reading,
the aim of this study was to examine whether stereotypical associations (e.g. Der
Handwerker strickt einen Pullover ‘the craftsman knits a sweater’) are activated during
sentence processing in older adults with and without signs of cognitive impairment.
It would be reasonable to assume little qualitative differences in processing gender
stereotypes between young and older adults, but there might be differences in the
magnitude of gender effects due to increased world knowledge as well as weakened
inhibitory mechanisms with age (e.g. Kemper and McDowd 2006; Radvansky et al. 2010). In
contrast to the majority of cognitive functions, world-knowledge and knowledge-based
verbal ability tend to increase with age (e.g. Park et al. 2002). While many aspects of
language comprehension remain intact until very late among healthy older adults (e.g.
Federmeier et al. 2010; for reviews, see Burke and Shafto 2008; Wingfield and Stine-Morrow
2000), older adults with pathological cognitive conditions such as dementia show patterns of
impaired comprehension (for a review, see Kempler and Goral 2008).
Page 5 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
6
Dementia refers to deficits in attention, memory and behavior that are associated with
neurological impairments, and are severe enough to reduce a person's ability to perform
everyday activities. Across different types of dementia, linguistic abilities and
comprehension are often severely impaired (Kempler and Goral 2008, Reilly and Peele
2008). It is often unclear whether changes in linguistic abilities indicate impairments of
genuinely language-related functions or stem from declines in general cognitive functions
such as working memory, attention and inhibition (Kempler and Goral 2008). In general,
while syntactic processing may remain intact, semantic capacities are often affected at the
onset of dementia (e.g. Clark-Cotton et al. 2007). Along with impairments of semantic
knowledge, world-knowledge and gender stereotypes may become less accessible during
language comprehension in dementia. Some studies suggest that stereotype processing is
distinct from retrieval of general semantic knowledge about objects and non-human living
things, but is strongly related to social cognition, including the representation of others’
mental states (e.g. Contreras et al. 2012). Thus, it remains an open question whether gender
stereotypes are accessible to dementia patients during sentence processing. In the current
study, we investigate the extent to which gender stereotypical information is activated
during reading among healthy older adults and among older adults who meet clinical criteria
for dementia.
2 Current study
Using self-paced reading, we examine whether stereotypical associations of verbs with
women or men as prototypical agents (e.g. the craftsman knits a sweater) are activated and
integrated during sentence processing in older adults with and without dementia. A variety
of methods have been used to investigate stereotypical inferences. Explicit ratings such as
Page 6 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
7
judging whether two nouns refer to the same referent (Siyanova-Chanturia et al. 2015) or
priming studies (White et al. 2018) usually present lexical elements in isolation. As such, they
provide useful, yet limited, information with respect to stereotype integration in on-line
sentence processing. One of the most frequently used methods for examining integration of
stereotypical inferences in sentence comprehension is the self-paced reading paradigm. It
allows on-line evaluation of sentence processing while requiring minimal equipment, easy
set up and no specific calibration procedures (in contrast to EEG or eye-tracking). The
experimental sessions can be kept relatively short, and can be conducted within the familiar
environment of the participants, minimizing the participation burden on patients. In line
with previous research, we chose a word-by-word presentation to measure processing time
at each word in the sentence, and we used comprehension questions following each
sentence to assess comprehension accuracy, task difficulty and memory limits.
So far, most studies investigated the influence of stereotypical nouns (e.g. occupations) on
reading behavior in the context of pronominal and noun phrase anaphors, adjectives or
other discourse contexts. In this study, we examine reading behavior in the context of verbs
that are associated with women or men as prototypical agents. We test whether occupation
nouns – explicitly marking biological gender using grammatical features – affect the
integration of these verbs. The verb to knit may be associated with a female agent, while a
male agent performing this action may induce a mismatch effect, leading to increased
processing difficulty and longer reading times. In a self-paced reading study with university
students, Müller-Feldmeth and Hanulíková (submitted) presented participants matching or
mismatching gender stereotypes in German sentences such as der Handwerker strickt einen
Pullover und gibt ihn seiner Großmutter ‘the craftsman knits a sweater and gives it to his
grandmother’. Upon reading a noun representing a human referent, readers activate its
Page 7 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
8
grammatical and biological gender, and maintain this information for contextual integration.
