Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Neurocase

ISSN: 1355-4794 (Print) 1465-3656 (Online) Journal homepage: https://www.tandfonline.com/loi/nncs20

Do subcortical structures control ‘language


selection’ in polyglots? evidence from pathological
language mixing

Jubin Abutalebi , Antonio Miozzo & Stefano F. Cappa

To cite this article: Jubin Abutalebi , Antonio Miozzo & Stefano F. Cappa (2000) Do subcortical
structures control ‘language selection’ in polyglots? evidence from pathological language mixing,
Neurocase, 6:1, 51-56, DOI: 10.1080/13554790008402757

To link to this article: https://doi.org/10.1080/13554790008402757

Published online: 17 Jan 2008.

Submit your article to this journal

Article views: 510

View related articles

Citing articles: 5 View citing articles

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


https://www.tandfonline.com/action/journalInformation?journalCode=nncs20
Neurocase (2000) Vol. 6, pp. 51-56 0 Oxford University Press 2000

Do Subcortical Structures Control ‘Language Selection’ in


Polyglots? Evidence from Pathological Language Mixing

Jubin Abutalebi‘, Antonio Miozzo’ and Stefan0 F. Cappal**


’Neuropsychology Laboratory, Clinica Neurologica, University of Brescia Medical School, Spedali Civili, Piazzale Spedali Civili 1,
25100 Brescia and *IRCCS S.Giovanni di Dio-S. Cuore Fatebenefretelli Hospital, Brescia, Italy

Abstract

In the field of multilinguism, ‘switching’ defines a mechanism operating automatically when speakers shift among
different languages. While the neural basis of this mechanism is unknown, recent studies of polyglot aphasia have
suggested a possible role of subcortical structures. We report a case of subcortical polyglot aphasia which provides
further evidence for a role of the basal ganglia in the switching mechanism. A polyglot female (Armenian-English-
Italian), after a subcortical infarction in the language dominant hemisphere, developed a non-fluent aphasia
characterized by pathological mixing among these languages in oral production tasks. This case confirms that damage
to subcortical structures may result in different types of dysfunction in the mechanisms implicated in the selection of
languages.

Introduction
Polyglottism is a frequent and worldwide condition which resulting either in a rigid fixation in one language or,
provides unique information about the cerebral substrate of conversely, in a pathological mixing (Albert and Obler, 1978).
language processing. One of the most important questions is Aglioti and Fabbro (1993) and Aglioti et al. (1996) reported
whether the polyglot’s languages are functionally indepen- a particular recovery pattern for a patient who was unable to
dent, or to what extent they constitute a single interdependent ‘switch’ between languages because of a rigid fixation on
functional language system. The results of both electrical one language (so-called selective recovery; Paradis, 1977).
cortical stimulation (Ojemann, 1983) and functional brain and suggested that damage to the left basal ganglia may
imaging (Perani et al., 1996; Kim et al., 1997) of bilingual result in switching difficulties. Evidence for participation of
subjects have in general indicated differences among the subcortical nuclei in cognition has been relatively well defined
representations of different languages in the brain. Lesion for monolinguals (Cappa and Vallar, 1992). Most studies
studies carried out to investigate the clinical features of carried out in monolinguals with damage to the dominant
polyglot aphasia have provided a complex pattern of results. subcortical nuclei have converged by proposing a crucial
In particular, recovery studies have clearly indicated that one role of these structures in the lexicosemantic level of language
of the polyglot’s languages may recover in a different manner organization, in particular for the processes involved in
from the others. It is still debated whether it is the language lexical retrieval (Wallesch and Papagno, 1988; Crosson,
learnt earlier in life, or the one spoken most frequently and 1992; Nadeau and Crosson, 1997).
intensively before brain damage, that recovers first (Pitres’ Pathological language mixing may result from a lesion to
rule). Several recovery patterns have been described (Paradis, the control systems for language output and, in particular,
1977, 1998); of particular interest, from the neurolinguistic because of the dysfunction of a putative mechanism for
switching among languages. We report a case of subcortical
standpoint, is the so-called blending recovery, characterized
polyglot aphasia with pathological mixing which provides
by pathological mixing among languages. Language mixing
complementary evidence for the role of the dominant basal
in polyglot aphasia may offer a direct window into the
ganglia in switching mechanisms among languages.
mechanism of interaction among language systems, and
therefore provide valuable information for modelling the
neural organization of multiple language systems (Perec-
Case report
man, 1989). AH is a right-handed woman born in the Armenian commun-
Switching mechanisms might be impaired in both senses, ity of Tripoli in Northern Lebanon in 1923, whose native

