Striking Loss of Second Language in Bilingual Patients With Semantic Dementia (Journal of Neurology) (2019)
Striking Loss of Second Language in Bilingual Patients With Semantic Dementia (Journal of Neurology) (2019)
Striking Loss of Second Language in Bilingual Patients With Semantic Dementia (Journal of Neurology) (2019)
https://doi.org/10.1007/s00415-019-09616-2
ORIGINAL COMMUNICATION
Abstract
Background Studies of bilingual or multilingual patients with neurodegenerative diseases that disrupt language like the
primary progressive aphasias (PPA) may contribute valuable information on language organization in the bilingual brain and
on the factors affecting language decline. There is limited literature on bilingual PPA and in particular on semantic dementia,
a type of PPA with selective loss of semantic memory. We studied the nature and severity of naming and comprehension
deficits across languages in bilingual patients with semantic dementia (SD).
Methods Sixteen bilingual patients with SD and 34 bilingual age-matched controls were administered the modified Boston
Naming Test and components of Cambridge Semantic Battery. The patients’ performance on picture naming and word
comprehension was compared across languages and with controls. The most proficient language on self-rating was labelled
as L1 and less proficient as L2.
Results We observed striking loss of second language (L2) in SD for both receptive and expressive language, even in patients
who were premorbidly fluent in their L2. Naming and comprehension in every patient’s L2 were impaired relative to both
their own first-language (L1) scores and controls’ L2 scores. Furthermore, item-specific correct responses in each patient’s
L2 were a subset of their successes in L1.
Discussion A striking contrast in performance between two languages in bilingual patients with SD indicates that a bilin-
gual’s L2 or less proficient language is more vulnerable to neurodegeneration. Our findings also support a common semantic
network in the brain for the different languages of bilinguals.
Keywords Primary progressive aphasia · Semantic dementia · Bilingualism · Naming · Word comprehension
Background
13
Vol.:(0123456789)
Journal of Neurology
knowledge [7–10]. The three PPA variants appear to be results may inform one of the central questions in this
important lesion models for understanding semantic mem- research topic: to what extent are the processes underlying
ory, speech fluency and syntax and phonological working a language skill such as object naming or word comprehen-
memory. An important question to be answered is how neu- sion shared or independent across the different languages of
rodegeneration affects these different linguistic capacities in a bilingual individual?
each language spoken by a bi- or multilingual patient. Con-
trasting performance in two languages might be informative
in the evaluation of PPA, as well as of theoretical interest for Methods
cognitive neuroscience and linguistics.
In spite of increasingly common presentations of bilin- Patients
gual aphasia in the clinic, there are a limited number of
research studies on PPA in bilinguals [11–21]. One describes Patients presenting with PPA attending the Cognitive Neu-
two multilingual patients with SD, with relative preservation rology Clinic, Manipal Hospital, Bangalore, India over a
of first language and substantial loss of the later-learned lan- 5-year period were identified. Of the 20 patients diagnosed
guages [11]. In both patients, words comprehended in one with SD, 4 were excluded because 2 were monolinguals and
language were not consistently named or comprehended in 2 were too impaired to complete the tests. A total of 16
the other languages they knew, leading the authors to pro- patients, 6 women and 10 men, formed the sample of the
pose separate lexical-semantic systems for the different study (Table 1). The mean age at diagnosis was 65.5 (8.2)
languages. Another brief report mentions loss of both first years and at reported onset was 62.9 (8.8) years. All patients
and second languages in a multilingual Taiwanese woman or their caregivers gave informed consent to participate in
with SD [12]. In the largest study on clinical features, soci- the study, according to the Declaration of Helsinki, which
odemographics and biomarkers of bilingual PPA published was approved by the hospital ethics committee.
to date, Costa et al. [21] combined 13 cases culled from a The diagnosis of SD was based on criteria for semantic
literature search with 20 new cases. Formal language assess- variant PPA [6]. All the patients were living at home with
ment was done in only 45% of patients. In addition to the family, and detailed history was taken from two reliable
above-mentioned papers, they added two more SD patients. caregivers (spouse and a son or daughter) as well as the
Despite the expected heterogeneity both between and within patient. A comprehensive neurological examination, neu-
subtypes, the authors reported largely parallel patterns of ropsychological testing and neuroimaging were performed.
