Lexical representation of gender: a quasiregular domain

A standardized approach to measuring gender transparency in languages

Languages can express grammatical gender through different ortho-phonological regularities present in nouns (e.g., the cues "-o" and "-a" for the masculine and the feminine respectively in Italian, Portuguese, or Spanish). The term "gender transparency" was coined to describe these regularities (Bates et al., 1995). In gendered languages, we can hence distinguish between transparent nouns, i.e., those displaying form regularities; opaque nouns, i.e., those with ambiguous endings; and irregular nouns, i.e., those that display the typical form regularities but are associated with the opposite gender. Following a descriptive analysis of such regularities, languages have been recently classified according to their degree of gender transparency, which seems relevant in regard to gender acquisition and processing. Yet, there are certain inconsistencies in determining which languages are overall transparent and which are opaque. In particular, it is not clear whether some other complex regularities such as derivational suffixes are also "transparent" cues for gender, what really constitutes an "opaque" noun, or which role orthography and morphology have in transparency. Given the existing inconsistencies in classifying languages as transparent or opaque, this work introduces a proposal to assess gender transparency systematically. Our methodology adapts the standardized factors proposed by Audring (2019) to analyse the relative complexity of gender systems. Such factors are adapted to gender transparency on the basis of the literature on gender acquisition and processing. To support the feasibility of such a proposal, the concepts have been instantiated in a quantitative model to obtain for the first time an objective measure of gender transparency using European Portuguese and Dutch as instances of target languages. Our results coincide with the theoretically expected outcome: European Portuguese obtains a high value of gender transparency while Dutch obtains a moderately low one. Future adaptations of this model to the gender systems of other languages could allow the continuum of gender transparency to sustain robust predictions in studies on gender processing and acquisition.

The role of word form in gender processing during lexical access: A theoretical review and novel proposal in language comprehension

In contrast to language production, there are few comprehension models of the representation and use of grammatical gender in long-term memory. To bridge this gap, we conducted a systematic review of empirical studies on the role of gender-form regularities in the recognition of nouns in isolation and within sentences. The results of a final sample of 40 studies suggest that there are two routes for the retrieval of gender during real-time comprehension: a form-based route and a lexical-based route. Our review indicates that the use of these routes depends on the degree of gender transparency of the language and the degree of overtness of the experimental paradigm. To accommodate these findings, we incorporate a dual-route mechanism within a general model of lexical access in comprehension, the AUSTRAL (Activation Using Structurally Tiered Representations and Lemmas) model, and identify directions for future research.

English Speakers' Implicit Gender Concepts Influence Their Processing of French Grammatical Gender: Evidence for Semantically Mediated Cross-Linguistic Influence

Second language (L2) learners often show influence from their first language (L1) in all domains of language. This cross-linguistic influence could, in some cases, be mediated by semantics. The purpose of the present study was to test whether implicit English gender connotations affect L1 English speakers’ judgments of the L2 French gender of objects. We hypothesized that gender estimates derived from word embedding models that measure similarity of word contexts in English would affect accuracy and response time on grammatical gender (GG) decision in L2 French. L2 French learners were asked to identify the GG of French words estimated to be either congruent or incongruent with the implicit gender in English. The results showed that they were more accurate with words that were congruent with English gender connotations than words that were incongruent, suggesting that English gender connotations can influence grammatical judgments in French. Response times showed the same pattern. The results are consistent with semantics-mediated cross-linguistic influence.

Form and Function: A Study on the Distribution of the Inflectional Endings in Italian Nouns and Adjectives

Inflectional values, such as singular and plural, sustain agreement relations between constituents in sentences, allowing sentence parsing and prediction in online processing. Ideally, these processes would be facilitated by a consistent and transparent correspondence between the inflectional values and their form: for example, the value of plural should always be expressed by the same ending, and that ending should only express plural. Experimental research reports higher processing costs in the presence of a non-transparent relation between forms and values. While this effect was found in several languages, and typological research shows that consistency is far from common in morphological paradigms, it is still somewhat difficult to precisely quantify the transparency degree of the inflected forms. Furthermore, to date, no accounts have quantified the transparency in inflection with regard to the declensional classes and the extent to which it is expressed across different parts of speech, depending on whether these act as controllers of the agreement (e.g., nouns) or as targets (e.g., adjectives). We present a case study on Italian, a language that marks gender and number features in nouns and adjectives. This work provides measures of the distribution of forms in the noun and adjective inflection in Italian, and quantifies the degree of form-value transparency with respect to inflectional endings and declensional classes. In order to obtain these measures, we built Flex It, a dedicated large-scale database of inflectional morphology of Italian, and made it available, in order to sustain further theoretical and empirical research.

