Cognition, 17 (1984) 275-287

Discussion

Cautiosnthe words you use may affect what you say:

A response to Au*
ALFRED H. BLOOM**

Swarthmore College
The claim that the language or languages we learn determine the ways we
think is clearly untenable. But it does not necessarily follow that language is
merely a code system which neither affects the process by which thinking
proceeds nor the nature of the thoughts manipulated in that process. Between
these two extremes there is a substantial middle ground which constitutes a
promising and important area of experimental research.
In Bloom (1981) I hypothesized that once linguistic labels, lexical or grammatical, have been mastered, their presence in an oral or written utterance
acts to trigger in the mind of the hearer or reader the specific cognitive
schemas associated with those labels, thereby not only providing a primary
level of comprehension of the utterance but also establishing specific direction
to further thought processing with respect to it.
This is not to say that contextual cues registered prior to th,e processing of
the utterance do not prepare the hearer or reader to respond appropriately
to the labels of the utterance. Nor is it to say that, in the case of an ambiguous
label, contextual cues do not assist the hearer or reader in disambiguating
which of the various cognitive schemas associated with that label is intended
in the particular context. Nor is it to say that the hearer or reader does not
call upon both world knowledge and familiarity with the present context in
order to construct an enriched interpretation of the utterance and to formulate a response to it. But it is to say that, whatever the context, the spe,cific
labels of the utterance act to elicit the construction of a specific perspective
on thst context; and it is that perspective in turn which, while not determining
the hearer or readers further thoughts, provides direction to those thoughts.
Cross-linguistically, it follows then that if on.e language has a specific label
for a given concept and another does not, the process by which that concept
is elicited in each language should differ in both form and cognitive implications, When a language has a direct label for a concept, through the use of

*A response to Tery Kit-fong Aus paper Chinese and English counterfactuals: The Sapir-Wharf
Hypothesis revisited.
**Requests for reprints should be sent to Alfred H. Bloom, Linguistics Program, Swarthmore College,
Swarthmore, PA 19081, U.S.A.

OOlO-277/84/$4.40

0 Elsevier Sequoh/Printed in The Netherlands

276

A.H. Bloom

its label, the concept can be triggered directly, thereby establishing it as a

cognitive standpoint from which to proceed. By contrast, in a language in
which a direct label for the concept is not available, other available labels
will be used to trigger their respective associated schzmas, with the expectation that these schemas will in turn, in conjunction with additional information, either provided or taken to be already known, lead the hearer or reader,
albeit indirectly, to the intended concept. An indirect process of this kind,
however, is subject, in a way that a direct process is not, to the influence of
intervening factors which may, for better or worse,, derail it from its intended
result. On the one hand, indirect elicitation of a concept is likely to leave the
hearer or reader freer to bring into consideration .alternative perspectives
which may in turn lead to a questioning of the validizy or appropriateness of
that concept. On the other hand, indirect elicitation is likely to leave the
hearer or reader more vulnerable to the effects of distracting complexities
which may simply interfere with his/her ability to arrive at the intended concept.
A comparison of English and Mandarin Chinese with respect to the ways
in which both languages express counter-factual thoughts provides a natural
testing ground for the hypothesized relationship between availability of labels
and modes of processing. English speakers make uS;eof specific counterfactual labels (Le., if x were . . . , y would be.. . , in the present, and if x had
been... , y would have been.. . , in the past) to mark the notion that a premise
(x), acknowledged to be false, is being entertained purely for the purpose of
drawing implications as to what would be or woulo have been the case if x
were true. By contrast, although speakers of Chinese express similar
thoughts, they do not do so by means of labels specific to them. Rather, to
express counter-factual thoughts, Chinese speakers make use of labels which
signal the existence of an implicational relationship in a context in which it
is already known, or directly stated, that the premise upon which the implication is based is false. For example, a Chinese speaker, in a situation in which
it is known that John is not taking linguistics, might say If John tak.es linguistics, he becomes very excited about it and the remark would be understood
as expressing a meaning roughly equivalent to the English, If John were
taking linguistics, he Twouldbe very excited about it. Or in a situation in
which it is not known whether in fact John took linguistics, a Chinese speaker
might
state explicitly John did not take linguistics and then follow that
statement by the past implicational statement If he did, then he was excited
about it and the remark would again be accorded a counterfactual interpretation-ie.,
be interpreted as roughly equivalent to the English, *If he had
taken linguistics, he would have been excited about it.
Two studies were conducted to test the impact of these cross-linguistic

Response to Au

differences on speakers processing of counterfactual thoughts.