impairment in the patients’ two languages across all sam- Magnetic resonance imaging (MRI) scans at presentation
ples. The study had several limitations as pointed out by the showed moderate atrophy of bilateral temporal lobes, espe-
authors, including its retrospective nature, limited sample cially anterior and medial, more prominent on the left side
size, heterogeneity of assessment tools and incomplete lin- in 13 and on the right in 3 patients.
guistic data. A recent review chapter attempted to analyse
the published cases of bilingual PPA and concluded that Language details
the language decline was either (a) parallel in L1 and L2 or
(b) differential with L2 more impaired [22]. There was no A questionnaire was used to collect information from the
patient where L1 was more affected. The authors concluded patient and the caregivers regarding languages spoken, read
that the order of acquisition of languages was a much more and written, age of acquisition, proficiency and medium of
important factor than either proficiency in or age of acquisi- education. Premorbid proficiency for speaking was assessed
tion of a patient’s two languages. on a scale of 0–7 (7 being most proficient) using self-ratings.
India is an ideal setting to study bilingual aphasia as it Seven patients could do a self-rating but nine patients had
is home to 234 dialects and 122 languages, of which 22 are poor comprehension, so the caregivers’ ratings were used in
officially recognized [23]. Around 25% of the population these cases. The most proficient language for speaking was
is bilingual, which amounts to roughly 250 million people. referred to as L1, which was the same as the mother tongue
Currently, we are investigating language profiles of bilin- in 14/16 patients. If two languages were rated equal, the
gual patients with mild cognitive impairment, Alzheimer’s preferred language was considered as L1. Six patients were
disease, PPA and frontotemporal dementia in our memory bilinguals and 10 multilinguals, three of whom spoke four
clinic [24]. Here, we report our first set of findings in bilin- languages premorbidly (Table 2). Thirteen patients could
gual patients with SD. The aim of the study was to examine read and write in L2 as well. Seven patients received their
the nature and severity of impairment across different lan- bilingual exposure before 5 years (early bilinguals). Most
guages in bilingual (multilingual) SD patients. The other frequent languages were Kannada as L1 in seven and Eng-
objective was to compare performance across each patient’s lish as L2 in eight patients. Two patients spoke dialects (L1)
languages at the item-specific level on the same tests. Such Bhojpuri and Tulu, which have no written form. English was
13
Journal of Neurology
Table 1 Patient characteristics and performance on the Addenbrooke’s Cognitive Examination-Revised (ACE-R) and the language subcompo-
nent of the ACE-R, with patients ordered from least to most impaired on total ACE-R
Patient Gender Education Occupation Age at diag- Reported age at Estimated dura- ACE-R (100) ACE-R-
(years) nosis (years) onset (years) tion (months) language
(26)
learnt formally in school in all the patients who could speak sisted of colour photographs of objects. The modified test
English (L2 or L3); it is also the language of instruction in was administered in a healthy elderly population.
college and university. Six patients had equal proficiency rat-
ings of 7 in both L1 and L2. Before the onset of symptoms, Cambridge Semantic Battery (CSB) Picture naming and
all the patients were using L2 every day, either socially (9) word-to-picture matching subcomponents of the Cam-
or at work (6) or both (1) and eight patients were still work- bridge Semantic Battery (CSB) [28] were administered
ing. For instance, patient 9 was teaching in English at an to evaluate semantic memory across different languages.
Engineering college at the onset of symptoms. Each test used the same 64 items, with 8 stimuli in each
of 8 categories: 4 groups of living things or natural kinds
(domestic animals, wild animals, birds, and fruits) and 4
of manmade objects or artefacts (large household items,
Assessment
small household items, tools and vehicles).
Naming test The patient is asked to name a line drawing
Cognitive assessment
of each of the 64 items. The stimuli were presented in a
random order. Maximum score is 64.
The Addenbrooke’s Cognitive Examination-Revised (ACE-
Word-to-picture matching Patients are presented with
R) [25] which has been adapted and widely used in an Indian
64 picture arrays, one for each item, with each array con-
population [26] was used as a global screen of cognitive
sisting of a target picture and 9 other pictures from the
function.
same category. They are asked to point to the target picture
named by the examiner. Maximum score is 64.
The language assessment was done over two different
Language assessment sessions. The tests were first administered in the patient’s
L1, followed by neuropsychological tests to reduce con-
Modified Boston Naming Test (mBNT) A 30-item nam- founding factors including test repetition. Then, the lan-
ing test similar to the Boston Naming Test (BNT) was guage tests in L2 and L3 were administered. The interval
developed from an Indian version of the Snodgrass and between the assessments was usually 3–7 days. Language
Vanderwart pictures [27] and their names, with high- and subcomponents and abilities were also assessed informally
low- frequency object names across three Indian languages as part of the cognitive examination by the neurologist.
