Gender and Generics: What makes a gender-specific generic sentence true? Alessia Passanisi (alessia.passanisi@unikore.it) Faculty of Human and Social Sciences, Kore University, Cittadella Universitaria - 94100 Enna (EN) James A. Hampton (hampton@city.ac.uk) Department of Psychology, City University, Northampton Square London, EC1V OHB UK Abstract instead of the sentence meaning “everything that is a duck Generic statements assert default properties of a kind. They lays eggs,” it means something like “a relevant fact to know reflect the relevant features of our concepts and are about ducks is that some lay eggs.” Quantifying the considered by people generally true of the entire class despite sentence with “all” may reduce the likelihood that people the existence of counterexamples (e.g. birds fly). We report will judge it to be true, but it does not trigger extensional three experiments which explore the factors that lead to the thinking to any great extent. The truth of generic sentences acceptance of generic statements. In particular we examine depends not on the absence of counterexamples but on what whether properties that relate to gender differences (lions is considered a relevant fact about the kind. For example, have manes, or ducks lay eggs) are more likely to be accepted than matched statements that refer to an arbitrary subclass although “Canadians are right-handed” is true of a majority unrelated to gender. Experiments 1 and 2 found surprisingly of Canadians, it is not judged as being generically true that gender-specific properties were less likely to be accepted (Leslie et al., 2011). than the neutral control properties. Experiment 3 showed by The present research set out to examine the factors affecting contrast that gender-specific properties are more acceptable the acceptability of minority characteristic generic when they relate to reproduction than to appearance-based statements relating to gender. All of the items in Leslie et gender differences. It is argued that reproductive properties al.’s set of minority characteristics used properties that were are more easily interpreted as referring to a kind rather than to the set of individual members. just true of one gender of an animal kind. For example there were sentences such as “lions have manes” or “insects lay Keywords: generics, gender-specific, concepts eggs”. Our research question was whether the acceptability of such sentences (even in universally quantified form) is Introduction the result of their association with gender. For example, When we describe the nature of the world, we typically use there may be some implicit pragmatic understanding that a “generic” sentences. These are unquantified statements gender modifier (male or female) is intended by the speaker. making general statements whose truth is resistant to When someone asserts “sheep give milk”, it could be argued counterexamples (Carlson & Pelletier, 1995). Recent that the hearer assumes that the speaker intends “sheep” to interest in generics has focused, for example, on the way in refer only to the females. Accordingly in Experiments 1 and which children understand gender-based statements 2 (which are close replications) we compared two (Cimpian & Markman, 2011) and how generic statements conditions, one in which a generic property was true of just serve to reinforce stereotypes and belief in the essential one gender of a fictitious animal, and one in which it was nature of social categories (Leslie, 2014). true of just one of two subtypes, unrelated to gender. We The logic of generic sentences and their truth evaluation has predicted that the gender-related generics would be more proven to be a highly complex issue, and is still the subject likely to be accepted as true than the gender-neutral of much debate (Greenberg, 2003; Leslie, 2014; generics. Liebersman, 2011). Our aim in the present paper is to explore one particular empirical question concerning Experiment 1 statements that describe “minority characteristics” true of The first experiment was designed to test whether minority less than 50% of the members of a kind. For example, the characteristics would be accepted as true more readily if sentences “Ducks lay eggs” and “Mosquitoes carry malaria” they relate to one particular gender rather than to a non- are typical generics. These sentences strike us as clearly gender based variety within a species. We hypothesised that true, even if we realize that only adult female ducks lay eggs people interpret the generic “ducks lay eggs” as applying to and only a small proportion of mosquitoes carry malaria. a subset of the kind, in particular just to the females. Hence Leslie, Khemlani, and Glucksberg (2011) established that they would accept a minority characteristic generic sentence generic statements about “minority characteristics” can still when it was true of females, but not males, but would reject be considered true when they are given universal it if it was true of just one half of a species, but regardless of quantification, as in the sentence “All ducks lay eggs”, gender. which is only true of a minority of ducks. It is as though 158 Method 3) It lives in forests (True) Participants. Fifty students at “Kore University” of Enna 4) It grows spots on its wings in spring (Generic-female) (Italy), participated voluntarily, 25 in each group. 