Mondorf (2002) Gender Differences in English Syntax

JEngL 30.2 Mondorf / Gender (June 2002) Differences in English Syntax

Gender Differences in English Syntax BRITTA MONDORF

Universität Paderborn, Germany

While the past three decades have witnessed an upsurge in research on gender and language, empirical investigations of gender differences in the area of syntax are still at a premium. The objective of this paper is to combine corpus-linguistic methodology with the theoretical framework provided by functional grammar in order to explore an area of marked gender differences in syntax (i.e., epistemic grounding strategies by means of adverbial clauses). The gender sensitivity of linguistic items that encode epistemic meaning has already been shown by Holmes (1984, 1986, 1990) for tag questions and hedges, as well as by Preisler (1986) and Coates (1987) for modal verbs. 1 The quantitative and qualitative in-depth analysis of syntactic variation in the use of four types of finite adverbial clauses according to gender in the LondonLund Corpus (LLC)2 not only introduces verifiable data on statistically significant differences but is also able to provide a more realistic point of departure for future theory building. This is more than that which is currently offered by less data-driven approaches. Moreover, a corpus-based approach is able to reconcile the apparently contradictory results of studies on gender-differentiated language use conducted so far. The semantic types investigated are causal, conditional, concessive, and purpose clauses. A careful breakdown of the data according to a range of factors, such as semantic type and position, allows us to discern the underlying motivation for the observed sex-differentiated use of adverbial clauses. The results will be analyzed with a view to several tenets postulated in expectation states theory.

Previous Research Although claims of sex differences in the use of subordination date at least to the beginning of the twentieth century, few studies have approached this question empirically. Sex differences with respect to subordinate clause usage were postulated by Jespersen (1922, 252) in his well-known quotation: A male period is often like a set of Chinese boxes, one within another, while a feminine period is like a set of pearls joined together on a string of ‘ands’ and Journal of English Linguistics, Vol. 30 / No. 2, June 2002 158-180 © 2002 Sage Publications

158

Downloaded from eng.sagepub.com at University of Wollongong on July 23, 2011

Mondorf / Gender Differences in English Syntax

159

similar words. . . . In learned terminology we may say that men are fond of hypotaxis and women of parataxis. Karin Aijmer (1986) has discussed adverbial clauses in terms of hedging, a phenomenon that (since Lakoff 1975) is one of the best-known characteristics differentiating female and male speech. Similarly, Brown and Levinson (1987, 146f.) have stressed that adverbial clauses can be used to hedge a main clause in order to signal negative politeness. In a diachronic study of gender difference in dramatic dialogue, Biber and Burges (2000, 35) analyzed ninety-five drama texts from A Representative Corpus of Historical English Registers (ARCHER) and observed “that female authors portrayed both female and male characters as being more involved and tentative than male authors.” Politeness, tentativeness, and hedging share the property of being expressed by linguistic devices that can signal epistemic meaning (i.e., the speakers’ commitment to the truth of the proposition expressed). The gender differences concerning position and semantic clause type that are identified in the present investigation can all be attributed to differences in the expression of epistemic meaning.

The Data The present analysis is based on the LLC of spoken British English, which comprises 100 texts exceeding a total of 500,000 words uttered by approximately 650 educated native speakers. This computer-readable corpus has been selected because, in addition to being large enough to give new insights into the structure investigated, it has the advantage of being prosodically coded.

Definition For present purposes, a grammatical unit is considered a finite adverbial clause if it 1. 2. 3.

is introduced by a subordinating conjunction, has a subject and finite verb, cannot function as subject or object of the main clause verb.

After automatically retrieving the major causal, conditional, purpose, and concessive subordinators in the corpus, each occurrence was manually edited in order to eliminate those cases to which the defining characteristics did not apply. This procedure provided approximately 4,500 finite adverbial clauses as the database for the present study.


160

JEngL 30.2 (June 2002)

Gender Differences in the Use of Finite Adverbial Clauses Statistically highly significant sex differences are revealed in an analysis of the correlation between finite adverbial clause usage and speaker sex in the LLC. A comparison of absolute differences is unrevealing since women’s and men’s overall contribution to the corpus measured in terms of the number of tokens differs. Men contributed 68 percent of all tokens to the corpus while the female contribution amounts to only 32 percent. In order to take account of these absolute differences, the expected value for each sex is computed. The expected value (EV) is the value that would be reached if the clause/token ratio had been proportionate to that of the other sex. The difference between actual and expected values then serves as a basis for comparison. The direction and extent of sex differentiation is depicted in the histogram in Figure 1. As the figure indicates, women’s proportion of adverbial clauses is higher than expected. This means that it is above the value that would have been expected if women had used the same proportion of adverbial clauses per token as men. The extent of sex differentiation is statistically highly significant. The level of statistical significance measured by the chi-square test is indicated at the bottom of the legend.3

Gender Differences According to Position and Semantic Clause Type The observation that women in the LLC use more finite adverbial clauses of the types analyzed is not in itself revealing. A further differentiation according to formal and functional criteria provides a clearer motivation for the encountered pattern. A strikingly consistent pattern emerges with respect to the positioning of adverbial clauses (see Figure 2). Statistical significance is ascertained for each position separately. A set of columns is labeled with its respective chi-square score only if gender differences reach the 5 percent level of statistical significance. The figure illustrates that no noteworthy differences emanate for preposed clauses. By contrast, the gender effect is statistically highly significant for postposed clauses and significant for clauses without main clause. The latter rubric comprises instances where an adverbial clause does not modify explicitly expressed material. The distribution according to position suggests the following: • Pre- and postposed adverbial clauses serve crucially different functions—a phenomenon that is largely uncontroversial to researchers in this field. • Female dominance in the use of finite adverbial clauses appears to be traceable to postposed adverbial clauses and those without main clause.