Corresportderice to: S. F. Cappa. Clinica Neurologica dell’ Universiti di Brescia. I1 Neurologia Spedali Civili, 25125 Brescia. Italy. Tel: +39 30 399 5823;
Fax: +39 30 338 4086: e-mail: cappa@master.cci.unibs.it
52 J. Abutalebi, A. Miozzo and S. F. Cappa

language, Armenian (L1), was the only language spoken


until the age of 4 years. At this age, she moved with her
family to Khartoum, the capital of Sudan, where she learned
English (L2) at school, and where she lived until 1971. She
had 13 years of schooling in English and her environment
in Sudan was predominantly English speaking (Sudan was
then a British colony). Her native language, Armenian, was
only used in the family.
In 1943, when she met her future husband, an Italian
trader, she began to learn Italian (L3) informally. In 1971,
she settled in Italy, and Italian became the most used language.
She continued using English and Armenian with friends and
relatives. To summarize, Armenian (LI) and Italian (L3)
were never learnt at school, but only acquired in an informal
manner, whereas English (L2) was also learnt by means of
13 years of schooling. Despite this fact, her reading and
writing proficiency was also very good in Armenian and
Italian. As testified by relatives, her general pre-morbid
linguistic performance was excellent in all three languages.
In April 1997, AH had a left hemispheric stroke, resulting
in severe language disturbances. She was referred to our
laboratory in May 1997 for language evaluation. Her relatives
had noticed that AH was no longer able to hold a conversation
in only one given language, even when specifically asked to Fig. 1. CT scan of AH showing a small area of infarctiim deep within
subcortical structures of the left hemisphere. More precisely. the l e w n site I\
do so. AH spontaneously switched from one language to the located in the periventricular white niattcr surrounding the left c;ludate nucleuh.
other, so that monolingual Italian subjects found it extremely
difficult to understand her. AH was fully aware of her
impairment.
The neurological examination was normal except for the
language deficit. AH was fully oriented in time and space,
and there was no evidence of apraxia or agnosia on mental
status testing. Right-left recognition and discrimination were
intact. She could adequately remember events before and
after her stroke, and her short-term memory was normal. A
CT scan showed a hypodense lesion in the white matter,
adjacent to the left caudate nucleus (see Fig. 1).
On neurolinguistic testing, a non-fluent aphasia pattern
was established, involving all three languages to the same Fig. 2. Results of formal language testing with the Boston Diagnostic Aphasia
Examination. The language profiles essentially indicate a parallel recovsry for
extent. Language assessment was ascertained for Italian and AH’S three languages. i.e. a similar degree of language deticit for Armenian
English by formal testing with the Boston Diagnostic Aphasia (LI),English (L2) and Italian (L3).
Examination (BDAE; Goodglass and Kaplan, 1983). Armen-
ian was tested with an adaptation of the Boston Diagnostic Italian, she produced the following mixings between lan-
Aphasia Examination prepared by a native Armenian speaker. guages (Italian elements are underlined):
A typical ‘transcortical motor-subcortical’ aphasia pattern, I can not comunicare con you;
characterized by the reduction of spontaneous speech (con- I can not say il mio nome to you;
sisting essentially of hesitation to initiate conversation and I am a disastro today.
prolonged pauses when speaking), perseverations, preserva- Furthermore, when asked to describe the BDAE picture
tion of auditory comprehension and normal word repetition card in the English session, AH blended in a similar manner:
with the presence of semantic and verbal paraphasia, was I bambini steal the biscuits from the armadio;
present in all three languages (Fig. 2). The water sta cadendo per terra.
A characteristic pattern of language mixing was found to AH was fully aware of her mixed language output. and
be present in the language production tasks. In spontaneous she complained that her monolingual friends had some
speech, AH mixed elements of one language with elements difficulty in understanding her. When asked to switch volun-
of the other language. tarily from one language to another, she had no difficulties.
For instance, when invited to describe her daily life in Moreover, she was able to answer in a language different
Control of language selection in polyglots 53

