Cognitive Neuroscience and Neuropsychology neuro(peport IN monolinguals, not only cortical areas but also specific subcortical structures are crucial for language and speech processing. While the role of the left basal ganglia in monolingual aphasia has been defined, its relevance in bilingual and polyglot aphasia is still unknown. Data have now been obtained on a patient who, following an ischaemic lesion not involving cortical structures and mainly confined to the left basal ganglia, showed severe impairments in mother tongue production, with signifi- cantly better performance in her hardly spoken second language. This dissociation remained stable for over a year and was observed both in spontaneous speech and in translation tasks. This pattern of linguistic performance, which has never been described in relation to subcortical lesions, suggests that the left basal ganglia play a relevant role in the output of a highly automatized language. Key words: Bilingual aphasia; Subcortical lesions; Implicit and explicit memory NeuroReport 4, 1359-1362 (1993) Paradoxical selective recovery in a bilingual aphasic following subcortical lesions Salvatore Aglioti’ and Franco Fabbro2 ‘stituto di Fisiologia Umana, Universita di Verona; Istituto di Fisiologia, Universita di Trieste, Via Fleming 22, 34127 Trieste, Italy A Corresponding Author Introduction Aphasic syndromes and recovery patterns in bilinguals with cortical lesions to the dominant hemi- sphere and data obtained by means of electrical cortical stimulation support the hypothesis that languages are functionally independent systems and that they are subserved by different neural substrates at the macro- anatomical and/or the micro-anatomical level.'~ Elec- trical stimulation and cerebral regional blood flow studies carried out on normal and aphasic monolin- guals reveal that the left basal ganglia and the left thala- mus are involved in language organization.** In monolinguals, a lesion to the left basal ganglia with or without involvement of the adjacent white matter, pro- vokes impairments in the speech and language output system, which are characterized by disturbances in speech rate and volume and by transcortical motor aphasia.’* Moreover, recent studies have shown that basal ganglia along with other subcortical structures and circumscribed neocortical structures are involved in implicit memory, i.e. in the learning and memoriza- tion of skilled, automatized functions which are acquired through many repetitions and not accessible to conscious introspection, whereas they are not involved in explicit memory, ic. when learning and memorizing knowledge which can be acquired from a single exposition and which individuals are aware of and may voluntarily recall.*"! Paradis” suggested that since an individual’s mother tongue (L1) is generally acquired in conversational set- tings with unconscious learning strategies typical of implicit memory, it should be anatomically and neuro- functionally organized in cortical and subcortical structures that differ from those subserving the second language (L2), which is generally learned in formal set- tings with conscious, metalinguistic learning pro- © Rapid Communications of Oxford Ltd cedures typical of explicit memory. Therefore, it would be reasonable to suppose that a lesion to the left basal ganglia should mainly impair L1, while sparing L2. Yet, not a single case of bilingual or polyglot apha- sia due to subcortical lesions has so far been reported in the literature. We present the study of a remarkable case that provides evidence of a double dissociation be- tween first and second language following subcortical lesions mainly involving the left basal ganglia. Materials, Methods and Results E.M. is a 70-year-old right-handed housewife who, in November 1990, presented a mild right sensory- motor hemisyndrome with a period of mutism lasting about two weeks. According to the description delivered by the patient and her relatives, the main resi- dual impairments consisted in a slowing down of movements and a remarkable speech deficit character- ized by the fact that the patient was no longer able to speak Venetan (Veronese dialect), which was her mother tongue and the language she had been using all her life, whereas she suddenly could resort to standard Italian, although her relatives and the medical staff addressed her in Venetan. Venetan is a Romance lan- guage spoken in northeastern Italy: it includes the dia- lects spoken in Venice (Venetian), Verona (Veronese), Treviso (Trevisan), and Padua (Paduan), which are pre- dominantly oral dialects.» During the first 3 months from the brain insult the patient’s mother tongue was so severely impaired, that she could hardly interact linguistically with her family and friends. In October 