(Kannada, Hindi and Tamil) and English. The pictures con-
13
13
Table 2 Language details of patients
Patient Number of Mother tongue L1—self-rating L2—self-rating L3 and L4—self- Read and write Reads newspapers Type of Proficiency-bal- Method of
languages for proficiency for proficiency rating for profi- bilingual anced or dominant acquisition
spoken (estimated AOA) (estimated AOA) ciency (estimated (AOA) (comments) of L2
AOA)
L1 most proficient language, L2 second most proficient language, L3 third most proficient language, L4 fourth most proficient language, AOA age of acquisition
Journal of Neurology
Journal of Neurology
Controls the two languages while speaking, both at home and during
testing.
34 bilingual healthy controls with a mean age of 65.5 (6.4)
years underwent a clinical interview, the ACE-R, mBNT Performance on tests
naming and the CSB naming and word comprehension tests.
The CSB and mBNT were administered in L1 and L2. Of Controls
the 34 controls, it was possible to match 10 to an individual
patient for age, sex, same L1 and L2 and even similar profi- As demonstrated in Supplementary Table 1 and Fig. 1a,
ciency ratings in L1 and L2. Details of language background b, the performance of the control participants on the two
and performance on tests are provided in Supplementary naming tests revealed a very small and not entirely consist-
Table 1. ent advantage for L1. In 80% of the control comparisons
Independent T tests were carried out comparing patients between L1 and L2 naming (the two naming tests com-
and the whole control group across age, education and lan- bined), naming scores for L1 vs. L2 were either equal or
guage proficiency ratings in L1 and L2. No significant dif- differed by only 1 or 2 points, with some actually a point
ferences were found between the groups on age (48, t = 0.46, or two higher in L2. This is a crucial finding: if language
p = 0.79, Hedges’ g = 0.02), education (48, t = 0.28, p = 0.78, performance was found to be significantly better in L1 than
Hedges’ g = 0.09), or language proficiency rating in L2 (48, L2 in healthy participants, such a finding in the SD patients
t = 1.2, p = 0.22, Hedges’ g = 0.38). Language proficiency could not be unambiguously interpreted as an effect of the
ratings in L1 were identical. brain disease.
The Wilcoxon Signed Rank Test comparing patients and
matched controls indicated no significant difference across Patients
age (Z = 1.83, p = 0.07), education (Z = 0.71, p = 0.48) or lan-
guage proficiency ratings in L2 (Z = 0.45, p = 0.66). Most patients performed poorly on the ACE-R, only four
scoring above 60/100 (Table 1), establishing that the group
has more advanced disease than the reported duration of
illness would suggest.
Results Table 3 and Fig. 1a, b present the number and percent
correct for the patients who managed to complete either
Qualitative observations the mBNT or the CSB naming tests, or both of these. With
the sole exception of patient 1 for CSB naming in his L1,
Relatives commonly reported the patients’ impoverished the naming scores were low, even in the patients’ L1s. Four
ability to converse in L2 or L3, in which they were previ- patients who were able to attempt these tests in their L3
ously fluent. All patients except two had reverted to use of had scores at floor, and these results will not be considered
L1 both at home and in social situations. Patients with Eng- further. Some of the patients would often respond in their L1
lish as L2 could no longer read the English newspaper. Eight even when the test was being administered in L2.
patients could speak a few words or phrases in L2 but speech In cases where naming performance was severely
was quite impoverished and they were unable to name most degraded in L1, it is obviously not meaningful to assess the
objects in L2 on bedside testing. Eight patients could not L1–L2 discrepancy. Consider, therefore, the 12 instances
understand even simple instructions in L2 and were unable where a patient’s naming score in L1 was ≥ 20% correct on
to speak a single sentence in the L2 in which they were pre- either the BNT or the CSB (Table 3). The average naming
viously fluent. In the examiner’s assessment, of the patients score for L1 in these 12 instances was 41%. The correspond-
who had spoken three or more languages premorbidly, five ing average for the same patients on the same tests in L2 was
demonstrated a similar extent of impaired ability to under- 13%. This is a striking result, because all these patients were
stand spoken L2 and L3, while in five patients, L2 was better proficient bilinguals premorbidly (7/9 had 6 or 7 proficiency
than L3 (patients 3, 5, 8, 9 and 12). ratings for L2). Furthermore, eight of them could previously
When spoken to in Kannada (L2), patient 16 responded, read and write L2 and one could read and write only in L2.