100 -
Table 1. The performance of AH in naming across her three languages. The
table includes all the responses for the subset of items for which naming was
required in all three languages
aNon-appropriate language
UAppropriate language
Item Armenian (L1) English (L2) Italian (L3)

Clock + L3 L2
Scissors L3 + +
Lighter L3 Semantic error (L3) L2
L1 L2 L3
Chair + + L2
Fig. 3. AH‘s performance in naming. The figure shows the percentage of Key L2 + L2
correct responses in the naming tasks performed during language evaluation. Glove + L3 +
The subcolumn of the ‘non-appropriate language’ responses represents the Feather + Anomia +
percentage of responses which are semantically correct. but not performed in Hammock L3 + +
the required language (i.e. the language used in the evaluation session by Guitar + + L2
the examiner). Square L3 + L2
Triangle Neologism L3 +
Running + + +
from the one used by the examiner. The two latter conditions Sleeping L2 + +
did not affect remarkably her spontaneous mixing rate. Drinking + + +
Smoking + + L2
A noticeable characteristic of her impairment was the Falling + L3 +
fact that, despite code mixing, noun phrases were always Dripping + Semantic error Perseveration
completed in the same language; furthermore, morphological Ear L3 + L2
Nose + L3 +
markers were always used appropriately across languages. Shoulder Anomia + +
For example, in the sentences reported above, the English Ankle L3 L3 +
Wrist L2 L3 L2
verb was in correct number agreement with the Italian noun Thumb + + +
(I bambini-plural form-steal...), and vice versa (The water Elbow + L3 +
sta-3rd person singular...). Her impairment thus appeared Eyebrow Perseveration + Perseveration
to involve a stage of lexical retrieval occurring after the Cheek + Semantic error Anomia
Knee L2 + Semantic error
specification of semantic and syntactic information (see the (L2)
discussion below).
+. the target item was named correctly in the required language; L2iL3, the
target item was named correctly, but using a different language (L2 or L3)
Multilingual naming from the one of the evaluation session; perseveration refers to the item
preceding the target item.
In order to assess language mixing more formally, we
developed a three-language picture naming task, which languages. It is noteworthy that several identical target items
included a subset of 27 ‘common’ items to be named in L1, were named using a language not required; however, when
L2 and L3. invited to name the same items using the non-required
AH’s naming deficit was mild in all three languages, but language, she sometimes named them using the previously
her naming performance was impaired by her inability to unavailable language. For example, when presented with the
respond consistently in the required language. This language picture of a clock, AH responded in the L2 session with
mixing was multidirectional, as it was present to the same ‘orologio’ (instead of clock), whereas she responded to the
degree in all three languages (see Fig. 3). In the L1 session, same stimulus presented in the L3 session using ‘clock’
she named correctly (from the semantic point of view) 24 of instead of ‘orologio’. The picture of a lighter was named in
27 stimuli (89%);however, 10 of the correct responses (42%) the L1 session using L3; in the L2 session, she again used
were not given in L1, but rather in L2/L3, even when she L3, but making a semantic error (‘sigaretta’l‘cigarette’instead
was invited to use L1. The same phenomenon was observed of ‘lighter’/‘accendino’), and finally, in the L3 session, she
in the L2 and L3 sessions. In the L2 session, naming was named the stimulus using L2.
correct for 45 of 55 items (82%). using L3 14 times (31%), Considering the whole corpus of naming stimuli, AH also
whereas in the L3 session she named 44 of the 52 items made other aphasic errors, consisting mostly of anomias
(85%) using L2 14 times (32%). The prevalence of the use (n = 7; 1 for L1, 3 for L2 and 3 for L3), semantic errors
of a different language while naming was therefore similar (n = 6; 4 for L2 and 2 for L3). perseverations (n = 5; 1 for
L1, 2 for L2 and 2 for L3) and neologisms (n = 4; 1 for L1,
among the three languages (x2 = 0.88; P = 0.64, not
1 for L2 and 1 for L3). It is noteworthy that none of the 27
significant). She mixed English with Italian, Italian with
‘common’ items gave rise to aphasic errors in all three
English, Armenian with Italian or with English. Characterist-
languages.
ically, Italian or English were never observed to be mixed
with Armenian, probably due to the fact that Armenian was
the less frequently used language.
Discussion
Table 1 shows the item-specific performance for the 27 Lexical access requires the selection of the meaning of words
items for which naming was required in each of the three (semantic representation), as well as access to morpho-
54 J . Abutalebi, A. Miozzo and S. F. Cappa

