Michel
Paradis
Department of Linguistics,
and
Centre de Neuroscience de
Because patients may recover their
languages to vastly different extents, in order to ensure a valid diagnosis, (1)
one should assess with a linguistically and culturally equivalent instrument all
languages previously spoken by the patient; (2) this instrument should not be
the mere translation of a battery designed for, and standardized in, another
language; and (3) it is important to know that manifestations of aphasia
symptoms are different in various languages. If therapy is not effective in one
language, it should be tried in another. In addition to reliance on pragmatics,
in some patients, metalinguistic knowledge (possibly more extensive in one of
the languages), and translation from one language to another may be used as a
compensatory strategies.
When proficient bilingual or
multilingual individuals regain consciousness after a stroke or a traumatic
accident, one would expect them to recover all of their languages at the same
time and to the same extent. If they were more fluent in one of their languages
premorbidly, one would expect this difference to be reflected in the extent of
impairment in each language, with higher scores in the previously stronger
language. This pattern is indeed exhibited by a number of patients. It has been
called parallel recovery because the
aphasic speech parallels the previous relative abilities in the two languages.
However, a number of exceptions have been reported where patients recover their
languages in one of several other, nonparallel,
patterns. The term differential recovery
refers to cases where one language is recovered much better than the other
relative to their premorbid fluency. When one language is recovered long before
the other, it is called a successive
recovery. Selective recovery
refers to the fact that only one of a patient’s languages is recovered. In
cases of antagonistic recovery, only
one language is available to the patient at first; then, as a second language
gradually becomes available, the first recovered
language proportionately regresses and eventually disappears. If this pattern
repeats itself, that is, if the available and unavailable languages switch more
than once over time, it is called alternating
antagonism, as the languages alternate over periods of from 24 hours to a
week or even months. Patients have also been reported to unwittingly mix their
two languages and not to be able to speak only one language at a time. This has
been termed mixed or blending recovery and refers to mixing or blending two languages in the way
that some unilingual patients blend words or syntactic constructions.
The practice of assessing bilingual
patients in only one of their languages was unfortunately prevalent in most
hospitals until very recently. Even in cities where the population is largely
bilingual, as in
Because no factor has been
identified that can predict the status of one language relative to another, it
is imperative that aphasic patients be assessed in all the languages that they
spoke premorbidly regardless of their respective premorbid fluency. Indeed,
some patients have been reported to have best recovered the language they spoke
the least well before insult, and for which they had no particular affective
bias. Assessment in only one language is therefore insufficient for at least
two major reasons. Not testing one of a patient’s languages may have
detrimental social and/or clinical consequences. On the one hand, patients may
be unaware of which of their two languages is more affected or may erroneously
believe that one of them has been better recovered. In one such case observed
in
Not only must all languages be
tested, but each must be assessed with a linguistically and culturally
equivalent instrument so that the results obtained may be validly compared. The
manifestations of aphasia are constrained by the corollary to Murphy’s Law that
states that only that which can go wrong will go wrong. In other words, the
structure of the language determines what types of errors may occur. The reason
why a certain type of error is salient or conspicuous in a given language may
be due to one or more of several factors: (1) the incidence of obligatory
contexts, (2) the importance of the form for the derivation of meaning, (3) the
frequency of use of the item in a language/culture, (4) the structural
complexity of the item (e.g., number of deviancies from the canonical form),
(5) the presence or absence of redundancy (e.g., word order and agreement vs.
either word order or agreement); (6) the presence or absence of a zero morpheme
and whether nouns and verbs exist as bare roots or must necessarily be inflected;
(7) whether, when inflections are omitted, the remaining form is pronounceable
or not; and (8) whether the form is memorized or derivable by rule, i.e.,
whether it is regular or irregular. The form of the error will likely depend on
the type of aphasia (e.g., omission versus substitution and/or type of
substitution), but the pool of possible errors (i.e., what can be omitted or substituted) is restricted by the grammar of the
language. In other words, the form of the errors may depend on the type of
aphasia, though potential errors are constrained by the structural
characteristics of each language. An agrammatic patient may substitute a
default morpheme (the zero morpheme when available), but when this is not an
option because it would lead to an unpronounceable form or merely to a nonword,
some other, less marked, morpheme, such as the nominative case for nouns and
the infinitive for verbs, again modulated by frequency of use, may be
systematically utilized.