“I don’t understand English”. She was unable to differentiate The same striking L1 > L2 percent correct contrast was
between the languages when asked to say which language observed in the test of single-word comprehension (forced-
she herself was using; in fact, failure to differentiate between choice word-to-picture matching): for the seven patients with
the languages was striking in five patients (patients 8, 11, performance ≥ 20% in their L1, the average percentage cor-
12, 14 and 16). Patient 12 frequently mixed Hindi and Eng- rect in L1 = 60% and in L2 = 17%. This is a very important
lish while speaking. Patient 8 tended to use both Kannada result to highlight, because it establishes that the L1 > L2
(preferred language) and Tamil (mother tongue) but mixed discrepancy is not confined to expressive language but also
13
Journal of Neurology
Percent Correct
matched controls (MC) across 60
most proficient language (L1)
and second most proficient 50
language (L2) on the modified 40
Boston Naming Test (mBNT). 30
b Naming (percent correct) per-
formance in individual patients, 20
semantic dementia (SD) patient 10
group, whole group controls 0
(WGC) and matched controls Pt 8 Pt 11 Pt 12
SD MC WGC
Pt 2 Pt 3 Pt 5 Pt 9 Pt 15
(MC) across most proficient Mean
language (L1) and second most
proficient language (L2) on the L1 L2
Cambridge Semantic Battery B 100
(CSB). Pt patient 90
80
70
Percent Correct
60
50
40
30
20
10
0
Pt 1 Pt 3 Pt 4 Pt 5 Pt 6 Pt 8 Pt 9 Pt 10 Pt 12 SD MC WGC
Mean
applies to receptive language/comprehension. Patients like (including surface dyslexia in English). Although sponta-
these who are towards the more impaired end of SD progres- neous writing was not possible in all the patients, they did
sion are, predictably, profoundly anomic even in their L1. write words and sentences to dictation but with errors (mis-
In some sense, therefore, the large L1 > L2 discrepancy on spellings, missing words).
word-picture matching, which is a less challenging receptive
test of language, is even more informative and striking than Whole‑group statistical analyses
the difference in naming.
On the comparison of item-specific naming performance Independent T tests compared performance of patients and
across languages, there was only one instance of a correct controls across L1 and L2 on mBNT, CSB naming and CSB
response in L2 when that patient did not correctly name word-to-picture matching (Table 4). The patients’ perfor-
the corresponding item in L1. Patient 9 named crocodile mance was significantly lower than controls across L1 (with
correctly in English but not in Oriya (L1). That is, with the medium effect sizes) and L2 on mBNT, CSB naming and
exception of one item for one patient, L2 correct naming CSB word-picture matching (with medium to large effect
responses were a proper subset of L1 correct responses. sizes).
At the level of item-specific responses in word-to-picture
matching, when patients’ levels of success in L1 were above Matched control statistical analyses
chance (the only situation in which it makes sense to do this
comparison), no patient ever chose the correct picture for the Wilcoxon Signed Rank Tests indicated that the median test
spoken name in L2 without a corresponding correct response ranks across BNT naming, CSB naming and CSB word-
to the same item in L1; in other words, L2 correct choices to-picture matching were significantly different between
were a fully proper subset of L1 correct choices. patients and their matched controls for L1 and L2 (Table 4).
Patients were tested informally on reading and writing. Each participant’s performance across L1 and L2 was also
Eight of the sixteen patients still retained some ability to analysed separately for SD patients and the control group.
read and write in L2, but reading errors were prominent Wilcoxon Signed Rank Tests indicated that the median test
13
Journal of Neurology
Table 3 Performance of patients in most proficient language (L1) versus second most proficient language (L2) on naming (both modified Boston
Naming Test—mBNT and Cambridge Semantic battery—CSB) and word-to-picture matching—CSB: number (and percent) correct
Patient Naming (mBNT)—N = 30 Naming (CSB)—N = 64 Word-to-picture matching (CSB)—
Correct responses (percentage) Correct responses (percentage) N = 64
Correct responses (percentage)
L1 L2 L1 L2 L1 L2
NA not administered
Table 4 Comparison of performance of semantic dementia (SD) matching—CSB in most proficient language (L1) and second most
patients and controls on naming (both modified Boston naming test— proficient language (L2)
mBNT and Cambridge semantic battery-CSB) and word-picture
Matched controls Whole group controls
Mean (s.d.) Median WSRT p value Mean (s.d.) t p value ES
Z
13
Journal of Neurology
ranks across mBNT naming (Z = 2.53, p = 0.01), CSB match- behavioural variant of frontotemporal dementia, all of which
ing (Z = 2.53, p = 0.01), CSB naming (Z = 2.67, p = 0.01) we are currently investigating.