161

Adverbial Clauses 3500 3000

AV (F) EV (F)

2500

AV (M) EV (M)

2000 x2 = 42.97*** 1500 1000 500 0 Figure 1: Finite Adverbial Clauses According to Sex (n = 4,471). NOTE: AV = actual value; EV = expected value.

2

x = 71.68*** Adverbial Clauses 2000 AV (F)

1800

EV (F)

1600

AV (M)

1400

EV (M)

1200 1000 800 600 2

x = 8.01**

400 200 0 Preposed

Postposed

No Main Clause

Figure 2: Finite Adverbial Clauses According to Sex and Position (n = 4,462).

In this context, a comment on hedges by Janet Holmes (1984, 52) is of interest. She states that positioning is often relevant to the functional differentiation of hedges and gives the following example, in which, in most contexts, the initial positioning of I believe can be interpreted as expressing more certainty than final positioning.


162


(1) I believe the students are responsible for this. The students are responsible for this I believe. (Holmes 1984, 52) Whereas initial placement of I believe may intensify or strengthen the assertion, final placement typically attenuates the strength of the speaker’s commitment to the truth of the proposition. In both cases, the expression of epistemic modal meaning is involved—in one case indicating certainty and in the other uncertainty.

Causal Clauses The LLC contains 1,792 finite adverbial causal clauses introduced by such conjunctions as because and its variants cause and cos, as well as since, for, as, in that, as long as, inasmuch, and in case. Causal clauses produce gender effects that reflect the same pattern as the aggregate results for all adverbial clauses (i.e., there is a statistically highly significant female lead in the use of this clause type, whereas males use comparatively few causal clauses) (see Figure 3). With respect to positioning, the emerging pattern also resembles that of the aggregate figures for all clause types. While there is no discernible correlation between gender and preposed causal clauses, sex differences for postposed clauses are statistically highly significant, displaying the—by now familiar—female lead (see Figure 4). The parallels between causal clauses and the aggregate pattern for all adverbial clause types raise the question as to whether the aggregate results are conditioned by the causal clause findings. Consideration of the remaining three clause types (i.e., conditional, concessive and purpose clauses) will demonstrate that this is not the case.

Conditional Clauses Conditional clauses included in the analysis are those introduced by if, unless, when, whenever, in case, supposing, provided, providing, so long as, as long as, on condition that, assuming, before, in the case that, just so, once, and save.4 At first sight, the 2,222 conditional clauses in the LLC do not exhibit statistically significant gender marking. However, taking the position of the conditional clause into account, the distribution resembles that of causal clauses: no gender differences emerge for preposed clauses, but a strong female dominance is apparent for postposed conditionals and the gender effect is statistically significant (see Figure 5). The same female lead may be observed for conditionals with no main clauses. Thus, as for causal clauses, the position of conditional clauses strongly correlates with gender. Women prefer postposed clauses, while men use postpositions comparatively less often.


Mondorf / Gender Differences in English Syntax Causal Clauses

163

1400 1200

AV (F) EV (F)

1000

AV (M) EV (M)

800 2

x = 68.53*** 600 400 200 0

Figure 3: Finite Causal Clauses According to Sex (n = 1,792).

2

x = 66.87*** Causal Clauses 1200 AV (F) 1000

EV (F) AV (M)

800

EV (M)

600 400 200 0 Preposed

Postposed

No Main Clause

Figure 4: Finite Causal Clauses According to Sex and Position (n = 1,792).

Purpose Clauses The third adverbial clause type, which is introduced by so, if, in case, lest, for fear that, and in order that, displays the same pattern as causal clauses. Women use comparatively more purpose clauses compared to men, who use relatively few. Since all 126 occurrences of purpose clauses in the LLC are postposed, the total distribution is the same as for postposed clauses (see Figure 6).


164


Conditional 900 Clauses 800

AV (F)

700

EV (F)

x 2 = 5.71*

AV (M)

600

EV (M) 500 400 300 x 2 = 5.91*

200 100 0

Preposed

Postposed

No Main Clause

Figure 5: Finite Conditional Clauses According to Sex and Position (n = 2,222).

Purpose Clauses

90

AV (F)

80

EV (F) AV (M)

70 60

EV (M)

50 40

2

x = 18.77***

30 20 10 0 Figure 6: Finite Purpose Clauses According to Sex (n = 126).