When reading the following action verb, the meaning and stereotypical association of the
verb should be accessed and integrated, with the agent performing the respective action. If
the agent’s biological gender generates expectations about what actions this agent is likely
to do, the reading times should increase in the case of incongruent associations between the
verb and the agent. This is exactly what Müller-Feldmeth and Hanulíková found. In line with
previous research, increased reading times on the word following a stereotypical verb were
observed and confirm integration of stereotypical associations of verbs during linguistic
processing in young adults.
In the current study, we applied the same design and materials as Müller-Feldmeth and
Hanulíková, using sentences where the subject’s gender either matched (e.g. Die
Handwerkerin strickt… / Der Kosmetiker prügelt… ‘the craftswoman knits… / the
beauticianmasc beats…’) or mismatched (e.g. Der Handwerker strickt… / Die Kosmetikerin
prügelt… ‘the craftsman knits… / the beauticianfem beats…’) the agent stereotypicality of the
verb (see Table 1). Under the assumption that gender stereotypes, as part of world-
knowledge, may not be accessed across different types of dementia, though should be
available to healthy older adults, we expected effects of gender stereotypes on reading to
diminish in dementia patients compared to healthy older adults. The mismatch effect
expected for healthy older adults should appear at the word position after the verb, in line
with Müller-Feldmeth and Hanulíková (submitted). We considered three regions of interest
for the analyses: the verb (verb), the position following the verb, which was either a
determiner or a preposition (verb+1), and the noun that completes the verb phrase (verb+2).
Since working memory is related to sentence processing and is commonly affected in
dementia patients, we also measured participants’ digit span performance. We predicted
Page 8 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
9
lower digit span score for dementia patients than for healthy older adults, and higher digit
span readers were expected to be faster and to reach higher accuracy scores.
3 Methods
3.1 Participants
Data collection took place at a nursing home. Fourteen older adults (2 men, mean age 79.5
years, SD 8.2, 11 with a high to medium educational background) without known cognitive
impairment and 9 older adults (3 men, mean age 86.5, SD 4.1, 6 with a high to medium
educational background n = 6) with clinically diagnosed mild to moderate dementia of
unspecified type completed the study after signing an informed consent (two additional
dementia patients were tested but excluded because they were not able to complete the
task). All participants were native speakers of German and did not show signs of hearing or
reading problems. Dementia patients were selected based on consultation with the medical
and nursing staff, which provided a subjective evaluation of patients’ language skills. In
addition, participants were asked to read simple sentences on the computer screen prior to
the experimental session to ensure that they had no difficulties with reading and would be
able to handle the task.
3.2 Materials
Sixteen German sentences were constructed, in which the sentence subject’s gender either
matched or mismatched the agent stereotypicality of the verb (see Table 1). Critical
sentences consisted of 10 words and had an identical word order.
Table 1. Example of sample sentences with English translations. Critical verbs are underlined.
Page 9 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
10
Die Handwerkerin/der Handwerker strickt einen Pullover und schenkt ihn ihrer/seiner Oma.
‘The craftswoman/craftsman knits a sweater and gives it to her/his granny.’
Die Kosmetikerin/der Kosmetiker prügelt den Boxsack und hört dabei laut Musik.
‘The beauticianfem/beauticianmasc beats the punching bag and listens to loud music.’
To control for stereotypicality of the occupations, half of the items consisted of
stereotypically male job roles (e.g. Handwerker/Handwerkerin ‘craftsman/craftswoman’),
and the other half of stereotypically female job roles (e.g. Kosmetiker/Kosmetikerin
‘beauticianmasc/fem‘). The job role stereotypicality was determined based on a large-scale
ratings study for several languages including German (Misersky et al. 2014). Note that
masculine nouns referring to human beings, such as Handwerker 'craftsman', are generic
and thus occur more frequently than the feminine counterparts. The verbs were selected
based on a rating study of 111 German action verbs by 60 native speakers of German (mean
age 26.1 years; 38 women). Each verb was rated on a 7-point Likert scale, where 1 meant
typically female activity and 7 meant typically male activity. The majority of the verbs were
rated as neutral. We selected 8 verbs that were rated as clearly female (kreischen 'to
scream’, kichern ‘to snicker’, bügeln ‘to iron’, schminken ‘to make so. up’, stricken ‘to knit’,
reiten ‘to ride’1, backen ‘to bake’, häkeln ‘to crochet’) and 8 verbs that were rated as clearly
male (zocken ‘to gamble’, kicken ‘to kick’, sägen ‘to saw’, kämpfen ‘to fight’, jagen ‘to hunt’,
saufen ‘to swig’, prügeln ‘to beat’, boxen ‘to box’). The agent-verb combinations were
selected such that the verbs were typical either for a male or a female agent but not for the
respective occupation. Target items were distributed across two lists such that an item
1 In German, reiten refers to ‘ride a horse’ (or other animals) and is usually associated with female referents.
Thus, stereotypical associations with the English verb ‘to ride’ may differ from the German verb reiten.