In the first study native speakers of Chinese and English responded
native language to the following question:

277

in their

If all circles were large and this small triangle A were a circle, would it be

large?
In Chinese, for lack of labels specific to the counterfactual,
wording of the question is closer to the English:

the actual

If all circles are large and if this small triangle A is a circle, is the triangle large?

On the basis of the view advanced above, the presence of the specific
counterfactual labels were and would be should act to trigger directly in
the minds of English speakers the establishment of a counterfactua! perspective as an orienting framework on the basis of which further processing proceeds. Taking that perspective as a given, consideration of the truth or falsehood of the premises themselves becomes irrelevant, and the question yields
a somewhat bothersome, self-evident affirmative response. By contrast, there
are no comparable labels which can be expected to trigger a counterfactual
perspective in the minds of Chinese speakers. Rather, the Chinese speaker
is first directed to the implicational premises, if all circles are large . . . and
if this triangle is a circle . . ., with the expectation that he or she will integrate
these premises with the knowledge that they cannot be true, and, thereby.
arrive indirectly at an understanding of the counterfactual intent of the question. However, with his/her attention focused initially on the premises themselves, in the absence of any direct signal tc preempt consideration of the
validity of the premises, the subject is likely to become involved in consideration of the very question of whether or not the premises are or could be
true. In face of the fact that both are blatantly untenable-all circles cannot
be large and a triangle cannot be a circle- and in the absence of any explicit
signal to indicate that that fact is irrelevant, the subject is likely to reject the
question itself as absurd or to respond negatively to it.
In confirmation of these predictions, 83% of the 115 English-speaking
subjects responded Yes, whereas only 25% of the 176 Taiwanese, native
Chinese-speaking subjects responded in this way, with the remaining subjects
responding either No or Unsure*. Moreover, when the question was
asked orally and informally, even more dramatic cross-linguistic differences
of the same type emerged. Furthermore, in elaborating upon their responses,
while English speakers tended to remark Of course, the circle would be
large; but it seems too easy, there must be some trick, Chinese-speakers
tended to wonder, How can all circles be large? How can a triangle be a.
Gcle? What are you talking about?.

2%

A.H. Bloom

Furthermore, the Japanese language, like the Chinese language, lacks

specific labels for counterfactual thoughts and, again like Chinese, signals
these thoughts by the use of implicational statements uttered in contexts in
which the premises of such statements are explicitly stipulated or otherwise
known to be false. Thus, if the hypothesized relationship between availability
of linguistic labels and mode of processing has cross-linguistic generality,
Japanese subjects, like their Chinese counterparts, should display lower counterfactual responding than comparable American subjects. In confirmation
of this expectation, of the 151 Japanese college students from Ryukoku University in Kyoto, Kyoto Bunkyo College and Kyoto University who were
asked to respond to the above question in Japanese, only 57% gave affirmative responses (cf. Bloom, 1984, Reference Note 1).
In the second study native speakers of Chinese and English responded in
their native language to paragraphs of the following general form: A is not
the case, but if A were the case, then B would have been the case, C would
have been the case, and D would have been the case. In Chinese, as a
consequence of the absence of any explicit label to signal counterfactual
intent, the paragraphs read in a manner roughly equivalent to the English,
A is not the case, but if A was the case, then B was the case, C was the
case and D was the case. After reading a paragraph the subject was asked
whether, according to that paragraph, the last of the implications presented,
(Le., D), actually took place. (For the text of the paragraphs and the response
formats cf. Blooim (1981)).
In order to assess the impact of distracting complexities on maintaining a
counterfactual standpoint, the counterfactual implications A through D were
presented in a moderately complex and in a highly complex version. In the
highly complex version, the statement of the false premise, which leads the
subject to construct a counter-factual interpretation, is followed first by a
distracting discussion of factual material relating to differences between Eastg:a and Western philosophy and then by the set of counterfactual implications. Thus, in order to arrive at the required counterfactual interpretation
of those implications, the subject must retain a counterfactual perspective
across the intrusion of distracting factual information. In the moderately compIex version, by contrast, the counterfactual perspective is reinforced by a
restatement of the false premise following the introduction of the distratiting
factual information. The subject can arrive at the required counterfactual
interpretation of the final implications without retaining a counterfactual
perspective across the distracting factual information. Furthermore, while
the highly complex version contains only two terms which point to possibility
rather than fact, the moderately complex version contains seven such terms,
haeasing the likelihood that the subject will consider a counterfactual in-