Concessive Clauses The subordinators retrieved for the present analysis are even if, even when, although, though, whereas, while, and whilst, where these encode a primarily concessive function. Concessive clauses pattern very differently from the other three adverbial clause types. In contrast to causal, conditional, and purpose clauses, they are comparatively often used by men. This male preference is statistically highly significant (see Figure 7).


Mondorf / Gender Differences in English Syntax Concessive Clauses

165

250 AV (F) EV (F) 200

AV (M) EV (M)

150

2

x = 7.48**

100

50

0

Figure 7: Finite Concessive Clauses According to Sex (n = 322).

Concessive 140 Clauses 120

x 2 = 8.73**

AV (F)

100

EV (F) AV (M)

80

EV (M)

60 40 20 0

Preposed

Postposed

No Main Clause

Figure 8: Finite Concessive Clauses According to Sex and Position (n = 322).

The “deviant” distribution for concessive clauses raises the question of whether this sentence type also exhibits a different pattern as regards positioning. Figure 8 illustrates that this is indeed the case. The pattern for concessives differs in three respects from the pattern for causal, conditional, and purpose clauses. • There is no gender effect observable for postposed clauses. For the remaining three clause types, this position was characterized by frequent female usage.


166


• For the first time, a statistically highly significant sex difference for preposed clauses is apparent. • This clause type is the only one that is predominantly used by males rather than females.

Explanation of Gender Differences According to Position Female Preference for Postposed Clauses The finding that women use more postposed adverbial clauses in the LLC than men do has been observed throughout all four adverbial clause types. This finding is most noteworthy with respect to conditional clauses that are generally indiscriminate with respect to gender in the corpus as a whole. However, when differentiated according to position, the gender-neutral pattern disappears and a clearly differentiated pattern emerges for women and men, a pattern that is concealed by the aggregate figures. Females clearly favor postposed conditionals, whereas men favor preposed conditionals. The positional preference can be explained in terms of the information structure of finite adverbial clauses. The postposed clauses favored by women are mainly asserted rather than presupposed. Women score particularly high on postposed clauses produced under a separate intonation contour (cf. Mondorf 1996). Such clauses express lower commitment than presupposed information. Postposition has also been shown to be the typical location for clausal hedges. Thus, one of the reasons why women apparently use these final clauses is to epistemically modify the proposition expressed in the main clause. Postposition appears to be the default location for signaling limited commitment toward the truth of the proposition expressed in the main clause.

Male Preference for Preposed Clauses Men’s adverbial clauses in the LLC are far more often of the kind conveying presupposed information, thereby expressing high commitment toward the truth of the proposition expressed. The difference in information status of women’s and men’s subordinate clauses is reflected in the distribution of clauses as regards positioning. Presupposed information usually appears in the canonical given information slot (i.e., in prepositioning), whereas asserted information typically appears in final position. This is consistent with the finding that women are, comparatively, more likely to use postposed adverbial clauses, while men use mainly preposed clauses. Overall, then, men tend to use finite adverbial clauses to convey high commitment to the truth of the propositions expressed, whereas women use them to the opposite effect.



Causal

Purpose

causality asserted

Conditional

167

Concessive

causality hypothesized

causality denied

Figure 9: Continuum of Causality Relations. SOURCE: Based on Harris (1988).

Women

Causal

Purpose

causality asserted

Conditional

causality hypothesized

Concessive

Men causality denied

Figure 10: Continuum of Causality Relations Extended by a Gender Component.

In addition, the discourse function of preposed clauses has been described by Ford (1993) as text structuring rather than of a main clause modifying kind. Preposed clauses generally have a less localized scope and often serve to introduce new topics (cf. Ramsay 1987). This feature can also be related to status differences between the sexes, since topic control tends to be a privilege of those in higher status positions.

Explanation of Gender Differences According to Semantic Clause Type Harris’s Semantic Continuum Harris (1988, 71) has suggested a semantic continuum covering causal relations in causal, conditional, and concessive clauses. By adding purpose clauses to the spectrum—where the relation between the subordinate and the main clause also asserts causality rather than hypothesizing or denying it—the semantic continuum can be represented in Figure 9. The findings presented have shown that in comparison with men, women in the LLC use strikingly more causal and purpose clauses and fewer concessives. By contrast, conditional clauses proved to be indifferent to the gender factor. Thus, extending the continuum by a female and male pole presents the distribution given in Figure 10.


168


Male Preference for Concessive Clauses Concessive clauses have been discussed in terms of hedges (i.e., in terms of negative politeness, which takes the face needs of interlocutors into consideration). Preisler (1986) assumes that concessives facilitate a compromise. His analysis groups together concessive clauses that appear within the same intonation contour as the main clause as “tentativeness signals”—clauses used to mitigate the statement made in the main clause. Preisler predicted that such concessives would occur more frequently in women’s speech, a hypothesis that was not borne out by his data. I argue that concessives have exactly the opposite effect: they do not mitigate the main clause proposition but rather reinforce it. Thus, a concessive clause introduced by even if can signal that the main clause proposition holds even if there may be arguments against it. Such concessives cannot be analyzed in terms of politeness strategies or tentativeness. They rather invoke some kind of “infallibility clause,” which is why they are intensifying rather than mitigating. The following example uttered by a politician during a radio discussion illustrates this point: (2) [@] and . ^though I don’t know much about the :Cr\oydon C/ouncil# I’m ^sure they’re :wrong about th\at# . (LLC 5.1 TU 254090)5 The concessive clause belongs to the type that Preisler (1986)—on prosodic grounds—groups among tentativeness signals. However, although the speaker admits his knowledge to be rudimentary, he intensifies his claim that “they’re wrong.”