Page 10 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
11
appeared in a list in one of the two conditions (match or mismatch). In addition, 16 simple
German sentences with either male or female agents and neutral verbs were constructed as
fillers. Item order was randomized for each participant. For each sentence a simple
comprehension question was constructed (e.g. Schenkt der Handwerker seiner Oma einen
Schal? ‘Does the craftsmen give a scarf to his granny?’).
3.3 Procedure
Participants were tested individually in their private rooms (dementia patients) or in a quiet
room in the nursing home. They were seated in front of a portable computer and were given
instructions to press the spacebar to read sentences word by word. After a short practice
with two sentences and two comprehension questions, participants read 32 sentences and
answered comprehension questions after each sentence. No accuracy feedback was given.
Participants were then asked to complete a digit-span task. Each participant was presented
with sequences of digits ranging from 3 to 7 on the computer screen and was asked to write
down the digit sequences on a structured answer sheet. Each sequence length was
presented twice. Participants started with two practice trials of length 2 and 3. The test
stopped as soon as the participant made errors in both sequences for a particular length.
The maximal number of digits with at least one correct sequence and percentage of errors
were used as a measure of digit span. After this task, participants completed a questionnaire
about their reading habits, languages, professional and educational background.
3.4 Data Analysis
Reading times and comprehension accuracy were used as dependent measures. To avoid
spurious results due to extreme values, we excluded outliers based on boxplots, separately
for each participant and critical word region (verb, verb+1, verb +2). Values below the first
Page 11 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
12
quartile minus 3 times the interquartile range or above the third quartile plus 3 times the
interquartile range were conceived as extreme values and excluded (3.7% of the data).
The remaining data for the three regions of interest were analyzed separately with linear
mixed effect models using the lme4 package (Bates et al. 2015a) in R (R Core Team 2018).
Reading times were log-transformed to achieve the precondition of normally distributed
residuals. Sum coding was used to allow the interpretation of the main effects of predictors
included in an interaction. The models included compatibility (compatible/incompatible)
between the agent’s gender, verb stereotype (male/female), and group (dementia/control)
as fixed factors as well as the interaction term of compatibility and group. Two additional
fixed effects were included; the log-transformed reading times of the preceding item
(log.rt.min.1), because reading times of a preceding word are highly predictive of the reading
times of the following word (Mitchell 1984; Vasishth 2006). To account for effects of word
length, the number of letters within the region of interest was also included (word length).
The maximal model that converged (Barr et al. 2013) and/or did not show strong signs of
overparametrization (Bates et al. 2015b) was calculated by stepwise reduction of complexity
of random effects structure. The final model included random intercepts for both participant
and item, and a random slope of word length for the participant term. In the following
sections, only the results of these parsimonious models are reported. P-values were
calculated using the afex package (Singmann et al. 2018) with Kenward-Roger approximation
for degrees of freedom, as recommended in Luke (2017).
Page 12 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
13
4 Results
4.1 Comprehension accuracy and digit span
Mean proportion of correctly answered comprehension questions (including filler sentences)
was 0.91 (SD 0.29; range 0.81 - 0.97) for the healthy older adults and 0.70 (SD 0.46, range
0.5 - 0.88) for the dementia patients. Healthy older adults achieved a mean digit span of 5.35
(SD 1.34). One of the dementia patients was not able to complete the first level of the test
(length 3). Mean digit span for the remaining patients was 4.375 (SD 0.92).
A logistic regression model predicting correctness of answers (correct/incorrect) from
group, age and digit span performance was fitted to the data, including random intercepts
for item and participant and a random slope of group in the item term. As expected,
dementia patients performed considerably worse in the offline comprehension task
compared to healthy older adults (z = 3.97, p < 0.001). Higher working memory abilities,
measured by the digit span task, contributed positively to performance in the
comprehension task (z = 2.05, p < 0.05). Age did not prove to be a significant predictor of
comprehension accuracy (z = -0.29, p = 0.8).