Response to Au

279

terpretation. Finally, the moderately complex version, in contrast to the

highly complex one, contains an intentional logical contradiction, which is
resolvable only through the adoption of a counterfactual interpretation and
which was included to encourage such an interpretation (cf. Bloom, 1981,
pp. 26-27).
Given the presence in the English versions of the paragraphs of multiple
counterfactual labels, it was expected that English-speaking subjects would
immediately adopt a counterfactual standpoint, maintain it without difficulty,
and, as a consequence, interpret the last implication as a statement about
what would have occurred if the premise were true. Xnety-seven percent of
the 107 subjects did just that, regardless of level of complexity. By contrast
only 53% of the 82 Taiwanese, native Chinese-speaking subjects responded
counterfactually to the moderately complex version. And, even more dramatically, only 7% 08 the 103 Taiwanese, native Chinese-speaking subjects responded counterfactually to the highly complex version
A few native Chinese-speaking Chinese-English bilinguals who offered
counterfactual responses to an earlier Chinese version of the paragraph, in
fact, wrote the English words would have in the margins of the Chinese
text, which suggests that these English labels assisted them in maintaining a
counterfactual stanclpoint in Chinese. Moreover, a parallel but smaller sanpie of Hong Kong subjects, who I was assured were fully educated within the
Chinese stream of the Hong Kong school system, and who were at the time
of the study enrolled in special summer courses in English at Hong Kong
University, showed almost identical results to those of the Taiwanese subjects, with 50% of the 20 subjects who responded to the moderately complex
version responding counterfactually and only 1 of the 17 subjects who responded to the highly complex version responding counterfactually. And
again, lending a further measure of cross-cultural generality to the findings,
of the 151 Japanese college students who responded to a moderately complex
Japanese version of this paragraph, only 40% interpreted its final implication
counterfactually.
In summary, the results of the two studies clearly demonstrate (A) that
the presence of counterfactual labels elicits with virtual certainty the adoption
of a counterfactual standpoint; and (B) that the absence of such labels makes
couRterfactual processing subject to derailment, either because it leaves the
subject freer to question the validity of adopting such a standpoint or because
it leaves the subject more vulnerable to distraction.
Terry Kit-fong Au (1983) attempted to replicate th!e results of the second
study and to identify more precisely the psycholinguistic mechanisms which
account for the English-speakers maintaining a counterfactual standpoint.
She proceeded, however, on the basis of three severely misguided assumptions.