Female Preference for Causal and Purpose Clauses Günthner (1992) has reported how conversational partners can turn a conversation into a confrontation by not giving a reason or justification for a proposition. This supports the view that causal clauses can serve as attenuators for a given statement. By contrast, the omission or nonutterance of an expected causal clause can signal that there is nothing to add to the proposition, since providing a reason is regarded as unnecessary; the interlocutor is not considered important enough to deserve such effort. In the data analyzed, it has been shown that purpose clauses are predominantly used by women. Hopper and Thompson (1985, 174f.) emphasize that a purpose clause denotes an unrealized state of affairs. (3) I’ll just break one open so that you can see the rich mushroom filling. (LLC 10.11 TU 7380ff.) The clause provides motivation for the event. Since the verb of the purpose clause does not report any event but only an unrealized state of affairs, it scores low with



Presupposition Negative Assertion

Realis Assertion

169

Irrealis Assertion

Figure 11: Discourse Pragmatic Scale According to Givón (1984, 322).

respect to categoriality. This is reflected cross-linguistically by the fact that in many languages, this verb is an infinitive form or otherwise marked as irrealis. It normally lacks morphological indications of tense, aspect, and person. The subordinate clause proposition (i.e., “to see the filling”) encodes an irrealis component. Irrealis, in turn, is related to epistemic meaning. Irrealis assertions have been characterized in Givón’s (1984, 25) epistemic contract as open to challenge. They score comparatively high in the encoding of epistemic meaning. If something is expressed in the irrealis mode, the degree of commitment to the proposition is comparatively low. In his functional grammar, Givón (1984) has expressed this relation in the scale in Figure 11. The scale indicates that presupposed material is put on a par with negative assertions as regards their ability to express the speaker’s commitment. Next on the scale are realis assertions, which do not require “evidentiary support” (Givón 1984, 322) but convey lower commitment on the part of the speaker than presuppositions and negative assertions. Irrealis assertions are considered to be open to challenge. The scale illustrates that presupposed material ranks on an equal “commitment level” to negative assertions. Thus, a denial as it is implicitly expressed in concessive clauses encodes very strong commitment. It is an assertive strategy by speakers who, according to the epistemic contract, do not typically expect their statement to be challenged. Denial implies the negation of a presupposed belief, which is why it scores so high on the commitment scale. The reason for placing negative assertion on a par with presuppositions is that affirmative utterances are commonly used to inform the addressee of something new. By contrast, negative declaratives deny a proposition against the background of the addressee’s presumed belief in the proposition or familiarity with it (cf. Givón 1984, 324). (4) a. What’s new? b. Oh, my wife is pregnant. c. Oh, my wife is not pregnant.

affirmative negative

A likely answer to the negated sentence might be as follows:


170


(5) Wait a minute, I didn’t know she was supposed to be pregnant. (Based on Givón 1984, 323) Thus, the negation can be regarded as denying a presupposed belief. In the example, the negation “My wife is not pregnant” denies the presupposed assumption of the conversation partner that the wife of the speaker is pregnant. As Givón (1984, 324) puts it, Affirmative declarative: The hearer does not know and the speaker knows. Negative declarative: The hearer knows wrong and the speaker knows better. One of the functions of a negative declarative can therefore be to signal the addressee that the speaker does not share the belief in the corresponding affirmative. This is why negatives are considered less polite, contentious, unpleasant or downright threatening. It is one thing to tacitly add to a person’s knowledge on the implicit background of non-knowledge. It is another thing altogether to challenge a person’s already existing, strongly held (and oft strongly asserted) belief. (Givón 1984, 324fn.) It might be tempting to account for the female preference for causal and purpose clauses by falling back on the hypothesis that women, in general, are more tentative than men, and they therefore use more adverbial clauses to justify the propositions in their utterances. However, this begs the question of how to account for the multifunctionality of causal clauses (i.e., not all causal clauses have a justifying function). Likewise, one might be tempted to argue that males as a group are so assertive that they can afford to use concessive clauses that state a contrast toward what is conveyed in the main clause and might even provide an interlocutor with arguments that can be used to challenge the speaker. However, this would not explain why some semantic clause types are differentiated for gender while, for instance, conditional clauses are not, even though the importance of conditional clauses in signaling politeness is so pervasive in language that one type of conditional clause has been characterized as a conventional expression of politeness (cf. Quirk et al. 1985). Further analyses concerning the pragmatic function of those adverbial clauses that encode primarily epistemic meaning versus those that mainly convey propositional meaning show that for all four semantic clause types, females use signifi6 cantly more adverbial clauses that signal epistemic meaning than men. For space reasons, this analysis cannot be provided within the scope of this paper.