syntactic information (word class, number, etc.) and the to one language represents a further crucial step towards the
corresponding phonological features (phonemic and syllabic correct production of meaningful language. Fabbro and
structure). While psycholinguistic models of lexical pro- Paradis (1995) proposed that the basal ganglia are an essential
cessing differ considerably (Dell, 1986; Levelt, 1989; Caram- structure for the link between the mere automatic functions
azza, 1997), they all include a stage in which the phonological of language and the propositional, emotional and switching
form of a word must be retrieved: this is the process which aspects, which would be subserved by the so-called ‘frontal
appears to be involved in this patient, since her responses lobe system’. One widely accepted model of basal ganglia-
were always semantically and syntactically appropriate. thalamo-cortical circuits (Alexander and Crutcher, 1990)
The deficit observed in AH could be considered to arise identifies four separate circuits (motor, oculomotor, limbic
because of language-specific word-finding difficulties, which and ‘prefrontal’). It is the latter one, which projects to the
were consciously avoided by selecting the correct item in a dorsolateral and lateral orbitofrontal cortex, that has been
different language. However, in our opinion, this is unlikely associated with cognitive processes. Considering the tight
to be the exclusive explanation of our patient’s unusual relationship between left basal ganglia and left frontal cortex,
behaviour. At the observational level, AH’S code mixing did such a neural network represents an ideal candidate for a
not appear to be under voluntary control, as the language system responsible for the control of correct language output.
shifts were not regularly preceded by hesitations, as can be In the specific topic of polyglottism, a model dealing with
expected in the case of word-finding difficulties. She also speech control in multilinguals was proposed by Green
denied the use of this strategy to circumvent naming difficul- ( 1986). This model presupposes the existence of reciprocally
ties, and appeared to be genuinely puzzled by the phenom- interdependent language subsystems for the polyglot’s lan-
enon. The unpredictable direction of the mixing, which did guages. Language output is therefore regulated by a device
not show any advantage of the most used language, also (so-called ‘specifier’) that prevents code mixing. This stage
appears to militate against a ‘strategic’ interpretation. is followed by the operation of a resource generator which,
The dysfunction of a ‘control mechanism’, subserving the following the specification of the language, activates the
lexical selection across languages, which should deactivate appropriate output system. Briefly, if the polyglot intends to
one language while speaking in a different language, seems speak in LI, the ‘language specifier’ selects the intended
to be a more plausible mechanism. This hypothesis does not language. Then, the resource generator activates the subsys-
exclude the possibility that the same ‘device’ also intervenes tems for word output in L1 and at the same time suppresses
in the complex mechanism of word finding. the word output subsystem for L2, and vice versa, in order
The neurological mechanisms responsible for lexical selec- to avoid code mixing. A lesion within this framework 01‘
tion in monolinguals are incompletely known. In the model control and activation may result in a variety of language
proposed by Wallesch and co-workers (so-called ‘Lexical deficits, essentially represented at one extreme by pathological
Decision-Making Model’; Wallesch, 1985; Wallesch and code mixing and at the other by a difficulty i n switching
Papagno, 1988), a crucial role is assigned to the subcortical among languages, as in pathological fixation on a language.
structures of the dominant hemisphere. The basic assumption The latter entity constitutes the main clinical feature observed
is that multiple lexical alternatives (e.g. the response alternat- in the phenomenon of selective recovery in polyglot aphasia
ives in a naming task) are formulated and released in the (Aglioti and Fabbro, 1993; Aglioti et d., 1996). Conversely,
posterior perisylvian cortex, then carried to the anterior the language deficit observed in AH is compatible with the
perisylvian cortex and the striatum in parallel modules. The opposite mechanism of dysfunction, i.e. an inability of the
role of the striatum is to monitor the lexical alternatives and control mechanism to deactivate the subsystems of one
to select the most appropriate one, and therefore control the language while speaking in a different language.
correct language output. The main reason for postulating a The anatomical correlation of damage to the switching
control function for the dominant basal ganglia (and the mechanism is not known. Some findings have suggested that
striatum, in particular) lies in their key position within the the neural basis for language switching may be located in
system, due essentially to their multiple afferents from the the prefrontal and frontal cortices (Stengel and Zelmanowicz,
cortex, including limbic cortex. The striatum is thus consid- 1933; Zatorre, 1989). Considering the close links between
ered to be responsible for the integration of various situational, these areas and the basal ganglia, this hypothesis is not in
emotional and motivational inputs in order to drive the contradiction with evidence from this case, as well as the
selection of the most fitting lexical item among multiple patient reported by Aglioti and Fabbro ( I 993) and Aglioti
alternatives. Aphasia reports in monolinguals with damage e? al. (1996), indicating a role for the basal ganglia region.
confined within the left basal ganglia are in agreement It is noteworthy that lesions in the basal ganglia region
with the hypothesis of an involvement of these subcortical appear to be associated with both pathological tixation
structures in the process of coordination and selection of (Aglioti and Fabbro, 1993; Aglioti rt uf., 1996). and code
language items that precede verbal production (Crosson, mixing (our patient). It may be speculated that a strategically
1992). In the specific case of bilinguals and polyglots, not located lesion (e.g. the caudate nucleus lesion in AH and the
only has the most appropriate alternative to be chosen capsulo-putaminal lesion in the patient reported by Aglioti
correctly, but the selection of linguistic elements belonging et al.) within this network of interconnected structures, which
Control of language selection in polyglots 55