280

A.H. Bfoom

In the first place she assumed that her subjects were comparable to the
subjects who had participated in my studies. My subjects were college students and adults whose education, with the exception of foreign language
courses, had been fully in Chinese and who were iimmersedprincipally within
a Chinese psycholinguistic world. Hers, by contrast, were all students in the
Anglo-Chinese track of the Hong Kong educational system Nhich instructs
primarily in English and which prepares students for entry into English universities. They were subjects, in other words, who were thoroughly immersed
in an English psycholinguistic world. This difference between subject pools,
moreover, becomes particularly consequential when one considers, as I argued in my earlier discussion of this work (cf. Bloom, 1981, p. 32), that native
Chine:;e-speaking subjects who have been deeply acculturated into the English psycholinguistic world are likely, as a consequence of the influence of
that world, to attach a counterfactual interpretation, as a second meaning,
to their Chinese labels for if . . . then, and in so doing to transform those
labels into direct, though ambiguous, signals of both counterfactual and implicational thoughts. At that point, these bilingual Chinese-speakers come to
have Chinese labels for counterfactual thoughts and thus can be expected to
respond directly, just as English speakers do, to invitations into the counterfactual realm.
Au (1983, foot of p. 172) acknowledges the possibility of this phenomenon,
but fails to recognize that it renders her subject pool inappropriate for testing
the distinctive cognitive effects of the Chinese language on its speakers processing of counterfactual thoughts. In fact, in that Aus subject population
permits the testing of the cognitive effects on native Chinese-speakers of
immersion in Western languages and thought, it might even be seen as constituting a useful control group for my own work.
Secondly, claiming that my paragraphs were non-idiomatic, Au set about
to construct one of her own which she presumed, idiomatic quality aside, to
be comparable to mine in other respects. However, while mine took the
genera? form:
Bier was an Eighteenth Century philosopher who wanted very much to investigate the principles of the universe and the laws of Jature. He could not read
Chinese., but if he had been able to read Chinese, he woulid have discovered
Chinese philosophical works relevant to his investigation. He would have been
particularly influenced by the fact that Chinese philosophers generally focused
on the interrelationships between natural phenomena, while Western
philosophers generally focused on such phenomena as individual entities. He
would havesynthesizedChinese and Western views, created a new philosophical
theory, overcome weaknesses in Western thought and brought Western

philosophycloser to science.

Response to A;c

28 1

Aus took the general form:

A Dutch explorer in Central Africa witnessed an event involving natives who
were first observed throwing a dead human body into a pot of boiling water and
then drinking collectively from the pot. Upon seeing the event the explorer fled,
but if he had been able to understand the native language and had not fled so
quickly he would have learnt that the dead native was actually a hero of the
tribe, and was killed in an accident; that the natives drank the broth of their
hero because they believed that only by doing so could they acquire the virtues
of the hero; that the natives were very friendly, and were not cruel and savage
as he thought.

My paragraphs involve rather abstract, intellectually demanding ccntent

and, moreover, limit factual statements to the original premise (i.e., Bier was
an Eighteenth Century Western philosopher who wanted . . . and who could
not speak Chinese), casting all subsequent implications as counterfactuals,
presumptions as to what would have happened if. Aus, by contrast, involve,
more concrete, less intellectually demanding content, and relate as facts not
only the original observation of the body being thrown into the boiling pot,
the shared drinking of the broth, the explorers inability to spetik the native
language and his fleeing from the scene, but all of the subsequent Implications
as well-i.e., the fact that the dead native was a hero of the t&r;: who had
died in an accident, the fact that the natives believed that only by drinking
the broth would they acquire the virtues of their hero and the fact that the
natives were friendly and not cruel. In fact, the only aspect of Aus story
which is counterfactual is that the explorer did not know these facts, bu.t
would have if he had been able to speak the language and had stayed around.
Not only, then, are our two test instruments not comparable with respect to
the complexity of their content, but they are quite distinct as well with respect
to the degree of counterfactual interpretation they demand. On both accounts
we would expect Aus paragraphs to present to Chinese-speakers fewer impediments to counterfactual interpretation than mine.
Finally, based again on the claim that my paragraphs were non-idiomatic, Au dismisses my most complex version and assumes thiat the simpler
version can suffice for her work. But in so doing, she again misses the point
that the hypothesis is not that Chinese-speakers do not arrive at counterfa+
tual interpretations, but rather that they do so in a different way from 2:;.
glish-speakers and that those differences are only likely to emerge in contexts
of increasing distractive complexity--or where attenuating considerations
may be brought to bear, as in the triangle-to-circle questions which she ignores altogether.
Proceeding then on the basis of these assumptions, Au presented to her
subject pool of Chinese-speakers, who are thoroughly immersed in an English