171

Epistemic Meaning and Gender Gender differences exist in the use of adverbial clauses as well as most of the linguistic units that have been discussed in terms of gender differences (e.g., hesitations, rising tone, tag questions, lexical hedges, intensifying particles, etc.), largely because these constructions encode epistemic meaning (i.e., the speaker’s commitment to the truth of the proposition expressed). This is consistent with Chafe’s (1987) observation that a subordinate clause can provide the epistemic background for the main clause proposition. The question remains as to why epistemic meaning appears so closely linked to gender. Givón (1990, 821) postulates an indirect inference relation between truth and power. He draws on the epistemic contract, a set of conventions that govern human communication. This contract links epistemic dimensions, such as knowledge and subjective certainty, to more sociopersonal dimensions, such as status or power. Hedges are a case in point, where speakers downtone the assertion expressed to indicate deference. Givón (1990, 822) holds that toning down is a hedge against the ‘possibility’ that the higher authority ‘might’ hold a contrary belief. Such ‘epistemic deference’ to power realities is a pervasive feature of many, perhaps all cultures. He assumes a continuum ranging from truth to power. truth

>

knowledge

>

certainty

>

status

>

power

Since gender effects are closely related to power and status, the continuum is easily applicable to gender variation. Truth and knowledge. The typical expectation if a speaker claims to express the truth of a proposition is that she or he has the relevant knowledge. If this is not true, speakers are expected to signal this, for instance, by means of a nondeclarative speech act or downtoning modifiers. Thus, when uttering (6) The train arrives at eight. the speaker is assumed to be more knowledgeable about arrival times than when uttering (7) Does the train arrive at eight? or I suppose the train arrives at eight.


172


I maintain that speakers who are aware they may not be expressing the truth usually indicate this “deficit” by means of certain disclaimers (i.e., expressions that bear epistemic significance). Among such disclaimers are certain tag question types and finite adverbial clauses expressing epistemic grounding. This does not mean that speakers cannot flout this principle by claiming something to be true, for instance, for rhetorical purposes. Conversely, speakers can use disclaimers even though they do know what they say to be true, in order to appear less knowledgeable and more polite. The present study provides empirical support for the claim that it is specifically women who use these expressions of epistemic modality in order to signal that what they are saying is not claimed to be the absolute truth. Whether these women really believe their propositions to be true or not is secondary in the present context because what is at stake is their intention of not appearing too knowledgeable. Knowledge and certainty. A speaker is only expected to express certainty about a proposition if he or she knows about it, unless of course expressions of certainty are deliberately used as a rhetoric device in persuading others. Certainty and status. The expression of certainty can be the privilege of highstatus speakers. The expression of uncertainty about a proposition can be a means of signaling modesty or deference. Thus, Syder and Pawley’s (1974) “modesty principle” states that speakers sometimes pretend to know less than they actually do, when their knowledge might positively reflect on their personal stature. This relates to a power maxim that states the following: In communicating to an interlocutor of higher status, one downgrades one’s own subjective certainty. (Givón 1990, 823) This factor is likely to contribute to the expression of politeness. As Lichtenberk (1995, 318) holds, Crosslinguistically, epistemic downtoning is a common feature of polite speech. Politeness is often effected by figuratively taking less space for oneself and leaving this space to others. Using disclaimers reduces the space for one’s own certainty or knowledge and increases that of others. This is where epistemic and interpersonal meaning are related. Epistemic meaning, in the form of mitigated expressions of commitment, can be signaled to achieve interpersonal ends. It therefore represents one strategy to signal interpersonal meaning. Another strategy of negative politeness is to soften disagreement. There is often a connection between downtoners and



173

negation. Commenting on combinations of negation plus yes/no question markers, conditionals, subjunctives, and modals, Givón (1990, 823) holds that its deference value might derive from the overlap between negation and irrealis, along the psychological dimension of subjective certainty. Status and power. If a speaker is in a powerful position, this often derives from the speaker’s status in society. Most frequently, power is associated with socioeconomic status but it need not be. The factors contributing to a person’s status are multifarious, and what conveys status on a person in one situation need not do so in another. The relation between status and expectations has been discussed under the heading of expectation states theory, which will play a crucial part in the explanation of sex differences introduced in the present study.

Expectation States Theory Expectation states theory accounts for the fact that what is at stake in the relation between certainty and knowledge in the above continuum is not necessarily the association of objective certainty or knowledge with status or power but expectations about speakers’ certainty or knowledge. Thus, a speaker in a high-status position might be expected to be more knowledgeable by deferent interlocutors. Conversely, a speaker in a comparatively low-status situation is likely to find that lack of power is frequently associated with expectations of uncertainty, lack of knowledge, or even the statement of what is not true. Whether the relation between truth and power is based on actual differences in knowledgeability is secondary in this context because the expectations themselves can result in different behaviors without ever being questioned or verified. This is where expectation states theory relates to the observed sex differences in syntax. Givón (1990, 807) remarks that the grammar of verbal manipulation shades into the grammar of deference, honorification and even the epistemics of certainty. Expectation states theory postulates seven central assumptions for interactants in task-oriented settings (cf. Meeker and Weitzel-O’Neill 1985, 387, 389): (A) Group members who are expected to do better at a given task receive and take more opportunities to make task contributions; they have more influence and prestige and receive more expressions of agreement and approval. (B) In the absence of information to the contrary, communication partners assign performance expectations on the basis of external status characteristics.