were significantly lower in L2 as compared to L1 in the SD Evidence from a number of recent studies suggests that
patient group. In contrast, the median test ranks between L1 semantic knowledge/processing for the different languages
and L2 on mBNT naming (Z = 1.45, p = 0.15), CSB match- of multilingual individuals is subserved by a common net-
ing (Z = 1.71, p = 0.09), CSB naming (Z = 1.59, p = 0.11) work in the brain [29, 30]. In healthy bilingual participants,
showed no significant differences in the control group. the results of several functional imaging studies strongly
support the view (a) that semantic representations are shared
between an individual’s L1 and L2, and (b) that the most
Discussion significant shared neural activation is observed in the left
anterior temporal lobe which is, notably, the most consistent
This study of word production and comprehension across area of atrophy in SD [31, 32]. This same region was also the
different languages in a fairly large case series of bilingual source of differential functional MRI activation for semantic
patients with SD yielded results that address two major versus phonemic fluency in both languages of a large set of
issues. First, despite a large literature on the cognitive and healthy bilinguals [33].
clinical features of SD, very little has previously been docu- The item-wise correspondence of results between L1 and
mented on the fate of language abilities in the different lan- L2 in both naming and comprehension in our study seems
guages of bilingual individuals with SD. Second is a more fully compatible with the proposal that, when a bilingual
general question about the neural networks for word compre- person sees an object or picture of it, there is only one con-
hension and production in bilingual individuals. ceptual representation for that object which sends activation
Although all patients were impaired in both production to the name in either L1 or L2 (or both). Our interpretation is
(picture naming) and comprehension (spoken word-to- that, as the semantic system deteriorates in SD, the stronger
picture matching) in their L1, their corresponding levels of premorbid connections between L1 language representa-
success in L2 (and L3 where available) were dramatically tions (for both input and output) and language-independent
worse. The performance was significantly different from semantic representations will be more resistant than L2 to
controls especially for L2. Even at presentation, many of this deterioration [34]. Of course our data only support this
the patients had essentially lost the ability to converse in L2 claim for the concrete concepts studied here (objects and
or L3. Strikingly, some even failed to identify the language animals). Abstract words—even within one language—tend
in which the examiner or they themselves were speaking. to have meanings that are more context-specific and may
Severely deteriorated performance in L2 and L3 was seen have slightly different neural representations in L1 vs. L2.
in early and late bilinguals, in the two illiterate and all the It is likely that the advantage for L1 over L2 has funda-
remaining literate patients, and in the equally proficient as mentally the same basis as the dramatic impact of stimu-
well as the less proficient bilinguals. In addition, the phe- lus frequency, familiarity and/or age of acquisition that is
nomenon was seen whether L2 was similar to L1 (e.g., already well documented in all aspects of performance in
Bhojpuri and Hindi are both Indo-Aryan languages; Telugu SD, with a massively significant advantage for the higher
and Kannada are both Dravidian languages) or distinctively frequency, more familiar and earlier learned aspects of expe-
different (Kannada and English). When it is remembered rience [35]. Many studies have demonstrated significant age
that, premorbidly, all of these individuals spoke their L2 of acquisition effects for naming in monolingual speakers
daily, could read and write in L2 (13/16) and read newspa- after brain damage [36–38], in healthy participants [39] and
pers only in L2 or L3 (10/16), these results are very notable in computational models [40, 41]. Other factors may also
indeed. Two patients who spoke dialect at home but were contribute, however, such as degree of proficiency in, usage
using L2 for speaking socially and for reading and writing of and exposure to [30, 42, 43], or even actual AOA of L2.
also showed the striking loss. A point worth emphasizing is In this regard, it is worth noting that the relatively spared
the ubiquitous loss or impairment of L2 in spite of the het- language or L1 in our study was usually but not quite always
erogeneity of the patient sample in terms of disease severity, the mother tongue or primary language, with two patients
years of education, variety of languages spoken (8 as L1 and (patients 8 and 12) being exceptions. For the remaining 14
5 as L2) and premorbid proficiency in L1/L2. patients, L1 was also their mother tongue or primary lan-
These prominent findings were easily demonstrable in guage. Thus, what our results demonstrate is that the most
the clinic while eliciting history or on naming of common used or most proficient and usually first-acquired language
objects in L2. Of course, our findings need to be replicated was relatively spared and the other languages were largely
across other bilingual populations from other countries lost.