As predicted, participants in the control group reached higher comprehension accuracy in
the compatible condition (93.8 %) than in the incompatible condition (85.7 %), while
dementia patients reached accuracy rates of 68 % in both conditions. A logistic regression
model with random intercepts for participant and item confirmed a significant influence of
compatibility on accuracy for the control group (z = 2.153, p < 0.05), but not for the
dementia group.
Page 13 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
14
4.2 Reading times
As demonstrated in Figure 1, dementia patients show considerably lower reading times (on
average 1466 ms, SD 1400) and higher variance compared to healthy older adults (on
average 720 ms, SD 523) (Wilcoxon rank sum test: W = 2376500, p < 0.001). Figure 2 shows
the mean reading times per word for the initial part of the sentence, separately for each
participant group. Figure 3 highlights the correlation between the digit-span performance
and the mean log-transformed reading times (all items, including fillers; data for the first and
last words of each sentence were excluded due to substantial variation, in particular for the
dementia group). The correlation was significant for dementia patients (rpearson = -0.82, p <
0.05) but not for healthy older adults (rpearson= -0.43, p = 0.23).
In addition to differences in speed, there were – at least descriptively – differences in
reading times between compatible and incompatible sentences at the verb position in both
healthy older adults and dementia patients, and at the verb+2 position for dementia patients
only. In the following, we present separate analyses (see Table 2) for the three regions of
interest (verb, verb+1, verb +2).
4.2.1 Verb
The verb position analysis revealed a main effect of group (t = -3.701, p < 0.01), such that
dementia patients had significantly increased reading times than healthy older adults.
Compatibility also showed a main effect (t = -2.104, p < 0.05), indicating that verbs whose
gender associations do not match with the agent’s gender are read longer than compatible
ones. The interaction of group and compatibility was not significant.
Since both groups differ remarkably in their general pattern of reading times, additional
models were fitted to the respective subsets of the data, despite the insignificant
Page 14 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
15
interaction. Apart from the exclusion of the factor group, model specifications remained the
same. For healthy older adults, there was a marginal effect of compatibility (t = -1.875, p =
0.06), while no effect was found for the dementia group (t = -1.179, p = 0.25).
4.2.2 Verb+1
The analysis of the position verb+1 revealed no main effect of compatibility, nor an
interaction between group and compatibility. The main effect of group was significant (t = -
3.544, p < 0.01). Separate analyses of the groups did not reveal any effects of compatibility.
4.2.3 Verb+2
The analysis of the position verb+2 revealed main effects for both group (t = -2.924, p < 0.05)
and compatibility (t = -2.32, p < 0.05), and a marginally significant interaction between the
two factors, such that the effect of compatibility was more pronounced for the dementia
group (t = 2.836 p = 0.09). Separate group analyses revealed a main-effect of compatibility
for the dementia group (t = -2.619, p < 0.05), but not for the healthy older group (t = -0.478,
p = 0.63).2
5 Summary and Discussion
We examined whether stereotypical associations of verbs, with women or men as
prototypical agents, are integrated during reading in older adults with and without
dementia. Dementia patients showed considerably lower accuracy in comprehension
questions compared to healthy older adults, which is in line with reported problems in
sentence comprehension (e.g. Kempler et al. 1998; MacDonald et al. 2001; Peelle et al.
2008). Dementia patients also showed significantly longer average word reading times.
2 Within the dementia group, reading times tended to be higher in sentences with incorrectly answered
comprehension questions. Therefore, we reanalyzed reading time data using correctly answered trials only.
The general pattern of results with regard to the stereotype effect did not change.
Page 15 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
16
Interestingly, the effect of compatibility was reliable in reading latencies at the verb position
when computed across both groups. The effect was reliable one position earlier than in a
comparable study with younger adults (Müller-Feldmeth and Hanulíková, submitted).
Although we did not find an interaction of group and compatibility, large differences in the
reading latencies and high variability motivated separate group analyses. These analyses
showed a marginally significant compatibility effect on the verb position in the healthy older
adults only. The dementia group showed a numerical tendency in the same direction, but
the effect was not reliable. This may be due to considerably more individual variability in
reading times as well as the moderate sample size compared to older adults.
