Whats so sexy about degenderizing language? Investigating gender representations by readers and listeners in Norwegian, Finnish and French

When referring to a person's personal, social or professional role, such as a scientist, traveller or manager, knowing the person's biological sex is not always crucial for comprehending a discourse. Previous research suggests that readers and listeners nevertheless do so, and often do so in ways that unnecessarily narrow their mental representation of the referent, thereby biasing one gender or the other. To avoid gender-directional language and thus biased gender representations, in Norway a linguistic strategy called "degendering by neutralization" has been suggested. This strategy refers to the use of word forms which do not specify gender when referring to women or to men, such as dropping feminine suffixes (lærerinne => lærer, sykepleierske => sykepleier) as well as other means of highlighting gender information (ombudsmann => ombud). However, what remains to be addressed is what impact this strategy of removing linguistic gender cues has on how gender is comprehended. This project investigates this issue by conducting experimental research with three languages (Finnish, Norwegian, and French) and two modalities of language comprehension (reading and listening). These languages were selected based on the extent to which they encode gender, with Finnish not grammatically encoding gender, French encoding both masculine and feminine forms, and Norwegian intermediating these (though gender is seldomly marked explicitly in Norwegian, the predominance of the grammatically masculine form when referring to persons may be interpreted as a cue to biological gender, thereby enforcing a male bias). Further, compared to the acoustic signal used in listening, when a linguistic expression is visually available in writing, a reader can potentially update or revise a gender representation, thereby raising the possibility that use of cues for gender representation may differ between listening and reading. We adopt a new experimental approach, complementing traditional language comprehension with non-verbal comprehension, using a face recognition task. More specifically, participants read short sentences on a computer display, each of which is followed by a face pair which they have to classify according to its gender. The assumption is that if a sentence activates a mental representation of a female subject, then participants will be quicker to correctly classify female faces than male faces while this representation is active. The use of a visual target task allows us to examine whether gender-related differences in linguistic forms extend beyond language to also influence broader, non-verbal conceptual gender categories, a factor which has been unexplored until now. The development of this task resulted in the elaboration of an experiment design in which response times could be gathered for three response alternatives in parallel (i.e., male, female or mixed face pairs), while taking into account that the latter category constitutes a combination of the former two. Evaluation of the design confirmed that it is sufficiently sensitive to detect gender priming effects, which is a prerequisite for the full series of experiments. The number of response alternatives available in the present study goes beyond what is typical for comparable paradigms, and the fact that one of the alternatives constitutes an overlap of the other categories (the mixed category incorporates elements from both the female and the male categories) contributes to increasing task complexity. Therefore, the experimental task was designed with particular emphasis on keeping response registration as simple as possible. This was achieved by implementing unimanual responses for unigender face pairs and bimanual responses for mixed gender face pairs. Additional evaluation in which the elaborated task design was compared to a more traditional approach confirmed that the former indeed lead to more efficient processing. To investigate potential biases, the current project focuses on mental representations activated by gender neutral language, and the results from all languages indicate that the use of neutral forms are an effective means of avoiding gender biases. For example, the gender bias observed after a sentence such as "The teachers were tired" is comparable to that observed after "The tables were new". Even in cases where an initial male bias was detected as participants were quicker to recognize male face pairs than female face pairs in the absence of verbal information, this bias was neutralized after the presentation of neutral language primes. Though some results are still pending, there are currently no indications to suggest that neutralization has a different impact on online processing across the different languages, i.e. grammatical gender languages seem likely to benefit as much from the use of neutral language forms as non gendered languages, whether the grammatical cues are tightly or loosely associated with biological sex.

The project has successfully developed and tested an experimental paradigm adequate to assess linguistically primed gender representations across different languages and modalities, and also to handle the implementation of three simultaneously available and overlapping response alternatives. Initial results indicate that the use of gender neutral language contributes to the reduction of inherent social gender biases, and upcoming results are expected to shed more light on the underlying processes, particularly with respect to modality. The project thereby contributes to understanding the impact of using gender neutral language and also offers a central methodological contribution to future work.

To avoid gender-directional language and thus biased gender representations, degendering by neutralization (i.e., use of a gender-unmarked form to refer to both women and men) has been suggested in Norway as a linguistic strategy. However, the impact of this strategy in regard with the question of how gender is processed when gender cues are removed from language has not been addressed. This project investigates this question by conducting experimental research in three language contexts (Finnish, Norwegian, and French) and two modalities of language comprehension (reading and listening) Across four experiments, one experimental referent condition (1) and two control referent conditions (2 and 3) will be tested in a task that will reveal the processes involved in text (Experiments 1 & 2) and spoken language comprehension (Experiments 3 & 4) regarding the gender of human referents: (1) Selected role names (plural form) that will be neither semantically nor conceptually gender marked and that are not linked to gender stereotypical expectations (e.g., The athletes...) (2) Explicit male and female human referents, operationalized via first names (e.g., Magnus and Lars ...) (3) Non-human referents (e.g., The busses ...) To evaluate the impact of degendering by neutralization (Norway) on global gender representations we will run these experiments in Norwegian and Finnish (a language that does not encode gender grammatically), as well as in French (visibility by feminization, i.e., using two forms simultaneously to refer to both men and women) and contrast them in terms of mental representations of gender. We also adopt a new experimental approach to include non-verbal representations by complementing traditional text comprehension with perceptual measures (i.e. face recognition task). This will allow us to address factors unexplored until now, consequently examining whether gender-related differences in linguistic forms can also be traced in non-verbal representation.