282

A.H. Bloom

psycholinguistic world, a Chinese and an English version of her own relatively

simple, and only marginally counterfactual paragraph as well as a Chinese
and an English version of one of my moderately complex paragraphs. She
obtained the following results:
Nearly 100% of her subjects responded counterfactually to the Chinese
version of her story and, on average, 97% of her subjects responde:d counterfactually to the English version of her story. Eighty-six percent responded
counterfactually to the Chinese version of my story and 89% to the English
version of my story (cf. Au (1983, pp. 165 and 171)).
Au interprets her subjects almost perfect counterfactual response to both
the Chinese and English version of her story as demonstrating that the absence of counterfactual labels does not affect the processing of counterfactual
thought. She attempts to expkain the somewhat lower level of counterfactual
responding by her subjects to the Chinese version of my story (86%) as well
as the overall markedly lower level of counterfactual responding by my subjects, as an artifact of the supposedly non-idiomatic quality of my test
instruments and concludes that neither her results nor mine give evidence of
any linguistic effect on thought.
However, if we interpret Aus results in light of the differences outlined
above, between her subject pool and mine and her paragraph and mine, quite
a different story unfolds, and one which, moreover, lends strong support to
the claim that linguistic labels do play a significant role in orienting speakers
thoughts.
That Aus subjects did not have difficulty responding to the Chinese version of her story follows from the characteristics of her story and the characteristics of her subject pool. On the one hand, her story is both rather simple
and only barely counterfactual, unlikely to distract even the indirect processor
from its counterfactual intent. On the other hand, the large majority of her
subjects are likely, as a function of their immersion in the English psycholinguistic world, to have transformed their Chinese labels if/then so that those
labels trigger directly either a schema for implicational thought or a schema
for counterfa~zual thought. That her subjects did not have difficulty responding to the English version of her story follows from the relative simplicity of
the story and from the fact that, as proficient English speakers, they can be
expected to have acquired adequate mastery of English counterfactual labels.
That her subjects responded at a somewhat lower level of counterfactuality
to the Chinese version of my paragraph than they did to the Chinese version
of her own story (86% versus 100%) follows from the expected impact on
counterfactual response of the differing levels of complexity and counterfactuality of the two stories.
Finally, that her subjects responded to the Chinese version of my moder-

Response to Au

283

ately complex paragraph at a markedly higher counterfactual rate than did

my own (86% versus 55%) follows from the expected impact on counterfactual processing of the difference in the degree to which the two groups of
subjects had been exposed to the English psycholinguistic world.
Regarding Aus recurring claim that my test instruments are non-idiomatic, i.t should be stressed that the Chinese versions of my stories were all
written by native Chinese-speaking professors at the National Taiwan University and judged by them to be in grammatical, acceptable and comprehensible
Chinese. The aim in constructing the stories was, of course, not that of typifying the most colloquial of styles, for that was not the issue, but rather that
of varying the difficulty of style and content, within the range of grammatical,
acceptable and comprehensible Chinese, so as to be able to test the impact
of that variation on level of counterfactual response. Ironically, Aus results
confirm the fact that my test instruments do not violate that range. If a
problem of idiomatic quality were the reason for the lower counterfactual
responding of my subjects then how would Au explain the fact that 86% of
her subjects responded counter-factually to the identical test instrument?
Moreover, it is interesting to note that what counts for Au as a paragraph
written in more idiomatic Chinese is one that is markedly less countSerfactual
in content.
Au took two further steps to bolster her position. In the first, she presented
the original English version of both her story and my story as well as several
less idiomatic versions of each to a group of American high school students.
Ninety-seven percent of these high school subjects responded counterfactually to her story; only 72% responded counterfactually to mine (cf. Au, 1983,
p. 175). From Aus point of view, these findings, by demonstrating that
non-idiomatic language can lower counterfactual response, increase the
plausibility of her claim that the supposed non-idiomatic quality of my test
instruments was responsible for the lower level of counterfactual response of
my Chinese subjects. But in drawing this inference, Au fails again to recognize that levc! of complexity rather than idiomatic quality is the issue at hand
and fails to appreciate the dangers inherent in comparing two subject pools
of radically differing characteristics - English-speaking American high
school subjects and Chinese-speaking students of the most selective universities of Taiwan and Hong Kong, My paragraphs were definitely and intentionally complex in content, but my subjects, both Chinese and American,
were selected so that all would be able to handle comfortably the level of
complexity required. As a consequence, lower counterfactual responding
among my Chinese subjects could confidently be attributed to the unavailability of counterfactual labels. By contrast, the lower counterfactual responding
of Aus English-speaking high school subjects is likely to result from an intel-