174


(C) Gender functions as a status characteristic, with females having lower status than males. (D) In the absence of information to the contrary, task contribution is accepted to raise the status of the contributor relative to the status of others. (E) Raising one’s status is legitimate and will be accepted for persons with high status (e.g., men) but not for those with low status (e.g., women). (F) Assumptions (D) and (E) can be canceled either if a lower status person’s contribution is motivated by helping others in the group rather than raising her or his own status or if the person holding lower external status legitimately has been assigned higher status (e.g., by having been appointed as leader). (G) These statements hold independently of performance expectations. As regards linguistic behavior, Meeker and Weitzel-O’Neill (1985, 390) hold that lower status speakers “do more agreeing and expressing of approval” than higher status speakers. In terms of assumption (F), these are legitimate contributions for female speakers. If the external status characteristic of sex can be made to appear irrelevant to the task, or if the particular women involved can be made to appear competent, performance expectations will not be affected by sex. If, at the same time, high rates of task behaviour are legitimated for women, sex differences should disappear. (Meeker and Weitzel-O’Neill 1985, 390) And Lockheed (1985, 410) points out that status expectation theory predicts that actors holding relatively higher expectation advantages will be deferred to. They will be given more opportunities to contribute to the group task, and their contributions will be more positively evaluated. A result of this admittedly unconscious deference is the emergence of an observable power and prestige order in the group. Berger’s (1980) theory has a lot to offer for the explanation of sex differences in language. Proponents of this theory hold that different expectations and beliefs about oneself and others lie at the basis of many behavioral differences. Individuals constantly evaluate themselves in relation to others, thereby creating “self-other performance expectations.” These expectations directly influence interactants’ behavior. The expectations are formed on the basis of so-called status characteristics of interactants. A status characteristic is any characteristic that is socially valued, is meaningful, and has differentially evaluated states which are associated directly or



175

indirectly with beliefs about task performance ability-‘performance expectations’. Examples of status characteristics are race, sex, education, or organizational office. (James and Drakich 1993, 286) Expectations are particularly influential among interactions of people who do not know each other well, so that no other factors can override status characteristics. According to James and Clarke (1993, 261), Unacquainted individuals are more likely than those who know each other well to rely on characteristics such as sex to define status/power relationships. These findings show that the sex-linked effect can be overriden or neutralized by other factors. Interactions need certain conditions in order for the gender effect to become operative. For instance, researchers such as Coates (1988) and Tannen (1992) argue that women, in general, feel more comfortable in private than in public settings. In private settings, sex as a status characteristic is least salient. The consequence of possessing high or low states of a status characteristic is multifarious. For instance, high-status speakers can be regarded as more competent or as performing better on a given task. The crucial consequence of such expectations is that a speaker who is believed to perform better is given more opportunity to do so, while a speaker who ranks low on a given status characteristic might not even try. It is important to note that status characteristics and their associated performance expectations are relational; that is, we do not speak of performance expectations for women, but rather we speak of performance expectations associated with women in relation to those performance expectations associated with men. (James and Drakich 1993, 286f.) This is why the status characteristics approach is able to account for situational factors. Because status characteristics involve relational expectations females do not in this conception carry sex-related characteristics around with them in every situation. (Berger and Zelditch 1977, 35)

Discussion The expression of epistemic meaning influences the creation and establishment of social relations. To signal a low degree of certainty can serve as a politeness strategy. Huebler (1983) has stressed that uncertainty or noncommitment reduces the risk of being falsified. It is no coincidence that linguistic units that express


176


epistemic meaning are also the ones that can signal interpersonal meaning in the form of politeness. Politeness can manifest itself in claiming less space for one’s own certainty, thereby increasing the space of interlocutors. Indeed, Robin Lakoff (1977, 147) related this point to women’s language many years ago: Since deference is often expressed as uncertainty, a deferential sentence is subject to interpretation as true intellectual indecision based on lack of knowledge. it (sic) is the plausibility of this interpretation that makes it difficult for women to be taken seriously in fields dominated by men. Another related aspect of epistemic meaning has been formulated in Givón’s (1990, 824) hazardous information principle, which says that the demonstration of too much knowledge or certainty can be socially destabilizing. Expectation states theory predicts that lower status speakers who formulate their proposition in an environment that expresses uncertainty merely fulfill expectations and hence do not run the risk of appearing to challenge the status hierarchy. Adverbial clauses encoding epistemic meaning are well suited for this function. By contrast, linguistic means that express a high degree of commitment are generally the privilege of higher status speakers. In a complex sentence, only the asserted part can be negated, since the presupposed part is outside the scope of negation. This aspect can be easily linked with the male preference for preposed adverbial clauses. Preposed clauses are in the default position for presupposed material. The male preference for concessive clauses can be similarly explained. In terms of the degree of commitment, negated material is on a par with presuppositions since negated declaratives in the new information slot state a proposition against the belief of an interlocutor.