are associated with complex patterns of reciprocal excitation neural substrates of parallel processing. Trends in Neuroscience 1990; 13:
266-27.
and inhibition (Graybiel, 1995), may result in opposite effects, Cappa SF. Vallar G. Neuropsychological disorders after subcortical lesions:
on the basis of the final output of the system. implications for neural models of language and spatial attention. In: Vallar
G. Cappa SF, Wallesch CW, editors. Neuropsychological disorders associated
with subcortical lesions. New York: Oxford University Press. 1992: 7 4 1 .
Caramazza A. How many levels of processing are there in lexical access?
Conclusion Cognitive Neuropsychology 1997; 14: 177-208.
Crosson B. Subcortical functions in language and memory. New York: The
The case presented in this paper suggests, from a physio- Guilford Press, 1992.
logical viewpoint, an essential role for subcortical networks Dell GS. A spreading activation theory of retrieval in sentence production.
Psychological Review 1986; 93: 283-32.
in the switching mechanism among languages in polyglots. Fabbro F, Paradis M. Differential impairments in four multilingual patients
Our findings have shown that damage to the dominant basal with subcortical lesions. In: Paradis M, editor. Aspects of bilingual aphasia.
ganglia may not only produce an inhibition of an ability to Oxford: Perganion Press, 1995: 139-76.
Goodglass H. Kaplan E. The assessment of aphasia and related disorders.
switch across languages, but also an exaggeration of this Philadelphia: Lea & Febiger, 1983.
language property may occur. It is noteworthy that on the Graybiel AM. The basal ganglia. Trends in Neuroscience 1995; 18: 6&2.
basis of this evidence, it may be hypothesized that the Green D. Control. activation. and resource: a framework and a model for the
control of speech in bilinguals. Brain and Language 1986: 27: 21LK23.
polyglot’s lexical representations are accessible and control- Kiln KHS, Relkin NR, Lee KM, Hirsch J. Distinct cortical areas associated
lable through neural routes which pertain to a cortical- with native and second languages. Nature 1997; 388: 1 7 1 4 .
subcortical loop, and in particular to a left basal ganglia-left Levelt WJM. Speaking: from intention to articulation. Cambridge, MA: MIT
Press. 1989.
frontal cortex loop. More insight might be gained by exploring Nadeau SE, Crosson B. Subcortical aphasia. Brain and Language 1997: 58:
this topic with the use of modern technology, like cognitive 355-402.
brain activation investigations, which could solidify our Ojemann CA. Brain organization for language from the perspective of
electrical stimulation mapping. Behavioral and Brain Science 1983: 6:
knowledge about the neural basis of language switching. 189-230.
Paradis M. Bilingualism and aphasia. In: Whitaker HA, Whitaker H. editors.
Studies in neurolinguistics. Vol. 3. New York: Academic Press, 1977:
65-121.
Acknowledgements Paradis M. Language and communication in multilinguals. In: Stemmer B,
Whitaker H, editors. Handbook of neurolinguistics. New York: Academic
We wish to thank Dr G. H. Mardighian for his invaluable Press, 1998: 4 17-30,
help in the Armenian assessment; we are also grateful to Perani D, Dehaene S. Grassi F, Cohen L, Cappa SF, Dupoux E rt cd. Brain
Dr M. A. Lambon Ralph and two anonymous referees for processing of native and foreign languages. NeuroReport 1996: 7: 2 4 3 9 4 .
Perecman E. Language processing in the bilingual: Evidence from language
their thoughtful comments on an earlier version of this mixing. In: Hyltenstam K. Obler LK, editors. Bilingualism across the
manuscript. lifespan. Aspects of acquisition, maturity, and loss. Cambridge, MA:
Cambridge University Press, 1989: 227-44.
Stengel E. Zelmanowicz J. Ueber polyglotte motorische Aphasic Zeitschrift
fur die Gesamte Neurologie und Psychiatrie 1933: 149: 292-311.
References Wallesch CW. Two syndromes of aphasia occurring with ischemic lesions
involving the left basal ganglia. Brain and Language 1985: 25: 357-61.
Aglioti S. Fabbro F. Paradoxical selective recovery in a bilingual aphasic Wallesch CW, Papagno C. Subcortical aphasia. In: Rose FC. Whurr R, Wyke
following subcortical lesion. NeuroReport 1993; 4: 1359-62. MA. editors. Aphasia. London: Whurr, 1988: 256-87.
Aglioti S. Beltraniello A, Girardi F. Fabbro F. Neurolinguistic and follow-up Zatorre RJ. On the representation of multiple languages in the brain: old
study of an unusual pattern of recovery from bilingual subcortical aphasia. problems and new directions. Brain and Language 1989; 36: 12747.
Brain 1996: 119: 1551-64.
Albert ML. Obler LK. The bilingual brain. New York: Academic Press, 1978. Received 011 30 October; 1998; resubmitted on 19 April, 1999;
Alexander GE. Crutcher MD. Functional architecture of basal ganglia circuits: accepted on 29 April, 1999
56 J . Abutalebi. A. Miozzo and S. F. Cappa