The neural substrate of noun morphological inflection: A rapid event-related fMRI study in Italian

The present research investigated the neural correlates of nominal inflection and aimed at disclosing their possible link with the frequency distribution of noun inflectional features: grammatical gender, inflectional suffixes and inflectional classes. The properties of the Italian nominal system were exploited since it allows to explore exhaustively fine-grained phenomena in the inflectional processing. An event-related functional magnetic resonance imaging (fMRI) experiment was carried out where Italian masculine and feminine nouns were visually presented to 50 healthy participants in an overt inflectional task: the generation of the plural from the singular and vice versa. The grammatical gender and the citation form suffix of nouns were manipulated in a factorial design. Functional data showed that inflectional operations for nouns activate an extensive cortical network involving the left inferior and right superior frontal gyri, the left and right middle temporal gyri, the posterior cingulate cortex and the cerebellum. Activations were variably modulated by the distributional features of gender-dependent properties of nouns. Particularly, cortical activity increased during inflectional operations for small and/or scarcely consistent inflectional classes. These findings demonstrate the relevance of specific morphological (inflectional suffixes) and distributional features (size and consistency) shared by groups of words (inflectional classes) in a language, particularly when implementing cognitive operations required for language processing.

The nature of morphosyntactic processing during language perception. Evidence from an additional-task study in Spanish and German

The present study investigates in how far morphosyntactic processing is affected by an additional non-verbal task and whether this effect differs between German and Spanish, two languages with differences in processing grammatical gender (lexical vs. cue-based processing). By manipulating task load and language we aimed at getting an insight into subprocesses of morphosyntax and their dependence on resources of general and verbal working memory, respectively. In more general terms, this study contributes to the debate on the modularity of morphosyntax. Written German or Spanish sentences with or without gender violations were presented word by word to native speakers. The critical words temporally overlapped in different degrees with a non-linguistic stimulus (a high or low tone). In a single task (Experiment 1) participants judged sentence acceptability and ignored the tones. Experiment 2 required a response to the tones. Left-anterior negativity (LAN) and P600 components were analyzed in the ERPs to critical words. Whereas the LAN was not affected by any of the experimental manipulations, the P600 was modulated as a function of language during the single task conditions (Experiment 1). In Experiment 2 the additional task did not add up with this effect; instead, the differences between language groups vanished. This may indicate that the processes reflected in the P600 draw on resources of general working memory. The LAN data seem to be in line with modularity of first pass morphosyntactic processing, although this interpretation contradicts findings from other studies. The P600 results may highlight the flexibility of sentence-based syntactic processing.

The Processing of Grammatical Gender Agreement in Brazilian Portuguese: ERP Evidence in Favor of a Single Route

The present study used event-related potentials to investigate whether the processing of grammatical gender agreement involving gender regular and irregular forms recruit the same or distinct neurocognitive mechanisms and whether different grammatical gender agreement conditions elicit the same or diverse ERP signals. Native speakers of Brazilian Portuguese read sentences containing congruent and incongruent grammatical gender agreement between a determiner and a regular or an irregular form (condition 1) and between a regular or an irregular form and an adjective (condition 2). However, in condition 2, trials with incongruent regular forms elicited more positive ongoing waveforms than trial with incongruent irregular forms. We found a biphasic LAN/P600 effect for gender agreement violation involving regular and irregular forms in both conditions. Our findings suggest that gender agreement between determiner and nouns recruits the same neurocognitive mechanisms regardless of the nouns' form and that, depending on the grammatical class of the words involved in gender agreement, differences in ERP signals can emerge.

Incidental Learning of Gender Agreement in L2

Incidental learning of grammar has been an area of interest for many decades; nevertheless, existing research has primarily focused on artificial or semi-artificial languages. The present study examines the incidental acquisition of the grammar of a natural language by exposing adult speakers of an ungendered L1 (English) to the gender agreement patterns in Russian (a language that was novel to the learners). Both receptive and productive knowledge and the mediating role of working memory (WM) in learning were measured. Speakers of the ungendered language were able to successfully acquire receptive but not productive grammatical knowledge in a new language under incidental exposure. WM was engaged in production but not in a grammaticality judgment task in the incidental learning condition, indicating cognitive effort during knowledge retrieval.

Sublexical and lexico-syntactic factors in gender access in Spanish

The influence of sublexical and lexico-syntactic factors during the grammatical gender assignment process in Spanish was studied in two experiments using the gender decision task. In Experiment 1, the regularity of the ending of gender-marked nouns (masculine nouns ended in -o and feminine nouns ended in -a) and of nouns with gender-correlated but unmarked word-endings (e.g., -ad) was manipulated. The results showed that regularity affected reaction times and error rates only in the case of gender-marked nouns, suggesting that the mere statistical distribution of a word-ending across genders is not responsible for the regularity effect. In Experiment 2, gender-marked nouns and gender-unmarked nouns were preceded by a masked prime which could be a definite article (which provides information about the gender of the noun) or a possessive pronoun (which does not contain gender information). The presentation of the definite article led to shorter reaction times and less errors only when the word-ending was different from -o or -a. Taken together, these results indicate that gender assignment in Spanish is carried out through different processes depending on the noun ending: gender decisions for gender-marked nouns are based on the gender-to-ending distribution. Meanwhile, gender decisions for unmarked nouns seem to require the retrieval of the corresponding definite grammatical article, regardless of the statistical distribution of the noun ending across genders.