284

A.H. Bloom

Iectual inability, mode of counterfactual processing aside, to grapple with the

level of complexity of the content involved (for example, the distinction be-

tween focusing on natural phenomena as individual entities and focusing on

their mutual interrelationships, cf. Bier story above). If the inability of her
American subjects to offer counterfactual interpretations is, as she maintains,
the result of incomprehensible material, rather than simply the result of an
intellectual inability to handle it, then one would expect that intellectually
sophisticated English-speaking subjects would likewise have difficulty with
the test instrument in question. Yet it is clear that such is not the case. Nearly
100% of my intellectually sophisticated English-speaking subjects responded
counterfactually to the identical test instrument, as did, on average, 89% of
Aus own bilingual subjects, who were students in intellectually demanding
high schools in Hong Kong. Au cannot conclude that lower counter-factual
responding among my subjects and lower counterfactual responding among
hers h?s the same etiology. In the case of my intellectually sophisticated
Chinese subjects, lower counterfactual responding is likely to reflect experience in a psycholinguistic tradition which does not specifically label counterfactual thought. In the case of her American high school ,students, it is likely
to reflect a lower level of sophistication in a psycholinguistic tradition which
does.
As a second step, finally admitting that ... since the Chinese bilinguals in
my studies have mastered English quite well, it could be that they have
extended the counterfactual schema which was originally associated with the
English subjective to their Chinese linguistic world (p, 177), Au turned to
Chinese speakers who, in presumed contrast to those subjects who had participated in her major study, . . . used Chinese almost exclusively in daily life,
and very rarely spoke English spontaneously outside of class (p. 177).
The selected group of native Chinese fourth to seventh graders, who had
studied English for about four to seven years, was first asked to translate
from Chinese into English the following statement:
Mrs. Wangdoes not know English. If Mrs. Wong knew English, she would read
English books.
Of COW% for lack of counterfactual

markers in Chinese, the original

Chinese statement reads closer to the English:
Mrs. wong does not know English. If she kGo\vsEnglish, she is able to read
English books.
my one of the 169 subjects successfully translated the Chinese statement
into an English statement that included the correct counter-factualelements,
even though they were all able to provide basic translations of the sentence-

Response to Au

285

a rather dramatic demonstration of the fact that counterfactual markers represent a particularly difficult aspect of English grammar for Chinese students
of English to master. As I discussed in Bloom (1981, p. 17), the grammatical
characteristics of a second language which are particularly difficult for students of that language to master are likely to be those which represent ways
of configuring the world to which the student has not become habituated on
the basis of experience in his or her first language,
Following the translation task, the same group of fourth to seventh graders
was asked in Chinese Can Mrs. Wong read English books? Ninety-one
percent indicated that they interpreted the sentence as saying that she could
not, thereby giving evidence that they had grasped the counterfactual intent
of the statement. One month later the large majority of these subjects also
responded to an analogous single implicational version of Aus original story,
and 85% gave evidence that they grasped the counterfactual intent of that
single implication (Au, 1983, p. 179).
These subjects do not have command of counterfactual labels in English, as evidenced by their inability to provide correct translation, and they
are unlikely as a consequence to have lleen led to develop counterfactual
labels in Chinese, yet they grasp the count&actual intent of the statements;
therefore, Au argues that the absence of counterfactual labels does not affect
the processing of counterfactual thought. What Au has shown in this part of
her work is simply that Chinese-speakers, in the absence of specific counterfactual labels, do arrive at counterfactual interpretations in response to a
single negated premise followed by a simple implicational statement based
upon it-one of the fundamental premises of my work. In such a simple task,
as I have persistently argued, the effect of having or not having counterfactual
labels cannot be expected to emerge.
In sum, both of Aus attempts to bolster her position fail, for she once
again misses the point that the relationship I predicted between the availability of labels, on the one hand, and the direct and indirect processing of
counterfactual thought, on the other, can only be expected to emerge when
the testing materials are sufficiently complex to permit mode of processing
to make a difference, She cannot turn to subjects who are unable to handle
such complexity, or to test instruments which do not contain it, and hope to
provide a meaningful test of the predicted relationship.
I suggested (Bloom, 1981) that the color-naming studies, by attempting to
test the effects of language on thought in those areas where perceptual images
can most easily substitute for linguistic labels, had in fact attempted to test
these effects in just those areas where they would be least likely to emerge.
I, therefore, intentionally set out to investigate the impact of language on
thought in a more abstract area, such as that of counterfactual processing,