Summary This investigation of the semantics and positioning of finite adverbial clauses has provided a consistent picture. On the semantic dimension, clause types that signal the lowest commitment to the truth of the proposition expressed (i.e., causal and purpose clauses) were favored to a statistically significant extent by women. Conversely, in the men’s speech investigated, the semantic clause type that typically expresses the highest degree of speaker commitment by means of denied assertions (i.e., the concessive clause) was significantly more frequent. In the data analyzed, women’s and men’s speech also differed with respect to the positioning of adverbial clauses: preposed adverbial clauses were generally preferred by men, whereas postposed ones are the marked domain of women. These positional preferences have been shown to be explicable with a view to information management and epistemic grounding.



177

The findings provided in the present study contribute to our understanding of the varied ways in which human beings make use of language. The multifarious functions served by syntactic devices such as adverbial clauses emphasize that any approach that neglects the intricate semantic, pragmatic, and cognitive aspects involved is unable to cope with the sex-differentiated variation described in the present study. Therefore, the present investigation makes a case for the inclusion of functional aspects in linguistic analysis on each language level. Coates (1987, 114) emphasizes a prominent aspect not acknowledged by views assuming that language can be accounted for with respect to its transactional function: One of the crucial differences between computer ‘language’ and human language . . . is that computers have no attitude at all towards the propositions they utter. One consequence of the present investigation for “Building HAL” (cf. Williams 1981, 27ff.), the robot in 2001, would thus be the assignment of gender identities for Ms. and Mr. Hal, manifested in the programming component for the expression of epistemic meaning. Ms. Hal would need to use considerably more causal and purpose clauses than Mr. Hal. She would prefer postposed clauses and adverbial clauses with an epistemic meaning, and both would need to know when to background their respective gender identities in order to allow other factors to become salient.

Notes 1. The question of whether it is theoretically and methodologically justified to extend the concept of variation to syntax has been widely debated (cf. Lavandera 1978; Dines 1980; Jacobson 1982; Romaine 1984; Winford 1984; Cheshire 1987). In the present analysis, I follow Lavandera (1978, 181), who holds that a syntactic variable should only be termed as such if the following conditions are met: first, it must carry some nonreferential information, manifested in social, stylistic, or other (e.g., pragmatic) significance; second, it must be definable in terms of quantifiable covariation reflected in frequency relationships. 2. For a detailed description of the corpus, see Svartvik (1992). 3. Statistical significance for alpha = 0.05 and 1 degree of freedom (χ2(1;95 percent)) is indicated by one asterisk, while highly statistically significant results (χ2(1;99 percent)) are indicated by two asterisks and very highly significant results by three asterisks (χ2(1;99.9 percent)). In other words, the likelihood that a sex difference is accidental is 5 percent for those results with one asterisk, 1 percent for those with two asterisks, and only 0.1 percent for those with three asterisks.


178


4. Only those occurrences were included in which the subordinator encodes a primarily conditional relation. In the case of in case, each example is included in only one clausal category, determined by its primary function: causal, conditional, or purpose. 5. Examples from the London-Lund Corpus (LLC) are cited with accompanying reference to the corpus section and the tone unit number. 6. For a pilot analysis of causal and conditional clauses, see Mondorf (1996).

References Aijmer, Karin. 1986. Discourse Variation and Hedging. In Corpus Linguistics II: New Studies in the Analysis and Exploitation of Computer Corpora, edited by Jan Aarts and Willem Meijs, 1-18. Amsterdam: Rodopi. Berger, Joseph and Morris Zelditch Jr. 1977. Status Characteristics and Social Interaction. New York: Elsevier. Berger, Joseph. 1980. Status Organization Processes. Annual Review of Sociology 6:479-508. Biber, Douglas, and Jená Burges. 2000. Historical Change in the Language Use of Women and Men: Gender Differences in Dramatic Dialogue. Journal of English Linguistics 28 (1): 21-37. Brown, Penelope, and Stephen Levinson. 1987. Politeness: Some Universals in Language Usage. Cambridge, UK: Cambridge University Press. Chafe, Wallace. 1987. Cognitive Constraints on Information Flow. In Coherence and Grounding in Discourse, edited by Russell Tomlin, 21-51. Amsterdam: Benjamins. Cheshire, Jenny. 1987. Syntactic Variation, the Linguistic Variable, and Sociolinguistic Theory. Linguistics 25 (1): 257-82. Coates, Jennifer. 1987. Epistemic Modality and Spoken Discourse. Transactions of the Philological Society 85:110-31. . 1988. Gossip Revisited: Language in All-Female Groups. In Women in Their Speech Communities: New Perspectives on Language and Sex, edited by Deborah Cameron and Jennifer Coates, 94-122. London: Longman. Dines, Elizabeth R. 1980. Variation in Discourse “and Stuff Like That.” Language in Society 9:13-31. Ford, Cecilia E. 1993. Grammar in Interaction: Adverbial Clauses in American English Conversation. Cambridge, UK: Cambridge University Press. Givón, Talmy. 1984. Syntax: A Functional-Typological Introduction, I. Amsterdam: Benjamins. . 1990. Syntax: A Functional-Typological Introduction, II. Amsterdam: Benjamins. Günthner, Susanne. 1992. Die interaktive Konstruktion von Geschlechterrollen, kulturellen Identitäten und institutioneller Dominanz: Sprechstundengespräche