No main effect of compatibility was found at the position following the verb (verb+1), which
we expected to be relevant based on our previous findings with students (Müller-Feldmeth
and Hanulíková, submitted). Given that students were considerably faster (519 ms, SD 259),
they might have continued reading at a faster pace, without having completed processing or
integrating the information of previously read materials, leading to spillover effects. In
contrast, older adults performed slower, and possibly had more time to readily process a
word before the next word appeared, thus showing the compatibility effect on the verb
itself. This is in line with studies that find age-related changes in so-called ‘wrap-up’ effects –
peaks in reading times that occur at clause and sentence boundaries and are generally taken
as indicators of semantic integration of preceding materials. Younger readers tend to show
more sentence-final wrap-up, while older readers show more wrap-up on intrasentential
boundaries, indicating that the time course of temporary storage of information and
allocating cognitive resources for integration changes with age (Stine 1990; Payne and Stine-
Morrow 2012).
Page 16 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
17
At the noun completing the verb phrase (verb+2), we also found a significant effect of
compatibility across both groups, but also a marginal significant interaction of group and
compatibility. Separate analyses revealed a significant compatibility effect at the position
verb+2, but only for dementia patients, suggesting that the effect of compatibility is delayed
for dementia patients compared to healthy older adults. This may indicate a dementia-
related slow-down in retrieving extra-linguistic knowledge, leading to a delay in the impact
of probabilistic world-knowledge information. The observed delay also points towards a
problematic aspect of the word-by-word presentation used in the current study. While it
allows examining different effects in separate regions, individual differences in reading
strategies and speed may mask these effects. Some readers may display a different time
course of processing and integrating information, which are linked to both age and specific
cognitive impairments (e.g. Payne and Stine-Morrow 2016). In addition, since the self-paced
reading paradigm measures reading time based on button presses that include motor
reaction times, we cannot easily determine whether interindividual and between group
differences reflect general cognitive functioning or language related linguistic impairment,
such as slowed-down lexical retrieval, or/and motor-related differences (Ebaid et al. 2017).
Dementia patients showed lower comprehension questions accuracy than healthy older
adults and only healthy older adults showed a main effect of stereotype compatibility in
accuracy. This might suggest that the effort of integrating an agent with an incompatible or
unexpected verb requires more attention and thus distracts from other aspects of sentence-
meaning construction. Due to the word-by-word presentation, participants were not able to
re-read preceding words, posing an additional burden on processing, in particular when
memory skills are affected. Correlation analyses confirmed that working memory skills
affected reading times, most prominently in the group of dementia patients. Additional
Page 17 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
18
variation within the dementia group may stem from different types or stages of dementia
(Kempler and Goral 2008). Since more specific information about the dementia type was not
available, we were not able to relate interindividual variation to these factors. For further
research, however, differentiation of dementia type and stage could provide a more
complete picture of availability of stereotypical associations, and could be traced to specific
impairment patterns including semantic memory, integration skills, and also executive
functioning such as working memory and inhibitory skills.
Variation in performance can also result from potential gender differences. Some studies
(e.g. Siyanova-Chanturia et al. 2015) have found evidence that gender stereotype effects are
stronger for male participants. A valid analysis of gender effects was not possible in the
present study, but we would expect that a more balanced sample might lead to a more
pronounced effect of compatibility. Future studies may also consider cohort effects as an
additional source of processing differences that are not necessarily mediated by the
individual process of ageing, albeit by changes in the social environment that affect
probabilistic and stereotypical knowledge. For example, the verb reiten ‘to ride’ used to be a
stereotypically male activity (e.g. soldiers, farm workers, or policemen) and may be more
likely considered as such by older compared to younger people. A subset of the older
participants (n = 7) rated the verbs. While their overall ratings were comparable with the
gender associations found in our rating study, the verb reiten ‘to ride’ indeed was rated as
female by two participants only, while the majority rated it as neuter (4) or male (2).
Taken together, we were able to show that gender associations of verbs influence processing
in older adults with and without dementia, indicating that this type of knowledge is available
and active until old age (in line with Radvansky et al. 2010; Siyanova-Chanturia et al. 2015),
even under conditions of impaired semantic processes. Although the effect was present in
Page 18 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
19
RTs only in the dementia group, it suggests that gender stereotypes may be preserved and
activated, at least in mild to moderate stages of dementia. Patients’ significantly lower
comprehension accuracy and higher reading times suggest that language comprehension
may be affected by changes in semantic abilities (Kempler and Goral 2008). However, given
that gender stereotypical associations are activated, provides further evidence for the claim
that stereotype knowledge as an aspect of social knowledge is represented differently from
other semantic aspects (Contreras et al. 2012; Molinaro et al. 2016), since it seems to be
emotionally more relevant than knowledge about non-social entities (Norris et al. 2004).