1991 the patient applied to the speech therapy service in Verona, asking to be re-educated in the comprehension and production of her mother tongue. E.M. grew up and had always lived in a family environment where Vol 4 No 12 (December issue} 30 September 1993 1359 neuro(gpeport S. Aglioti and F. Fabbro only Venetan (L1) was spoken; she had attended only 3 years of elementary school in Italian (L2). She usually watched Italian TV and read Italian magazines, but she hardly ever spoke in her second language (2-3 times in a year, according to E.M. herself), and if so with great effort and several borrowings from her L1. A neurological examination one year after the stroke revealed a slight tendon hyperflexia of the right upper limb along with severe verbal-motor inertia. In the Wechsler Adult Intelligence Scale (administered in L2) E.M. scored 92 (Verbal I1Q=94, Performance 1Q = 90). No signs of ideational or ideomotor gestural apraxia were observed. Magnetic resonance was per- formed in April 1992 (Fig. 1). E.M. spoke spontaneously only in Italian (L2). Her speech production was non-fluent, slow and character- ized by a low voice, with foreign-accent syndrome in 2.5 The foreign-accent syndrome was judged by a group of 26 professors and medical students (20 males and 6 females) whose mother tongue was Italian. They listened to recordings of E.M.’s speech production in Italian and were simply asked to say whether the voice belonged to an Italian native speaker, and if not to tell what kind of accent they could detect. Twelve of them judged E.M.’s accent to be Italian-sounding but none said that the speaker had a typical accent of the region of Veneto, whereas 14 held the speaker for a foreigner indicating different accents (German-, Slovenian-, English-, and Arabian-sounding) to a varying degree (for 6 judges it was only slight, for 5 it was moderately strong, for 2 it was very strong). In her mother tongue E.M. did not produce spontaneous speech and only on request could she, after a great effort, produce a few words, again with a low voice and foreign accent. By means of a language test (Italian version of the Aachener Aphasia Test)" a slight Broca’s aphasia was assessed. The same test was partially administered in Venetan, so as to assess the patient’s linguistic com- petence in both production and comprehension of L1. In an object naming task E.M. scored 29/30 in L2 and 12/24 in L1 (x? = 15.88, 1 df. p < 0.0001), errors in L1 consisted of 4 semantic paraphasias, 1 anomia and 7 correct words uttered in L2 instead of Li. In the description of pictures she scored 28/30 in L2 and 3/30 in L1 (x? = 41.71, 1 df.,p < 0.0001). No significant dif- ference between languages was found in comprehen- sion tasks (Token test: L2 = 36/50, L1 = 35/50; oral comprehension of words: L2 = 27/30, Li = 14/15; comprehension of oral description of events: L2 = 9/10, L1 = 8/10). E.M.’s relatively good perform- ance in language comprehension in both L1 and L2 may be due to the fact that the cortical structures of FIG. 1. Magnetic resonance image (MRI} through basal ganglia showing E.M.’s lesions in transverse (A) and coronal section (B). The images are T2 weighted: damaged areas appear white. The left hemisphere is represented on the right. The lesions have been reconstructed by two neuroradiol- ogists who did not know about the patient's clinical history and symptoms. The extent of the lesions was assessed according to Damasio and Damasio.” Black arrows highlight a subcortical infarction area involving the left basal ganglia. Small bilateral infarctions in the para- and supra- ventricular frontal white matter can also be seen. Since the patient presents only extremely slight contralesional paresis, the internal capsule must be at least partially spared. 1360 Vol 4 No 12 (December issue) 30 September 1993 Bilingual aphasia following subcortical lesions neurofpeport Table 1. E.M.’s performance in different spontaneous speech tasks Table 3. Patient's (E.M.) and control's (A.B.} results in sentence trans- lation tasks Number of x? p Number of utterances words df=1 u1 L2 Mixed Stimuli from > to Correct x? p LI L2 responses df=1 Task 1 654 874 631.31 <0.0001 51 89 76 E.M. (patient) 42.8% 57.2% 23.6% 41.2% 35.2% Sentences L112 19/20 32.4 < 0.0001 Sentences L2>11 1/20 Task 2 121 1059 4 113 46 10.3% 89.7% 2.5% 69.3% 28.2% A.B. (control) aaa ne snninleay Task 1: dialogue with the examiner in Venetan (L1). Over a period of 40 Sentences L1>L2 18/20 24 0.14 min of voice tape-recording (4 sessions in total) the patient’s utterances Sentences L29L1 19/20 account for 21/12” (53.3%), with 1528 uttered words and a fluency of 72.07 words per minute. Task 2: dialogue in Italian (L2). Over a period of 30 min of voice tape-recording (4 sessions) the patient spoke