179

zwischen Deutschen und Chinesen/innen. In Die Geschlechter im Gespräch, edited by Susanne Günthner and Helga Kotthoff, 91-125. Stuttgart: Metzler. Harris, Martin B. 1988. Concessive Clauses in English and Romance. In Clause Combining in Grammar and Discourse, edited by John Haiman and Sandra A. Thompson, 71-99. Amsterdam: Benjamins. Holmes, Janet. 1984. Hedging your Bets and Sitting on the Fence: Some Evidence for Hedges as Support Structures. Te Reo 27:47-62. . 1986. Functions of you know in Women and Men’s Speech. Language in Society 15 (1): 1-21. . 1990. Hedges and Boosters in Women’s and Men’s Speech. Language and Communication 10 (3): 185-205. Hopper, Paul J., and Sandra A. Thompson. 1985. The Iconicity of the Universal Categories “Noun” and “Verbs.” In Iconicity in Syntax: Proceedings of a Symposium on Iconicity in Syntax, Stanford, June 24-26, 1983, edited by John Haiman, 151-83. Amsterdam: Benjamins. Huebler, Axel. 1983. Understatements and Hedges in English. Amsterdam: Benjamins. Jacobson, Sven. 1982. Syntactic Variation: Theory and Practice. Umea, Sweden: Umea University, Department of English. James, Deborah, and Sandra Clarke. 1993. Women, Men, and Interruptions: A Critical Review. In Gender and Conversational Interaction, edited by Deborah Tannen, 231-80. Oxford, UK: Oxford University Press. James, Deborah, and Janice Drakich. 1993. Understanding Gender Differences in Amount of Talk: A Critical Review of Research. In Gender and Conversational Interaction, edited by Deborah Tannen, 281-312. Oxford, UK: Oxford University Press. Jespersen, Otto. 1922. Language, Its Nature, Development and Origin. London: Allen & Unwin. Lakoff, Robin. 1975. Language and Woman’s Place. New York: Harper Colophon. . 1977. Women’s Language. In Language and Style, edited by Douglas Butturff and Edward L. Epstein, 139-58. Akron, OH: L & S Books. Lavandera, Beatriz R. 1978. Where Does the Sociolinguistic Variable Stop? Language in Society 7:171-83. Lichtenberk, Frantisek. 1995. In Modality in Grammar and Discourse, edited by Joan Bybee and Suzanne Fleischman, 293-327. Amsterdam: Benjamins. Lockheed, Marlaine E. 1985. Sex and Social Influence: A Meta-Analysis Guided by Theory. In Status, Rewards and Influence: How Expectations Organize Behavior, edited by Joseph Berger and Morris Zelditch Jr., 406-29. San Francisco: Jossey-Bass. Meeker, Barbara F., and Patricia A. Weitzel-O’Neill. 1985. Sex Roles and Interpersonal Behaviour in Task-Oriented Groups. In Status, Rewards and Influence: How Expectations Organize Behavior, edited by Joseph Berger and Morris Zelditch Jr., 379-405. San Francisco: Jossey-Bass.


180


Mondorf, Britta. 1996. Why Women Use More Postposed Adverbial Clauses. Working Papers in Language, Gender and Sexism 6 (2): 5-29. Preisler, Bent. 1986. Linguistic Sex Roles in Conversation: Social Variation in the Expression of Tentativeness in English. Berlin: Mouton de Gruyter. Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech, and Jan Svartik. 1985. A Comprehensive Grammar of the English Language. London: Longman. Ramsay, Violeta. 1987. The Functional Distribution of Preposed and Postposed “if” and “when” Clauses in Written Narrative. In Coherence and Grounding in Discourse, edited by Russel S. Tomlin, 383-408. Amsterdam: Benjamins. Romaine, Suzanne. 1984. On the Problem of Syntactic Variation and Pragmatic Meaning in Sociolinguistic Theory. Folia Linguistica 18:409-37. Svartvik, Jan. 1992. The London-Lund Corpus of Spoken English: User’s Manual. Lund: Department of English, Lund University. Syder, Frances Hodgetts, and Andrew Pawley. 1974. The Reduction Principle in Conversation. Auckland, New Zealand: Anthropology Department, Auckland University. Tannen, Deborah. 1992. That’s Not What I Meant! How Conversational Style Makes or Breaks Relationship. New York: Ballantine. Williams, Joseph M. 1981. The English Language as Use-Governed Behavior. In Style and Variables in English, edited by Timothy Shopen and Joseph M. Williams, 27-60. Cambridge, MA: Winthrop. Winford, Donald. 1984. The Linguistic Variable and Syntactic Variation in Creole Continua. Lingua 62:267-88.


Mondorf (2002) Gender Differences in English Syntax

Recommend Documents