286

A.H. Bloom ,

where perceptual images are less likely to substitute for linguistic labels and
where, as a consequence, the impact of language on thought should be more
pronounced. On the presumption that she has undercut my work, Au concludes that there is no evidence for linguistic effects on thought even in that
area where, on my account, those effects can be expected to be most pronounced. She goes so far as to suggest in her last paragraph that someday
Who& observations will themselves, likewise, be shown to be the result of
a lack of sensitivity to the idiomatic qualities of the languages under investigation and leaves the reader with the impression that the Sapir-Wharf
Hypothesis and, with it, the question of linguistic impact on thought have
been finally put to rest.
By projecting so markedly narrow a vision of the potential range of effects
of linguistic labels on thought, and thereby constraining the possibilities we
are likely to imagine, Aus work performs its greatest disservice. It seems
likely in fact that the specific labels we have available affect our thoughts in
much more general ways than have been discussed in this article. When a
label is attached to a schema, we can use the label as a symbol to stand for
the schema and, thereby, create an external representation, written or oral
of that schema. This representation, in turn, acts to trigger directly in the
mind of the hearer or reader the schema with which it is associated, and, as
such, communication proceeds. However, thoughts which are structured in
terms of schemas that do not have labels must be translated into schemas that
do in order to permit external representation. And in that process of translation, is it not likely that those thoughts must be reconfigured to conform to
the structural parameters of the schemas whose labels we use to rcgresent
them? Furthermore, is it not likewise the case that, in the process of thinking
to ourselves, we first make use of mental representations to extract from the
collective associativity of our thoughts discrete, distilled mental perspectives
to which we can then react in order to further develop our ideas? And in this
process, do we not again have to translate those thoughts which are structured
in terms of schemas that do not have labels into schemas that do, in order to
permit internal representation of them? Thus, within both the internal and
the external domain, are not the specific schemas for which we have labels
likely to play influential roles in shaping the processing of our thoughts?
Moreover, turning from the impact of linguistic labels on the processing of
thought to the impact of such labels on the development of thought, is it not
likely that the specific lexical and grammatical labels of a natural language
act, in much the same way as the core vocabulary and symbols of a disciplinary language, to lead each next generation of speakers to develop many
specific schemas for thought which these speakers would be unlikely to develop in the absence of such labels, and which they require for participation

Response to Au

287

in their particular culture or discipline? Isnt it likely, for example, that our
exposure to the specific labels Nepal, deterrence!, law, GNP,
phoneme, operant conditioning, cognitive dissonance, prototype
and Whorfian Hypothesis has had an impact on shaping the ways in which
we have come to think, whether or not these labels themselves are necessarily
activatecr when we use the schemas they represent? The Sapir-Whorf
Hypothesis suggests a range of important and exciting possibilities. To put it
to rest at this point constitutes a premature and unimaginative response.
References
Au, T.K.-F. (1983) Chinese and English counterfactuals: The Sapir-Wharf hypothesis revisited. Cog.. 15.
155-187.
Bloom, A.H. (1981) The Linguistic Shaping of 7%0&t: A Study in the hpact of Language on Thinking in
Chino and the West. Hillsdale, NJ, Erlbaum Associates.

Reference Note
I. Bloom, A.H. (1984) The linguistic shaping of thought in Japan. Manuscript in preparation.