5) It has crest on its head (Generic-male) 6) It eats nuts and seeds (False) 7) It has a yellow tail (False) 8) It eats small fishes (True) 9) It has a sharp beak (Generic-male) 10) It has a cooing call (Generic-female) The true sentences matched information given about the whole species, while the generic sentences were only true of one of the pictured creatures. Two booklets were created, one for each Condition. The order of the stories within the booklets was randomized. The task was translated into Italian by the first author (e.g. 8) Si nutre di piccoli pesci). Figure 1: Rattle Bird, Male (or Type A) on the left and Design and Procedure. Participants were randomly divided Female (or Type B) on the right, in Experiments 1 & 2. into 2 groups, each receiving one of the 2 booklets (either the gender or the neutral) with the four sets of stories. The Materials. Each booklet contained four sets of illustrated first page of each booklet contained a cover sheet with the descriptions of four fictional dimorphic species of creatures instructions “This experiment is simple and short. You will (Tabbie Toad, Rattle Bird, Crabby Beatle, Cranky Fish). In be shown a couple of pictures of fictional creatures, and a each story, a picture and description was given of two text that describes them. Then you will be asked to say different types, and of the species in general, which was whether a number of sentences are true or false for the referred to with a Definite Singular phrase. In the Gender species, based on the information you have been given”. group, the two types were labeled as male and female, while Participants circled one of 2 response options (true or false), in the Neutral group they were labeled as two sub-varieties printed to the right of each sentence. with both male and female members. For example, the gender-based story for the Rattle Bird was as follows (but in Italian): The Rattle Bird comes in two forms corresponding to the male and the female of the species. This is the male Rattle Bird (left). This is the female Rattle Bird (right). In spring, the female Rattle Bird grows spots on its wings. [Accompanying picture]. These Rattle Birds are found in France. The male has a sharp beak, and a crest on its head, whereas the female doesn’t have either. The female has a coo-ing call, but the male is silent. The Rattle Bird lives in forests and dense woodland and is related to the dove; it only eats worms, beetles and small fishes that can be found in small lakes and rivers. Figure 2: Distribution of responses across conditions in The gender-neutral version of the story was similar but Experiment 1 began: Results and discussion. Responses to True and False The Rattle Bird comes in two closely related forms with statements were strongly positive and negative as predicted. some minor differences. Both male and the female of the Interest centred on the acceptance of the generic test species can have either appearance, and the two forms, statements in each condition. Each participant saw four which are equally common, interbreed freely. This is the species and judged four generic statements about each, brown form of the Rattle Bird (left). This is the yellow form giving a total possible number of “true” judgments of 16. of the Rattle Bird (right) Figure 1 shows the distribution of number of true judgments to generics by condition (gender or neutral). The first point Each text was followed by 9 sentences: 4 generic (2 each for to note is that the distribution is non-normal. Both male and for female in the gender condition), 3 true and 3 conditions showed considerable variation between false. For instance, for the noun “Rattle Bird” the following individuals, with 13 of the 25 participants in the Gender 10 sentences were constructed: condition choosing to accept either all (6) or none (7) of the 1) It is only found in Asia (False) statements. The Neutral condition showed a more uniform 2) It is related to the dove (True) distribution, with a slightly greater preponderance of 159 participants accepting the majority of the generics (16 out of happier to allow a generic to be true of only one kind of 25, compared to 14 out of 25 for the gender condition). creature when it was NOT associated with a male/female Overall, 52% of generics in the Gender condition were difference. considered true, while 66% of the neutral generics were estimated true. Thus, the amount of agreement with the generic statements was in fact greater for the neutral stories than for the gender based stories. Because of the distributions, a median-split chi-square test was used to test for differences in frequencies across conditions, and showed no significant difference (χ2 = 0.3, p >,5). The results therefore leant no support to our hypothesis that minority characteristics would be better supported when they were specifically related to one gender. To test the robustness of this result we decided to replicate the study while making one change to the materials. In Experiment 1 we used the Definite Singular form for describing the kinds and an anaphoric singular sentence to express the generic statements. Previous research (e.g. Khemlani et al., 2007; Figure 3: Distribution of responses in Experiment 2 2012; Leslie et al., 2011) used bare plurals in their experiments (i.e. ducks lay eggs) and found high rates of Nevertheless, changing the noun from singular into bare acceptance (e.g. 89% of gender-based minority plural, and repeating the noun in the sentences also had an characteristics were accepted as true in Khemlani et al, apparent impact on the results. While acceptability was 2007). We therefore sought to replicate the results of much the same