but, in our experience, this phenomenon does not occur Limitations of our study include relatively low scores and
in patients with other PPAs, Alzheimer’s disease or the the inability of some patients to complete the tests in spite
13
Journal of Neurology
of the rather short duration of symptoms reported by their 2. Lorenzen B, Murray LL (2008) Bilingual aphasia: a theoretical
families. This may be in part due to cultural effects, where and clinical review. Am J Speech Lang Pathol 17:299–317. https
://doi.org/10.1044/1058-0360(2008/026)
even well-educated healthy people seem to do poorly on 3. Mesulam M (1982) Slowly progressive aphasia without general-
tests as compared to western populations [24]. A lack of ized dementia. Ann Neurol 11:592–598. https://doi.org/10.1002/
awareness of dementia in general, failure to recognize early ana.410110607
symptoms by the family and misdiagnosis by the primary 4. Mesulam MM (2001) Primary progressive aphasia. Ann Neurol
49:425–432. https://doi.org/10.1002/ana.91
physician may also have contributed to the apparent discrep- 5. Gorno-Tempini ML, Dronkers N, Rankin KP, Ogar JM, Phen-
ancy between duration and performance on tests. We made grasamy L, Rosen HJ et al (2004) Cognition and anatomy in
every effort to match controls to patients and succeeded three variants of primary progressive aphasia. Ann Neurol
in matching 10 with similar language characteristics. An 55:335–346. https://doi.org/10.1002/ana.10825
6. Gorno-Tempini ML, Hillis AE, Weintraub S, Kertesz A, Mendez
exactly matched control sample, though of course ideal, M, Cappa SF et al (2011) Classification of primary progressive
would be extremely difficult to identify. Furthermore, as in aphasia and its variants. Neurology 76:1006–1014. https://doi.
many clinical studies, it was not feasible to achieve an ideal org/10.1212/WNL.0b013e31821103e6
combination of extensive cognitive testing and sophisticated 7. Warrington EK (1975) Selective impairment of semantic mem-
ory. Q J Exp Psychol 27:635–657. https://doi.org/10.1080/14640
brain imaging in a large patient cohort. 747508400525
Future directions include a study of the interaction of fac- 8. Snowden JS, Goulding PJ, Neary D (1989) Semantic demen-
tors such as age of acquisition, degree of proficiency and tia: a form of circumscribed cerebral atrophy. Behav Neurol
usage of/exposure to L2 on language decline in SD as com- 2:167–182. https://doi.org/10.1136/jnnp.70.3.323
9. Hodges JR, Patterson K, Oxbury S, Funnell E (1992) Seman-
pared to the other varieties of PPA. tic dementia: progressive fluent aphasia with temporal lobe
In conclusion, we found a dramatic loss of less proficient atrophy. Brain 115:1783–1806. https://doi.org/10.1093/brain
and usually later-acquired languages (L2/L3) in our bilingual /115.6.1783
(and multilingual) patients with SD. Furthermore, any cor- 10. Hodges JR, Patterson K (2007) Semantic dementia: a unique clin-
icopathological syndrome. Lancet Neurol 6:1004–1014. https://
rect responses for both naming and word comprehension in doi.org/10.1016/S1474-4422(07)70266-1
L2 were virtually always matched by success on the same 11. Mendez MF (2004) Semantic dementia in multilingual patients.
items in L1. These results suggest (a) that less proficient J Neuropychiatry Clin Neurosci 16:381. https://doi.org/10.1176/
and usually later-learned languages are more vulnerable to appi.neuropsych.16.3.381
12. Liu YC, Yip PK, Fan YM, Meguro K (2012) A potential protective
neurodegeneration, and (b) that different languages connect effect in multilingual patients with semantic dementia: two case
to a common semantic network in the brain. reports of patients speaking Taiwanese and Japanese. Acta Neurol
Taiwanica 21:25–30
Acknowledgements The authors thank all the patients and their fami- 13. Filley CM, Ramsberger G, Menn L, Wu J, Reid BY, Reid AL
lies for participating in the study and colleagues for referrals. Some of (2006) Primaryprogressive aphasia in a bilingual woman. Neuro-
the findings from this study have been presented at the International case 12:296–299. https://doi.org/10.1080/13554790601126047
Neuropsychological Society (INS) meeting in London, 6–8 July 2016, 14. Hernandez M, Cano A, Costa A, Sebastian-Galles N, Juncadella
at the British Neuropsychological Society (BNS) meeting in London, M, Gascon-Bayarri J (2008) Grammatical category-specific
1 November 2018 and the International conference on Frontotemporal deficits in bilingual aphasia. Brain Lang 107:68–80. https://doi.