for 14’50” {49.01%), producing 1180 words with a verbal fluency of 79.55 words per min. This table shows on the /eft the number of words and on the right the number of utterances produced in tasks 1 and 2. both hemispheres, which are more involved in lan- guage comprehension, were not affected by the lesion. The rather uncommon dissociation in language pro- duction between L1 and L2 observed in E.M. was sys- tematically evaluated by analysing spontaneous speech performance in both languages and word and sentence cross-language translation tasks.'” All words and sen- tences administered in these tasks had been previously chosen from the Veronese-Italian dictionary.” When choosing the words and sentences for the translation tests, two main criteria were followed: (1) only words and sentences with exactly corresponding meaning in the two languages but different lexical stems and syn- tactic structures were chosen, (2) very common words were used. Since all sessions had been video recorded, scoring was checked with the help of the film sequences. All tests were administered by a Venetan/ Italian bilingual speech therapist. During spontaneous speech sessions only one of the languages was used all the time the examiner and the patient were together. During the sessions where only Venetan was allowed, the patient was reminded to speak Venetan several times, because she would inadvertently switch to Ita- lian, thereby violating instructions. Independent of the language to be used in each spontaneous speech task (either L1 or L2), the patient would produce most Table 2. Patient's (E.M.) and control’s (A.B.) results in word translation tasks Stimuli from—to Correct x p MEAN S.D, responses df=1 RT (s) E.M. (patient) Words L17L2 52/75 11.89 <0.0001 7.47 6.64 Words L271 31/75 10.91 11.05. A.B. (control) Words LI>L2 68/75 2.95 0.08 2.08 2.3 Words L2->11 73/75 1.88 21 Ss Errors in translation into L1 were non-responses, whereas 13 out of 23 errors in translation into L2 were either pertinent circumlocutions or semantic paraphasias (e.g. ‘rapola’ [wrinkle in L1] translated into L2 as ‘vecchiaia’ (old age]). Reaction times (RTs), reported on the right side of the table, were recorded for each item by means of a manually acti- vated stop-watch. E.M.’s RTs were significantly longer in translation tasks from L2 into L1 than from L1 into L2, whereas no significant differ- ence in translation direction was found in the control subject. Analysis of variance (ANOVA) on RTs with completely randomized design evinced F(3,220) = 29.39; p < 0.0001. Multiple comparisons (Newman— Keuls) also showed significantly slower performances of E.M. as opposed to A.B. OO SSSsSsSssSSSSSsSSs Task 7——Translation from L2 (Italian) into L1 (Venetan): out of 124 words uttered by E.M., 119 were in L2 and only 5 in L1. Task 2—Translation from L1 into L2: 121 words were uttered by E.M.., all of them in L2. A.B. did not show any mixing phenomenon: all words in task 1 {122} were in L1, and all words in task 2 (114) were in L2. utterances in L2 (Table 1). In the repetition of words and sentences E.M. did not show any paraphasia. Errors in repetition tasks in L1 were mainly due to the patient’s tendency to give a partial translation in L2 of the stimuli presented in L1, instead of simply repeating them (spontaneous translation). On the contrary, when speaking spontaneously the patient made several word errors: 20 neologisms, 28 phonemic paraphasias, 3 verbal paraphasias and 1 semantic paraphasia. Since the patient mainly spoke in L2, the majority of mis- takes were in this language. In word translation tasks (Table 2) the patient per- formed significantly quicker and better from L1 into the L2 than from L2 into L1, thus showing better results in what is considered to be the most difficult task between the two.2"° The same was true for sen- tence translation tasks (Table 3). Here again the patient performed significantly better in what is assumed to be the most difficult translation version (from L1 into L2). E.M.’s husband (A.B.), a 67-year-old bilingual (Ll =Venetan, also Veronese dialect; L2 = Italian) right hander with eight years of schooling in Italian acted as control subject to match translation perform- ance. As regards direction of translation, no significant difference was found in the control subject, both in word and sentence translation tasks (Tables 2 and 3). Discussion and Conclusion Studies on monolingual aphasics with lesions to the left basal ganglia showed that these subcortical struc- tures participate in the process of coordination and selection of language items that precede verbal pro- duction.