One reason why it is important for language
pathologists to be aware of the particular manifestations of aphasic symptoms
in a given language or family of languages is to avoid misdiagnosis. For
example, as long as agrammatism was defined as the omission of grammatical
morphemes in obligatory contexts and paragrammatism as their substitution,
there was a risk of diagnosing agrammatism in English (or Spanish or Hungarian)
but paragrammatism in Hebrew or Arabic, in an English- (or Spanish- or
Hungarian-) Hebrew or Arabic bilingual patient who was actually agrammatic in
both languages. Now, if we assume, say, that clinically, agrammatism is broadly
characterized as a reduction in the use of free and bound inflectional
morphology, lack of grammatical agreement, reduction in the use of complex sentence
structures, and use of telegraphic style, we may expect that, given the
structural characteristics of specific languages, some of these manifestations
will be more or less salient. At the very least, in a language that does not
contain particular items (e.g., definite articles),
aphasic patients will not have the opportunity to drop them in obligatory
contexts—in fact, in any context, since there are none to be dropped. Assuming
that such a feature does exist but occurs only rarely, again, there will be few
opportunities to show a deficit in its use. The clinician and researcher should
therefore be aware that what is generally true in their own language may not be
in the language of the patient under assessment.
Therefore, the instrument of
evaluation should not be the mere translation of a battery designed for, and
standardized in, another language. Each task requires a different criterion of
cross-language equivalence, depending on the rationale behind the design of the
task (i.e., what it is that we want to measure, and why) and on the structural
distance between the languages and the cultural differences between the
communities. For example, in the Bilingual
Aphasia Test (BAT), the criterion varies with each of its 32 tasks, for a
number of reasons. Stimulus items may be culturally inappropriate. They may
refer to objects that either are not part of the culture or may look or work
differently, and hence would not be recognized or would be misunderstood. Some
grammatical constructions may present very different levels of difficulty than
in the original language of the battery. Most obviously, any task that is based
on phonological minimal pairs or rhyming words will simply not work at all when
translated into any other language. Some series, such as days of the week or
months of the year, are not available in some languages. Words may have a
frequency of use that varies widely from that of their translation equivalent.
Common syntactic constructions, such as the passive in English, are rarely if
ever used in some other languages. The point need not be further belabored.
Suffice it to say that translations will usually not yield interpretable
results. Rather, corresponding items in another language must be selected so as
to tap the same information as the original, in accordance with the rationale
that served to construct the items in the first place. Such a transposition may
yield a test in which the items have no obvious resemblance with the
original—and yet they are much more valid than a mere translation would have
been (i.e., they test the same underlying characteristics as the original,
albeit through different means).
Some additional cultural concerns
related to testing implementation may have to be taken into account. The
patients’ responses are to be considered in light of their topolect and
sociolect, with respect to both pronunciation and morphosyntax. The way
patients are addressed will depend on local customs. In
languages with different pronouns for different categories of addressees (tu/vous
in French; Du/Sie in German; gij/U in Dutch, etc.). The BAT
specifies that the pronouns used should be a function of age, sex, and
perceived status relative to the interviewer, in order that the patient be made to feel as comfortable as possible. In some cultures,
men will feel uncomfortable being tested by women or by younger persons, a fortiori by young women. In other
countries, women will feel uncomfortable being tested by men. Test results may
be influenced by these factors.
In the literature, we find reports
of partial transfer of benefits from the treated to the
nontreated language; of benefits only in or greater in the nontreated mother
tongue subsequent to therapy in the second language; of lack of transfer to the nontreated language; as well
as of treatment in both languages resulting in improvement in only one. We may
therefore conclude, at the least, that it is possible for therapy to be effective in only one of a patient's
languages and not in the other, even when at the onset of therapy both
languages present the same qualitative and quantitative picture (i.e.,
equivalent symptoms to the same degree). Hence, if therapy shows no effect in
one language, that does not mean that it should not be
attempted in the other.
Based on the behavior of some
bilingual aphasic patients, three compensatory strategies that could be used in
rehabilitation come to mind: The use of metalinguistic knowledge, which may be
more extensive in the second, later learned, language; the substitution of
pragmatics where syntax and/or lexical access are unavailable; and the use of
translation, possibly via metalinguistic knowledge, at any level of linguistic
structure. Indeed, because some bilingual aphasic patients are able to translate
into a language that is not available to them for spontaneous speech or even
naming common objects, it has been suggested that translation might be used
when words in one language are not available. One patient was recently
reported, who had considerable language impairments following a left functional
hemispherectomy, was able to compensate for her severe lexical access deficit
by accessing words in the other language. She was also able to use gestures
when a word was not accessible in either language.
Conclusion
Bilingual
individuals with aphasia do not necessarily recover both or all of their
languages at the same time and to the same extent. There is no known
correspondence between pattern of recovery and any single biological or
ecological factor. Sometimes the second language may appear to be recovered
better than the native language because patients rely on metalinguistic
knowledge.
·
All
languages of a person with aphasia should be assessed with a linguistically and
culturally equivalent instrument, not a simple translation of a standardized
test from another language.
·
The
fact that therapy is ineffective in one language does not mean it will not work
in another.
·
Translation
may be available when word retrieval is not.
Sources:
Paradis,
M. (2001) The need for awareness of aphasia symptoms
in different languages. In M. Paradis (ed.), Manifestations of aphasia symptoms in
different languages (pp. 1-7).
Paradis, M. (2004). A neurolinguistic theory of
bilingualism (Chapter 3: Bilingual aphasia).
(and references found therein.)