Professional Documents
Culture Documents
Definiteness and Voice in The Interpretation of Active and Passive Sentences
Definiteness and Voice in The Interpretation of Active and Passive Sentences
To cite this article: Michel Hupet & Brigitte Le Bouedec (1975) Definiteness and voice in the
interpretation of active and passive sentences, Quarterly Journal of Experimental Psychology,
27:2, 323-330, DOI: 10.1080/14640747508400491
Article views: 44
AND
Downloaded by [Athabasca University] at 11:23 12 June 2016
BRIGITTE LE BOUEDEC
Department of Experimental Psychology, University of Angers, France
I n Experiment I, subjects were given active and passive sentences where the
definiteness of nominals is varied and about which uncertainty was expressed,
e.g. “I thought that the policeman had been injured by a gangster, but I was mis-
taken,infact.. .”. Theirtaskwastodecidewhatwasthetargetof themistakebycom-
pleting freely the sentence in such a way that the account would be corrected.
When the nominals are differentially determined (a-the, the-a), the pattern of
responses indicates that, for both active and passive, subjects were more likely to
conclude that it was the non-definitely marked nominal that had been involved in
the mistake rather than the definitely marked one. When both nominals are
similarly determined (a-a, the-the), subjects were more likely to indicate the
event itself as having been involved in the mistake. This supports the hypothesis
that subjects are able to utilize definiteness to determine the relation between pre-
supposed and assertional information. In Experiment 11, subjects were given
paired active and passive sentences with differentially determined nominals and
were required to choose between the two syntactic forms of each pair the one they
would prefer to use. The pattern of responses clearly shows that subjects were
more likely to choose the voice allowing the hypothesized preferential order
“Definitely marked grammatical subject-non-definitely marked grammatical
object”. Passive transformational rule is thus interpreted as a particular case of a
more general rule specifying the ways in which what is made known (comment)
is nested on to what is already assumed to be the case (topic).
Introduction
Since it received a formal expression in the generative transformational grammar,
the very close affinity between the active and the passive has proved to be a source
of some trouble to some psychologists, for “at first glance passives seem to have no
purpose” (Wales and Marshall, 1966, p. 77). Indeed, if active and passive can be
regarded as essentially redundant or “basically” synonymous, then “why is there a
need in the language for the further complication of passive constructions?”
(Grieve and Wales, 1973, p. 173). However, considering that an essential psycho-
323
324 M. HUPET AND B. LE BOUEDEC
linguistic distinction between the two voices may rest in their relative usage,
attention has been turned toward asking how passives function in language use.
Since the main structural difference between the active and the passive voice is in
the position of the logical subject and logical object, psychological studies have con-
centrated on the communicative function of this word-order difference. Attention
has been first focused on the importance of the grammatical subject in the beginning
of the sentence (for a review, see Anisfeld and Klenbort, 1973). T h e main con-
clusions were (a) that the grammatical subject is more “important” and carries
greater emphasis than the grammatical object, and (b) that the difference between
the grammatical subject and object in this regard is greater for the passive than for
the active. It has thus been claimed that the passive is more emphatic than the
active, the former being used to indicate that importance is attached to that entity
which occupies initial nominal position in the sentence.
Downloaded by [Athabasca University] at 11:23 12 June 2016
N , and the event in questions supplied for sentences like a N + V + a N (Grieve and
Wales, 1973). Since these results apply both to short passives and to active sen-
tences, Grieve and Wales concluded that the important entity of an utterance is not
solely determined by voice and word-order, but is more closely concerned with
definiteness which is supposed to relate to a feature of discourse they termed
topicalization.
T h e following two experiments attempted to test the usefulness of this account in
terms of topicalization. Specifically, employing active and full passive construc-
tions, the incidence of definiteness on the determination of the perspective structure
of a sentence was examined.
Experiment I
Consider the following distinction between topic and comment of sentences.