$ In bilinguals and polyglots, the selection of a set of items belonging to one language is a crucial step towards correct production without code mixing. Moreover, switching from one language set to another language set is a typical, effortless behaviour in normal bilinguals. Since cerebral lesions generally reduce the amount and variety of behavioural sets, several neurol- ogists have tried to explain some peculiar aspects of aphasia in polyglots, such as differential or selective recovery, in terms of an impairment of the ability to switch from one language to another. An alteration in Vol 4 No 12 (December issue) 30 September 1993 1361 neuro(gpeport S. Aglioti and F. Fabbro the switching mechanisms was thought to be the cause for the permanent ‘tuning’ on one language only.” Other authors suggested that the unrecovered lan- guage was not lost but only inhibited.' These explanations, however, do not account for the reason why a cerebral lesion impairs one specific lan- guage more than another. Several hypotheses have been suggested in the past: the native language was thought to be more resistant than the languages which have been acquired later (Ribot’s rule); the most fam- iliar language was said to be more resistant (Pitres’ rule); written languages were supposed to be less liable to brain damage than oral languages, whereas other authors'?”? stressed the importance of psycho-social factors influencing language resistance to brain dam- age. However, some clinical pictures peculiar to bilingual or polyglot aphasia can be both explained and predicted in terms of most recently developed con- cepts. These include the distinction between implicit and explicit memory, which are subserved by different neurofunctional and anatomical substrates, and the hypothesis that in certain types of bilingualism L1 is organized through implicit memory, whereas formally acquired L2 predominantly relies on explicit memory. Since basal ganglia are among the most relevant structures for implicit memory," and since implicit memory is related to one’s mother tongue, this might explain why a lesion to the basal ganglia of E.M.’s dominant hemisphere impaired access to the mother tongue but did not impair production in L2 to the same extent. Our findings clearly indicate for the first time that in bilinguals the basal ganglia of the dominant hemisphere are essential subcortical nodes in the neural 1362 Vol 4 No 12 (December issue) 30 September 1993 network for correct output in the mother tongue, or any automatized Janguage which has been acquired as a habit. References 1. Paradis M. In: Boller F and Grafman J, eds. Handbook of Neuropsychology. Amsterdam: Elsevier, 1989; vol. 2, chapter 6. . Albert ML and Obler LK. The Bilingual Brain. New York: Academic Press, 1978. . Ojemann GA and Whitaker HA. Arch Neurol 35, 409 (1978). ” Ojemann GA and Creutzfeldt OD. In: Mountcastle VB, Plum F, Geiger SR, eds. Handbook of Physiology. Bethesda: American Physiological Society, 1987: vol. 5, part 2, chapter 17. . Wallesch C-W, Henriksen L, Kornhuber H-H et af. Brain Lang 25, 224 (1985). . Démonet J-F, Puei M, Celsis P et al. J Neuroling 6, 319 (1991). " Alexander MP. In: Boller F and Grafman J, eds. Handbook of Neuropsychology. Amsterdam: Elsevier, 1989: vol. 2, chapter 2. 8. Crosson B. Subcortical Functions in Language and Memory. New York: Guild- ford Press, 1992. 9. Cohen NJ. In: Squire LR and Butters N, eds. Neuropsychology of Memory. New York: Guildford Press, 1984: chapter 10. 10. Tulving E. Human Neurobiol 6, 67 (1987). 11. Saint-Cyr JA, Taylor AE and Lang AE. Brain 111, 941 (1988). 12. Paradis M. In: Ellis N, ed. fmplicit and Explicit Language Learning. In press (1993). 13. Tagliavini C. Introduzione alla Glottologia. Bologna: Patron, 1969. 14. Graff-Radford NR, Cooper WE, Colsher PL et al. Brain Lang 28 (1986). 15. Blumstein SE, Alexander MP, Ryalls JH. Brain Lang 31, 215 (1987). 16. Huber W, Poeck K, Weniger D et al, Aachener Aphasietest. Italian version. Firenze: Organizzazioni Speciali, 1991. 17. Paradis M. The Assessment of Bilingual Aphasia. Hillsdale, NJ: LEA, 1987. 18. Beltramini G and Donati E. Piccolo Dizionario Veronese-Italiano. Verona: Vita Veronese Ediz., 1982. 49. Paradis M, Goldblum M-C and Abadi R. Brain Lang 15, 55 (1982). 20. Fabbro F, Gran L, Basso G et al. Brain Lang 39, 69 (1990). 21. Pétzl O. Zeitschr ges Neurolog Psychiat 124, 145 (1930). 22. Minkowski M. In: Halpern L, ed. Problems of Dynamic Neurology. Jerusalem: Hebrew Univ., 1963: 119-161. 23. Damasio H and Damasio AR. Lesion Analysis in Neuropsychology. New York: Oxford University Press, 1989. PWN Non ACKNOWLEDGEMENTS: We wish to thank the speech therapist Mrs Flavia Girardi, the staff of the speech therapy unit and the radiologist Dr A. Beltramello, all at the University Hospital ‘Borgo Roma’, Verona, Italy. Special thanks go to Professor M. Paradis for his valuable suggestions. Received 7 June 1993; accepted 28 July 1993