in the neutral case, acceptability of a gender- Experiment 1 with the same materials and design, but specific characteristic was much lower with the bare plural changing the definite singular phrase (The Rattle Bird) to noun placed at the head of each sentence. It is possible that the bare plural form (Rattle Birds). the definite singular form “The Rattle Bird as a species” In addition, to be sure that the scope of the sentences was encourages interpretation of the sentences as referring to the understood, we repeated the bare plural noun phrase at the relevant gender only, so making them more acceptable (but start of each sentence (e.g. Rattle birds are only found in still no more acceptable that a gender-neutral distinction). Asia). On the other hand the bare plural “Rattle Birds” clearly refers to the whole set of individuals, and so the fact that the Experiment 2 characteristic does not apply to one of the genders may be more evident. Further research is needed to explore the basis In Experiment 2 we aimed to replicate the first experiment of this effect. with some changes. In order to emphasize the generic nature of the sentences we used a bare plural form in both the story Experiment 3 and the sentences in place of the singular definite form “The It is striking that in spite of a high rate of acceptance of Rattle Bird” and anaphoric pronoun “it lives in forests”. gender-based minority characteristics in previous research Otherwise the procedure was unchanged. (e.g. Khemlani et al., 2007), our results were far less positive. One possible reason may be the fictional creatures Method that we have used. One aim of Experiment 3 was therefore Participants. A further 50 students at “Kore University” of to compare fictional cases with real-life examples, to see Enna (Italy), participated without any reward. whether the use of fictional cases is valid. The second aim Material. The materials were the same as in Experiment 1 was to test an alternative hypothesis about when gender- save for the changes described above. based minority characteristics would be acceptable. Design and Procedure. The design and procedure were the Gender-based generics in previous research have sometimes same as in Experiment 1. used reproductive properties (“lays eggs”, “suckles its Results and Discussion. As in Experiment 1 we counted young”) and sometimes differential appearance (“has a the number (out of 16) of generics that were judged as true mane”, “is red”). By contrast, all the sentences used in by each participant. The distributions for each condition are Experiments 1 and 2 were about appearance only. We shown in Figure 3. This time, agreement with the generic hypothesized that a key factor in accepting the truth of statements was clearly greater for the neutral stories (58%) gender-specific minority characteristics may relate to than for the gender based stories (12%). reproduction per se, rather than to other gender related This time the difference between the generalization of features. We therefore constructed gender stories generics in the neutral and the gender condition was much differentiating between features of physical appearance and greater and significant (χ2 = 13.3, p < .001 on a median reproduction. We then expected a stronger True response for test). Thus, contrary to our prediction, people were much 160 the generics about reproduction than for the generics about first, because of a contrast with the less convincing fictional appearance. cases which they followed. The Experiment therefore had two factors, fictional versus real-life creatures and reproductive versus appearance gender-based properties. We predicted that acceptance would be stronger for real-life than for fictional cases, but that the difference between reproductive and appearance features would be evident in both kinds of case. Method Participants. A further 50 students at “Kore University” of Enna (Italy), participated voluntarily. Materials. Each booklet contained four sets of descriptions with pictures: two based on real creatures (lions and deer) and two on fictional dimorphic creatures (Tabbies and Rattles). Each story was followed by 6 sentences: 2 true, 2 false, 1 generic on physical appearance (i.e. lions have manes), and 1 generic on reproduction (i.e. lions give live birth to their young). The order of the stories within the booklets was randomized. The form of the sentences was a bare plural as in Experiment 2. Design and Procedure. Participants were randomly divided into 2 groups, one group first saw the two stories based on real animals and then the two fictional creatures, while the other group saw the stories in the opposite order. Thus, the Figure 5. True Responses by condition In Experiment 3. condition of order of appearance was between subjects and Error bars show Standard Error. the type of creature (real or fictional) and type of generic (reproduction or appearance) was within. The first page of These results supported our prediction that reproduction is each booklet contained a cover sheet with instructions. an exceptional case that encourages generics because the Participants circled one of 2 response options (true or false), predicate isn’t so much “false” of the males, as that it just printed to the right of each sentence. doesn’t sensibly apply. On the other hand appearance features are quite free to be true of either gender, and so are less likely to be accepted as true. Real cases were better liked than the fictional ones, but both showed the same effect. This