Downloaded by [Athabasca University] at 11:23 12 June 2016
Topic of a sentence may be defined as that part of the sentence which states what
the sentence is about, i.e. which refers to what the speaker is talking about, and
comment as the rest of the sentence which provides new information about the
topic. As Grieve and Wales (1973, p. 175) make clear, it can be assumed that in
discourse the speaker presupposes that the hearer has knowledge as to topic and
ignorance concerning comment, “for when this assumption is not met, communica-
tion fails in the former case and is redundant in the latter”. Furthermore, since
definitely marked nominals seem to require presupposition of previous knowledge,
it can be hypothesized that in strings with two differentially determined nominals,
presupposed information is indicated by that nominal which is definitely marked
while asserted new information is indicated by the remaining non-definitely
marked nominal.
T h e present investigation of this hypothesis is based on the fact that when a pro-
position is negated or doubted, these judgments are interpreted as applying to the
assertional focus rather than to the presupposition (see Langendoen and Savin,
1971; Klenbort and Anisfeld, 1974). This phenomenon was used to test the
hypothesis that in the interpretation of the passive as well as of the active, definite-
ness is used as a cue for distinguishing between presupposed and asserted informa-
tion. Here, the subjects were given active and passive sentences about which un-
certainty was expressed, and their task was to decide what was the target of the
negative expression ; but contrary to Klenbort and Anisfeld’s procedure which
restricted the choice to two alternatives, the subjects were asked to complete freely
the sentence in such a way that the account would be corrected. As an example,
consider the sentences (I) “I thought that the policeman had been injured by a
gangster, but I was mistaken. In fact . . .”, and (2) “I thought that a gangster had
injured the policeman, but I was mistaken. I n fact. . .”. Here the sentence is
embedded in the context “I thought. . . but I was mistaken. I n fact. . .” which
denies some part of the information in the sentence. But which part? Our hypo-
thesis was that the non-definitely marked nominal (“a gangster”) is more likely to be
interpreted as having been involved in the mistake than is the definitely marked one
(“the policeman”), regardless of their position in the surface structure. I n other
words, it was predicted that subjects are more likely to conclude from (I) as well as
from (2) that the policeman had been injured by someone else than to conclude that
326 M. HUPET AND B. LE BOUEDEC
someone else had been injured by a gangster. When we consider sentences where
neither nominal is definitely marked: (3) “. . . a policeman had been injured by a
gangster. . .”, and (4)“. . . a gangster had injured a policeman . , .”, it may be
hypothesized that it is the event, itself, which will be interpreted as being involved
in the mistake, for knowledge of neither nominal has been presupposed in the
listener, the assertional focus being thus distributed over the entire sentence. In
other words, it was predicted that subjects will be more likely to conclude from (3)
and (4)that there was no injured man at all. When we turn to sentences where
both nominals are definitely marked: ( 5 ) “. . . the policeman had been injured by
the gangster. . .”, and (6) “. . . the gangster had injured the policeman. . .”,
definiteness criterion offers no firm predictions since it suggests that the speaker of
( 5 ) and (6) has presupposed in his listener knowledge of both definitely marked
.nominals.
Downloaded by [Athabasca University] at 11:23 12 June 2016
Method
Subjects
Since the experiment was conducted in French, 42 naturally French-speaking under-
graduates in the age range of 17-26 years participated as subjects.
Passive
( 5 ) The N l + V + t h e Nz 2 23 I3 (38) E ( X z : 16.13)t
(6) The N1 + V + a Nz 4 6 31 (41) Nz (x2: 32’35)t
(7) A N l + V + t h e Na 24 5 I0 (39) N1 ( X 2 :14.06)t
+
Downloaded by [Athabasca University] at 11:23 12 June 2016
Experiment I1
Results of Experiment I clearly show that word order is not the only determinant
of the perspective structure of active or passive sentences. Rather, it is apparent
that the subjects were able to utilize other aspects of surface structure such as
definiteness to determine the relation between presupposed and assertional in-
formation. However, while these results are of some interest, it remains to be
seen how this feature should be related to the topic-comment distinction. Al-
though this distinction is somewhat different from what has been called the gram-
matical subject-predicate distinction, and is also different from the notion of
logical subject and predicate (Hornby, 1971), it has been suggested by several
writers that temporal order plays a significant role in the determination of the topic-
comment relation. The general principle here is that the topic of active and
passive sentences precedes the comment (Chafe, 1970). If topic is indicated by
Downloaded by [Athabasca University] at 11:23 12 June 2016
that nominal which is definitely marked when the nominals are differentially
determined, it could then be hypothesized that the subjects will see that the
temporal order “Definitely marked grammatical subject-non-definitely marked
grammatical object’’ is preserved. If such is the case, it may be predicted that
definiteness will affect the voice in which sentences will be produced.