The current study shows that investigating age effects in the time course of world-
knowledge integration, using self-paced reading, may constitute a valuable source of
understanding the dynamics of extralinguistic information use in language comprehension.
Due to large variation in clinical populations, more conclusive results would require
substantially larger samples and supporting data from other methods that do not require
word-by-word presentation. A larger sample may allow a more in-depth examination of the
self-paced reading paradigm as an instrument for early diagnosis in dementia.
References
Barr, Dale J., Roger Levy, Christoph Scheepers & Harry J. Tily. 2013. Random effects structure
for confirmatory hypothesis testing: Keep it maximal. Journal of Memory and Language
68(3). 255–278.
Bates, Douglas, Martin Mächler, Ben Bolker & Steve Walker. 2015a. Fitting linear mixed-
effects models using lme4. Journal of Statistical Software 67. 1–48.
Page 19 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
20
Bates, Douglas, Robert Kliegl, Shravan Vasishth & Harald R. Baayen. 2015b. Parsimonious
mixed models. Available from arXiv:1506.04967 (stat.ME).
Burke, Deborah M. & Meredith A. Shafto. 2008. Language and aging. In Fergus I. M. Craik &
Timothy A. Salthouse (eds.), The Handbook of Aging and Cognition (3), 373–443. New-
York: Psychology Press.
Carreiras, Manual, Alan Garnham, Jane Oakhill & Kate Cain. 1996. The use of stereotypical
gender information in constructing a mental model: Evidence from English and Spanish.
The Quarterly Journal of Experimental Psychology Section A 49(3). 639–663.
Clark-Cotton, Manuella R., R. K. Williams, Mira Goral & Loraine K. Obler. 2007. Language and
communication in aging. In James E. Birren (ed.), Encyclopedia of gerontology: Age,
aging, and the aged (2nd ed.), 1–8. London: Elsevier.
Contreras, Juan Manuel, Mahzarin R. Banaji & Jason P. Mitchell. 2012. Dissociable neural
correlates of stereotypes and other forms of semantic knowledge. Social Cognitive and
Affective Neuroscience (7). 764–770.
Duffy, Susan A. & Jessica A. Keir. 2004. Violating stereotypes: Eye movements and
comprehension processes when text conflicts with world knowledge. Memory &
Cognition 32(4). 551–559.
Ebaid, Deena, Sheila G. Crewther, Kirsty MacCalman, Alyse Brown & Daniel P. Crewther.
2017. Cognitive Processing Speed across the Lifespan: Beyond the Influence of Motor
Speed. Frontiers in Aging Neuroscience 9. 62.
Page 20 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
21
Esaulova, Yulia, Chiara Reali & Lisa von Stockhausen. 2014. Influences of grammatical and
stereotypical gender during reading: eye movements in pronominal and noun phrase
anaphor resolution. Language, Cognition and Neuroscience 29(7). 781–803.
Federmeier, Kara D., Marta Kutas & Rina Schul. 2010. Age-related and individual differences
in the use of prediction during language comprehension. Brain and Language 115(3).
149–161.
Garnham, Alan, Jane Oakhill & David Reynolds. 2002. Are inferences from stereotyped role
names to characters’ gender made elaboratively? Memory & Cognition 30(3). 439–446.
Kemper, Susan & Joan McDowd. 2006. Eye movements of Young and Older Adults while
Reading with Distraction. Psychology and Aging 21(1). 32–39.
Kempler, Daniel, Amit Almor, Lorraine K. Tyler, Elaine S. Andersen & Maryellen C.
MacDonald. 1998. Sentence Comprehension Deficits in Alzheimer’s Disease: A
Comparison of Off-Line vs. On-Line Sentence Processing. Brain and Language 64(3).
297–316.
Kempler, Daniel & Mira Goral. 2008. Language and dementia: Neuropsychological aspects.
Annual review of applied linguistics 28. 73–90.
Kennison, Shelia M. & Jessie L. Trofe. 2003. Comprehending pronouns: a role for word-
specific gender stereotype information. Journal of psycholinguistic research 32(3). 355–
78.