dementias in Sydney, 11–14 November 2018. org/10.1016/j.bandl.2008.01.006
15. Machado A, Rodrigues M, Simoes S, Santana I, Soares- Fer-
Funding This research did not receive any specific grant from funding nandes J (2010) The Portuguese who could no longer speak
agencies in the public, commercial or not-for-profit sectors. French: primary progressive aphasia in a bilingual man. J Neu-
ropsychiatry Clin Neurosci 22:E31–E32. https://doi.org/10.1176/
jnp.2010.22.1.123.e31
Compliance with ethical standards 16. Zanini S, Angeli V, Tavano A (2011) Primary progressive apha-
sia in a bilingual speaker: a single-case study. Clin Linguist Phon
Conflicts of interest On behalf of all the authors, the corresponding 25:553–564. https://doi.org/10.3109/02699206.2011.566464
author states that there is no conflict of interest. 17. Kambanaros M, Grohmann KK (2011) BATting multilingual
primary progressive aphasia for Greek, English, and Czech. J
Ethical standard The study has been approved by the hospital ethics Neurolinguist 25:520–537. https://doi.org/10.1016/j.jneuroling
committee. .2011.01.006
18. Larner AJ (2012) Progressive non-fluent aphasia in a bilingual
Informed consent All patients or their caregivers and controls gave subject: Relative preservation of “Mother Tongue”. J Neuropsy-
informed consent according to the Declaration of Helsinki. chiatry Clin Neurosci 24:E9–E10. https://doi.org/10.1176/appi.
neuropsych.11010019
19. Druks J, Weekes BS (2013) Parallel deterioration to language pro-
cessing in a bilingual speaker. Cogn Neuropsychol 30:578–596.
https://doi.org/10.1080/02643294.2014.882814
References 20. Karpathiou N, Papatriantafyllou J, Kambanaros M (2018) Bilin-
gualism in a case of the non-fluent/agrammatic variant of pri-
1. Grosjean F (2010) Bilingual: life and reality. Harvard University mary progressive aphasia. FrontCommun 3:1–16. https://doi.
Press, Cambridge org/10.3389/fcomm.2018.00052
13
Journal of Neurology
21. Costa AS, Jokel R, Villarejo A, Llamas-Velasco S, Domoto-Reil- representations in anterior temporal lobe. J Neurosci 34:332–338.
ley K, Wojtala J, Reetz K, Machado Á (2019) Bilingualism in https://doi.org/10.1523/JNEUROSCI.1302-13.2014
primary progressive aphasia: a retrospective study on clinical and 33. Grogan A, Green DW, Ali N, Crinion JT, Price CJ (2009) Struc-
language characteristics. Alzheimer Dis Assoc Disord 33:47–53. tural correlates of semantic and phonemic fluency ability in first
https://doi.org/10.1097/WAD.0000000000000288 and second languages. Cereb Cortex 19:2690–2698. https://doi.
22. Malcolm T, Lerman A, Korytkowska M, Vonk JMJ, Obler LK org/10.1093/cercor/bhp023
(2019) Primary progressive aphasia in bilinguals and multilin- 34. Kroll JF, Stewart E (1994) Category interference in translation
guals. In: Schwieter (ed) The handbook of neuroscience of mul- and picture naming: evidence for asymmetric connection between
tilingualism. Wiley, Hoboken, pp 572–591 bilingual memory representations. J Mem Lang 33:149–174. https
23. Census India (2001). http://www.census india .gov.in. Accessed 31 ://doi.org/10.1006/jmla.1994.1008
May 2019 35. Patterson K, Lambon Ralph MA, Jefferies E, Woollams A, Jones
24. Ramanan S, Narayanan J, D’Souza TP, Malik KS, Ratnavalli E R, Hodges JR et al (2006) “Presemantic” cognition in semantic
(2015) Total output and switching in category fluency successfully dementia: six deficits in search of an explanation. J Cogn Neurosci
discriminates Alzheimer’s disease from Mild Cognitive Impair- 18:169–183. https://doi.org/10.1162/089892906775783714
ment, but not from frontotemporal dementia. Dement Neuropsy- 36. Ellis AW, Lum C, Lambon Ralph MA (1996) On the use of
chol 9:251–257. https: //doi.org/10.1590/1980-576420 15DN9 3000 regression techniques for the analysis of single case aphasic data.