Method
Subjects
Forty naturally French-speaking undergraduates in the age range of I 6-29 years partici-
pated as subjects.
that this order provides discourse unity by placing the theme or presupposed in-
formation in grammatical subject position, whilst at the same time highlights the
comment or assertional information by placing it in the grammatical object position
at the end of the sentence. Such a finding, indicating a decided effect of the
Downloaded by [Athabasca University] at 11:23 12 June 2016
definiteness manipulation upon the readiness to perform active and passive forms,
corroborates the Tannenbaum and Williams’s rationale (1968) that while these two
forms may be linguistically treated as being somewhat redundant, there is a func-
tional distinction between them arising from different demand characteristics of the
encoding situation. In this perspective, sentence types 3 and 7 used in Experiment
I should be regarded as unnatural sentences since they were violating the prefer-
ential order that is highlighted by the present test. This illuminates a possible
weakness of the experiments reported by Grieve and Wales (1973), since such un-
natural sentences (e.g. “a toy was stolen by the boy”, “a taxi bumped the lorry”)
were used to test the hypothesis that the important entity of an utterance is not
solely indicated by voice and word order. Without such sentences in the material
used, their experiments would have lent considerable support to the voice word-
order hypothesis that they considered could not accommodate the data, as the sub-
jects selected the definitely marked nominal to indicate the important entity of
sentences in which the definiteness of nominals was varied. It could be argued
(Grieve, personal communication) that these sentences are perfectly natural given
an appropriate context, e.g. Q. “What was stolen by the boy?” A. “A toy”, or Q.
“Was it a bus that bumped the lorry?” A. “No, a taxi”, with it being understood
that the entire sentence frameworks would read, “A toy was stolen by the boy”,
and “NO, a taxi bumped the lorry”. It may, however, just as easily be argued that
the inferred sentences are, “He stole a toy”, and “No, it was bumped by a taxi”
which retain the preferential order highlighted in our Experiment 11. Thus, as the
sentence must be inferred by the recipient and not the responder, it is meaningless
to talk of “appropriate context” unless fully complete and grammatical replies are
given.
After all, this sort of puzzle really seems to proceed from the misapprehension
that there must be an entity which is more important than another, the notion of
i<-importance’’ remaining fairly mysterious and very much dependent upon the
experimenter’s own conception of the term. In fact, the results reported here
suggest that this notion is a misleading guide to linguistic performance and should
be reviewed in the perspective of the topic-comment articulation and its various
aspects in the surface syntactic structure. If the phenomena of this articulation
are rendered in the semantic representations of sentences by a linear ordering of the
330 M. HUPET AND B. LE BOUEDEC
lexical units (Segall, I972), the transformational rule for passivization would then
be formulated so that it cannot apply in a sentence of which the first participant is
included in its topic. In that case, since such a rule can account for many other
inversion forms, the passive transformation rule appears to be only a particular case
of a more general rule specifying the ways in which what is made known is nested
on to what is already assumed to be the case.
This research was supported by grants from the National Foundation of Science of
Belgium to the first author. The careful reading of an earlier draft by M. Anisfeld and R.
Grieve is gratefully acknowledged.
References
ANISFELD, M. and KLENBORT, I. (1973). On the functions of structural paraphrase: the
Downloaded by [Athabasca University] at 11:23 12 June 2016