Kreiner, Hamutal, Patrick Sturt & Simon Garrod. 2008. Processing definitional and
stereotypical gender in reference resolution: Evidence from eye-movements. Journal of
Memory and Language 58(2). 239–261.
Page 21 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
22
Linger [Computer software]. Linger - a flexible platform for language processing experiments.
Retrieved from http://tedlab.mit.edu/~dr/Linger/.
Luke, Steven G. 2017. Evaluating significance in linear mixed-effects models in R. Behavior
Research Methods 49(4). 1494–1502.
MacDonald, Maryellen C., Amit Almor, Victor W. Henderson, Daniel Kempler & Elaine S.
Andersen. 2001. Assessing working memory and language comprehension in
Alzheimer’s disease. Brain and Language 78(1). 17–42.
McKoon, Gail & Roger Ratcliff. 1992. Inference during reading. Psychological review 99(3).
440–466.
Misersky, Julia, Pascal M. Gygax, Paolo Canal, Ute Gabriel, Alan Garnham, Friederike Braun,
Tania Chiarini, Kjellrun Englund, Adriana Hanulíková, Anton Öttl, Jana Valdrova, Lisa Von
Stockhausen & Sabine Sczesny. 2014. Norms on the gender perception of role nouns in
Czech, English, French, German, Italian, Norwegian, and Slovak. Behavior Research
Methods 46(3). 841–871.
Mitchell, D. C. 1984. An evaluation of subject-paced reading tasks and other methods for
investigating immediate processes in reading. New methods in reading comprehension
research. 69–89.
Molinaro, Nicola, Jui-Ju Su & Manuel Carreiras. 2016. Stereotypes override grammar: Social
knowledge in sentence comprehension. Brain and Language 155–156. 36–43.
Müller-Feldmeth, Daniel & Adriana Hanulíková (submitted). The craftsman knits a sweater:
Activating gender stereotypes during the processing of verbs.
Page 22 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
23
Norris, Catherine J., E. Elinor Chen, David C. Zhu, Steven L. Small & John T. Cacioppo. 2004.
The interaction of social and emotional processes in the brain. Journal of Cognitive
Neuroscience 16(10). 1818–1829.
Oakhill, Jane, Alan Garnham & David Reynolds. 2005. Immediate activation of stereotypical
gender information. Memory & Cognition 33(6). 972–983.
Park, Denise C., Gary Lautenschlager, Trey Hedden, Natalie S. Davidson, Anderson D. Smith &
Pamela K. Smith. 2002. Models of visuospatial and verbal memory across the adult life
span. Psychology and Aging 17(2). 299–320.
Park, Denise & Angela Gutchess. 2006. The cognitive neuroscience of aging and culture.
Current Directions in Psychological Science 15(3). 105–108.
Payne, Brennan R. & Elizabeth A. L. Stine–Morrow. 2012. Aging, parafoveal preview, and
semantic integration in sentence processing: Testing the cognitive workload of wrap-up.
Psychology and Aging 27(3). 638–649.
Payne, Brennan R. & Elizabeth A. L. Stine-Morrow. 2016. Risk for mild cognitive impairment
is associated with semantic integration deficits in sentence processing and memory. The
Journals of Gerontology Series B: Psychological Sciences and Social Sciences 71(2). 243–
253.
Peelle, Jonathan E., Vanessa Troiani, James Gee, Peachie Moore, Corey McMillan, Luisa
Vesely & Murray Grossman. 2008. Sentence comprehension and voxel-based
morphometry in progressive nonfluent aphasia, semantic dementia, and nonaphasic
frontotemporal dementia. Journal of Neurolinguistics 21(5). 418–432.
Page 23 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
24
R Core Team. 2018. R: A language and environment for statistical computing. R Foundation
for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.
Radvansky, Gabriel A., Nicholas A. Lynchard & William von Hippel. 2009. Aging and
Stereotype Suppression. Aging, Neuropsychology, and Cognition 16(1). 22–32.
Radvansky, Gabriel A., David E. Copeland & William von Hippel. 2010. Stereotype activation,
inhibition, and aging. Journal of Experimental Social Psychology 46(1). 51–60.
Reali, Chiara, Yulia Esaulova & Lisa von Stockhausen. 2015. Isolating stereotypical gender in a
grammatical gender language: Evidence from eye movements. Applied Psycholinguistics
36(4). 977–1006.