result validates our use of fictional examples. While participants found them less convincing overall, the same preference for reproductive characteristics was shown with both types of material. General discussion Our three experiments are an initial exploration of why and when people are willing to accept minority characteristics as Figure 4: Male and Female Lions in Experiment 3 generically true. Previous research (Khemlani et al., 2007) has shown that people accept the truth of sentences which Results and Discussion. The data were submitted to are only true of a minority of a class. However all the ANOVA with within-subjects factors of Real vs Fictional sentences of this kind that they studied related to gender- and Reproduction vs Appearance. There were significant specific features such as “lions have manes” or “ducks lay main effects of Reproduction vs. Appearance (F(1,49) = eggs”. Our results provide some further understanding of 18.02, p < .001) and of Fictional vs. Real (F(1,49) = 8.73, p these interesting cases. First, in Experiments 1 and 2, we < .005). There was no interaction, F<1, (see Figure 5). showed that, surprisingly, people are less willing to accept We found also a significant effect of the order of appearance that a minority characteristic is true of a class when it is of the two conditions, which however did not interact with linked to gender. In Experiment 2, 58% of generics were the other factors. There was a contrast effect. When the accepted when they were true of an arbitrary subset of the fictional cases were seen first, they received higher ratings class, but only 12% were accepted when they were true of generally than when seen second, following the more only one specific gender. convincing real cases. By contrast when the real cases were Experiment 3 explored a further question about minority seen second they received higher ratings than when seen characteristics, namely whether gender based features that 161 relate to reproduction may be more acceptable than those Khemlani, S., Leslie, S.J. & Glucksberg, S. (2012). that refer to appearance. In Leslie et al.’s data, the Inferences about Members of Kinds: The Generics overgeneralisation to universally quantified sentences (e.g. Hypothesis. Language and Cognitive Processes, 27, 887- “All ducks lay eggs”) was more convincing with 900 reproductive features than with appearance (e.g. “All Khemlani, S., Leslie, S., Glucksberg, S., & Fernandez, P.R. kangaroos have pouches”), possibly because of the (2007). Do Ducks Lay Eggs? How People Interpret ignorance of some participants. To avoid problems of Generic Assertions. Proceedings of the 29th Annual ignorance, we provided our participants with the relevant Cognitive Science Society. Nashville, TN: Cognitive information in a short text, and then tested the acceptability Science Society. of the generics. We found that indeed reproductive features Leslie, S. (2014). Carving up the social world with generics. were generally better accepted (76% of judgments in the Oxford Studies in Experimental Philosophy, 1, 208-232. case of real creatures) than appearance features (51%). Leslie, S. (2015). Generics Oversimplified. Nous, 49, 28-54. Leslie (2015) discusses a suggestion by Liebesman (2011) Leslie, S., Khemlani, S., & Glucksberg, S. (2011). All ducks to the effect that generics are statements not about lay eggs: The generic overgeneralization effect. Journal individuals (singular or plural) but about kinds. Similarly of Memory and Language, 65, 15-31. Hampton (2012a; 2012b) proposes that generics are Liebesman, D. (2011). Simple Generics. Nous, 45, 409–442. considered true when they include information that is a part of the intensional prototype that represents the kind. That could explain why very rare but striking properties are also commonly accepted as true (e.g. sharks attack bathers). They are a part of our knowledge base about the kind, because of the need to take suitable precautions when swimming near sharks, and the spectacularly dire consequences of failing to do so. Given this framework, it is possible that reproductive features are more easily attributed to the kind than are gender-based appearance features. Ducks laying eggs is relevant to both male and female ducks, simply because they all, as individuals, began life this way. On the other hand lions having manes is quite specific to the males. Another explanation for our result in Experiment 3 relates to a hypothesis put forward by Andrei Cimpian (Cimpian & Markman, 2011) that generics are more likely to be acceptable when they reflect essential or deep causal properties of a kind. Reproductive features are a part of a network of strongly related properties of a kind, embedded in theories of sexual reproduction and the rearing of offspring. They would therefore be more likely to find acceptance as generics than gender-differentiating appearance features. References Carlson, G. N., & Pelletier, F. J. (1995). The generic book. Chicago, IL: Chicago University Press. Cimpian, A., & Markman, E.M. (2011). The Generic/Nongeneric distinction influences how children interpret new information about social others. Child Development, 82, 471-492. Greenberg, Y., (2003). Manifestations of Genericity. Routledge. New York. Hampton, J. A. (2012a). Generics as reflecting conceptual knowledge. Recherches Linguistiques de Vincennes, 41, pp. 9-24. Hampton, J. A. (2012b). Thinking intuitively: The rich (and at time illogical) world of concepts. Psychological Sciences, 21, 398-402. 162