Do subcortical structures control


'language selection' in polyglots?
Evidence from pathological language
mixing

J. Abutalebi, A. Miozzo and S. F. Cappa


Abstract
I n the lield of multilinguism, 'switching' detines a mechanism operating
~iutoinaticallywhen speakers shift among different languages. While the neural
hnsis of this mechanism is unknown. recent studies of polyglot aphasia have
wggested a possible role of subcortical structures. We report a case of
\ubcortical polyglot aphasia which provides further evidence for a role of the
basal ganglia in the switching mechanibm. A polyglot female (Armenian-
English-Italian). after a subcortical infarction in the language dominant
hemisphere. developed a non-fluent aphasia characterized by pathological
mixing aiiiong these languages in oral production tasks. This case confirms
that damage to subcortical structures may result i n different types of dysfunction
in the mechanisms iniplicated in the selection of languages.

Journal
Neurocase I O W ; 6 : 5 1-6

Neurocase Reference Number:


0183

Primary diagnosis of interest


Polyglot aphasia

Author's designation of case


A I~I

Key theoretical issue


8 A n eahential role for subcortical networks in the language selection
mechanism among the polyglot's languages
Kcv I I ' o d S : language switching: polyglot aph ; basal ganglia lesion;
language output control

Scan, EEG and related measures


CT scanning

Standardized assessment
Bo\ton Diagnostic Aphasia Examination (BDAE)

Lesion location
0 Periventricular white matter surrounding the caudate nucleus of the left
hemisphere

Lesion type
Infiirction

Language
English

You might also like