Reilly, Jamie & Jonathan E. Peelle. 2008. Effects of semantic impairment on language
processing in semantic dementia. Seminar in Speech and Language 29(01). 32–43.
Singmann, Henrik, Ben Bolker, Jake Westfall & Frederik Aust. 2018. afex: Analysis of factorial
experiments. R package, version 0.19-1.
Siyanova-Chanturia, Anna, Francesca Pesciarelli & Cristina Cacciari. 2012. The
electrophysiological underpinnings of processing gender stereotypes in language. PLOS
ONE 7(12). e48712.
Siyanova-Chanturia, Anna, Paul Warren, Francesca Pesciarelli & Cristina Cacciari. 2015.
Gender stereotypes across the ages: On-line processing in school-age children, young
and older adults. Frontiers in Psychology 6. 1388.
Stine, Elizabeth L. 1990. On-line processing of written text by younger and older adults.
Psychology and Aging 5(1). 68–78.
Page 24 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
25
Vasishth, Shravan. 2006. On the proper treatment of spillover in real-time reading studies:
Consequences for psycholinguistic theories. In: Proceedings of the International
Conference on Linguistic Evidence.
von Hippel, William, Lisa A. Silver & Molly E. Lynch. 2000. Stereotyping against your will: The
role of inhibitory ability in stereotyping and prejudice among the elderly. Personality
and Social Psychology Bulletin 26(5). 523–532.
White, Katherine R. G., Stephen L. Crites, Jennifer H. Taylor & Guadalupe Corral. 2009. Wait,
what? Assessing stereotype incongruities using the N400 ERP component. Social
Cognitive and Affective Neuroscience 4(2). 191–198.
White, Katherine R. G., Rose H. Danek, David R. Herring, Jennifer H. Taylor & Stephen L.
Crites. 2018. Taking Priming to Task. Social Psychology 49(1). 29–46.
Wingfield, Arthur & Elizabeth A. L. Stine-Morrow. 2000. Language and speech. In Fergus I. M.
Craik, Timothy A. Salthouse (eds.), Handbook of Aging and Cognition, (2nd ed.), 359–
416. Mahwah, N.J.: Erlbaum Associates.
Page 25 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
Figure 1. Boxplots of log-transformed word reading times, separated by group. Left hand
side: All data points, including filler sentences. Right hand side: data points in the three
regions of interest (verb, verb+1, verb +2) after the outlier correction.
Page 26 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
27
Figure 2. Reading times (ms) on the three regions of interest (verb, verb+1, and verb+2) as
well as the word before (agent) and after (verb+3) the regions of interest, averaged over
participants and items, split by group and compatibility. Outliers are excluded. Red lines
display incompatible items, blue lines compatible ones. Dementia patients are represented
by the two lines on top, healthy older adults by the bottom lines. Whiskers indicate standard
errors.
Page 27 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
28
Figure 3. Scatterplot of digit span score (maximal digits) and mean log-transformed word
reading time (av.word.log.rt; all items, including fillers; data for the first and last words of
each sentence are excluded). Each dot and triangle represents one participant. The red, blue
and grey lines represent regression lines.
Page 28 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
For Review Only
Running title: Gender stereotypes in dementia and older adults
verb
verb+1
verb+2
Est.
SE
t
p
Est.
SE
t
p
Est.
SE
t
p
(Intercept)
4.455
0.284
15.673
4.675
0.225
20.753
4.714
0.323
14.599
compatibility
-0.031
0.015
-2.104
0.038
0.003
0.012
0.255
0.802
-0.034
0.015
-2.32
0.022
group
-0.19
0.051
-3.701
0.002
-0.143
0.04
-3.544
0.002
-0.152
0.052
-2.924
0.01
log.rt.min.1
0.309
0.041
7.562
<0.001
0.281
0.033
8.645
<0.001
0.267
0.045
5.969
<0.001
word length
0.022
0.011
1.907
0.091
0.001
0.016
0.075
0.942
0.035
0.023
1.498
0.155
compatibility X
group
0.004
0.015
0.266
0.792
0.002
0.012
0.128
0.9
0.025
0.015
1.693
0.093
Table 2. Estimated coefficients (Est.), standard errors (SE), t-values (t) and p-values (p) for the linear mixed-effect model fitted to the self-paced reading
latencies in three regions of interest (verb, verb+1, verb+2).
Page 29 of 29
Linguistics Vanguard
Linguistics Vanguard
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