007 J Neurolinguistics 9:165–174. https://doi.org/10.1016/0911-
25. Mioshi E, Dawson K, Mitchell J, Arnold R, Hodges JR (2006) 6044(96),00008-5
The Addenbrooke’s Cognitive Examination Revised (ACE-R): a 37. Lambon Ralph MA, Graham KS, Ellis AW, Hodges JR (1998)
brief cognitive test battery for dementia screening. Int J Geriatr Naming in semantic dementia–what matters? Neuropsychologia
Psychiatry 21:1078–1085. https://doi.org/10.1002/gps.1610 36:775–784. https://doi.org/10.1016/S0028-3932(97)00169-3
26. Mathuranath PS, Hodges JR, Mathew R, Cherian PJ, George A, 38. Woollams AM, Cooper-Pye E, Hodges JR, Patterson K (2008)
Bak TH (2004) Adaptation of the ACE for a Malayalam speaking Anomia: a doubly typical signature of semantic dementia. Neu-
population in southern India. Int J Geriatr Psychiatry 19:1188– ropsychologia 46:2503–2514. https://doi.org/10.1016/j.neuro
1194. https://doi.org/10.1002/gps.1239 psychologia.2008.04.005
27. George A, Mathuranath PS (2007) Community-based naming 39. Morrison CM, Ellis AW (1995) Roles of word frequency
agreement, familiarity, image agreement and visual complexity and age of acquisition in word naming and lexical decision.
ratings among adult Indians. Ann Indian Acad Neurol 10:92–99. J Exp Psychol Learn Mem Cogn 21:116–133. https : //doi.
https://doi.org/10.4103/0972-2327.33216 org/10.1037/0278-7393.21.1.116
28. Adlam AL, Patterson K, Bozeat S, Hodges JR (2010) Cambridge 40. Ellis AW, Ralph MAL (2000) Age of acquisition effects in
Semantic Memory Test Battery: detection of semantic deficits in adult lexical processing reflect loss of plasticity in matur-
semantic dementia and Alzheimer’s disease. Neurocase 16:193– ing systems: insights from connectionist networks. J Expt
207. https://doi.org/10.1080/13554790903405693 Psychol Learn Mem Cogn 26:1103–1123. https : //doi.
29. Green DW (2003) The neural basis of the lexicon and the grammar org/10.1037/0278-7393.26.5.1103
in L2 acquisition: the convergence hypothesis. In: van Hout R, 41. Lambon Ralph MA, Ehsan (2006) Age of acquisition effects
Hulk A, Kuiken F, Towell R (eds) The Interface between syntax depend on the mapping between representations and the frequency
and the lexicon in second language acquisition. John Benjamins, of occurrence: empirical and computational evidence. Vis Cogn
Amsterdam, pp 197–218 13:928–948. https://doi.org/10.1080/13506280544000110
30. Perani D, Abutalebi J (2005) The neural basis of first and second 42. Perani D, Abutalebi J, Paulesu E, Brambati S, Scifo P, Cappa SF
language processing. Curr Opin Neurobiol 15:202–206. https:// et al (2003) The role of age of acquisition and language usage
doi.org/10.1016/j.conb.2005.03.00 in early, high-proficient bilinguals: a fMRI study during verbal
31. Crinion J, Turner R, Grogan A, Hanakawa T, Noppeney U, Devlin fluency. Hum Brain Mapp 19:170–182. https://doi.org/10.1002/
JT et al (2006) Language control in the bilingual brain. Science hbm.10110
312:1537–1540. https://doi.org/10.1126/science.1127761 43. Wartenburger I, Heekeren HR, Abutalebi J, Cappa SF, Villringer
32. Correia J, Formisano E, Valente G, Hausfeld L, Jansma Bonte A, Perani D (2003) Early setting of grammatical processing in
M (2014) Brain-based translation: fMRI decoding of spoken the bilingual brain. Neuron 37:159–170. https://doi.org/10.1016/
words in bilinguals reveals language-independent semantic S0896-6273(02)01150-9
13