Implicit knowledge of grammatical gender in preschool children

The study aimed at investigating the role of nominal gender in animal categorization in preschoolers. Given the regularities characterizing gender system, at both syntactical and morphological level, Italian language is suitable to address this issue. In three experiments, participants were asked to classify pictures of animals as male or female. Half stimuli had names of feminine gender and half of masculine gender. In Experiment 1, Italian speaking adults and preschoolers classified animals according to the nominal gender. This effect was not found with English speaking participants (Experiment 2) but confirmed with 3-, 4-, and 5-year-old Italian-speaking children (Experiment 3). These results showed an implicit knowledge of grammatical gender in preschoolers, suggesting that semantic processing may be modulated by linguistic information.

Corroborating biased indicators: global and local agreement among objective and subjective estimates of printed word frequency

The internal validity of several types of experiments in experimental psychology and neuroscience depends in part on the possibility of controlling or manipulating critical lexical variables such as word frequency of occurrence. Two ways of estimating this variable are (1) objective frequency counts and (2) subjective ratings of word frequency. Each method produces estimates that generally agree (i.e., they are highly correlated) but that disagree substantially concerning the relative frequency of a number of words. To investigate this issue more closely, the global and local agreement of subjective frequency estimates was examined in detail for a pool of 6,202 words drawn from the OMNILEX database of French words (Desrochers, 2006; www.omnilex.uottawa.ca). The results indicated that objective and subjective frequencies are strongly correlated, subjective frequencies share a significant amount of bias variance with other lexical characteristics (e.g., imageability), and the codeterminants of subjective frequency are in an antagonistic relationship with one another. The implications of these results for the selection of lexical stimuli are discussed, and multiple variables to aid in item selection are reported. Supplemental materials for this study may be downloaded from brm.psychonomic-journals.org/content/supplemental.

Phonological regularities and grammatical gender retrieval in spoken word recognition and word production

Two experiments investigate whether native speakers of French can use a noun's phonological ending to retrieve its gender and that of a gender-marked element. In Experiment 1, participants performed a gender decision task on the noun's gender-marked determiner for auditorily presented nouns. Noun endings with high predictive values were selected. The noun stimuli could either belong to the gender class predicted by their ending (congruent) or they could belong to the gender class that was different from the predicted gender (incongruent). Gender decisions were made significantly faster for congruent nouns than for incongruent nouns, relative to a (lexical decision) baseline task. In Experiment 2, participants named pictures of the same materials as used in Experiment 1 with noun phrases consisting of a gender-marked determiner, a gender-marked adjective and a noun. In this Experiment, no effect of congruency, relative to a (bare noun naming) baseline task, was observed. Thus, the results show an effect of phonological information on the retrieval of gender-marked elements in spoken word recognition, but not in word production.

The relation between language and cognition in 3- to 9-year-olds: the acquisition of grammatical gender in French

The French language has a grammatical gender system in which all nouns are assigned either a masculine or a feminine gender. Nouns provide two types of gender cues that can potentially guide gender attribution: morphophonological cues carried by endings and semantic cues (natural gender). The first goal of this study was to describe the acquisition of the probabilistic system based on phonological oppositions on word endings by French-speaking children. The second goal was to explore the extent to which this system affects categorization. In the study, 3- to 9-year-olds assigned gender categorization to invented nouns whose endings were typically masculine, typically feminine, or neutral. Two response conditions were used. In the determiner condition, children indicated the gender class by orally providing the determiner un or une marked for gender. In the picture condition, responses were given by pointing to the picture of a Martian-like female or male person that would be best called by each spoken pseudoword. Results indicated that as young as 3 years, children associated the determiner corresponding to the ending bias at greater than chance levels. Ending-consistent performance increased from 3 to 9 years of age. Moreover, from 4 years of age onward, sensitivity to endings affected categorization. Starting at that age, pictures were selected according to endings at greater than chance levels. This effect also increased with age. The discussion deals with the mechanisms of language acquisition and the relation between language and cognition.

Gender processing in first and second languages: The role of noun termination

In 2 picture-naming and 2 grammaticality judgment experiments, the authors explored how the phonological form of a word, especially its termination, affects gender processing by monolinguals and unbalanced bilinguals speaking German. The results of the 2 experiments with native German speakers yielded no significant differences: The reaction times were statistically identical for items from gender typical, ambiguous, and gender atypical groups. The 2 experiments with English bilinguals who had learned German as a second language (L2), however, provided evidence that the L2 word's termination plays a role in L2 gender processing. Participants were fastest when producing gender-marked noun phrases containing a noun with a gender typical termination and slowest when the noun had a gender atypical termination. Analogous results were obtained in the grammaticality judgment experiment. These findings support the assumption that there is interaction between the levels of phonological encoding and grammatical encoding at least in bilingual processing.

Assigning grammatical gender during word production

The present study was designed to examine the processes by which grammatical gender is assigned during word production. French words varied in strength of sublexical cues, based on whether the word ending was typical for one gender rather than neutral about gender, and lexical cues, derived from the associated definite article being uninformative about gender for words beginning with a vowel, but informative for words beginning with a consonant. In Experiment 1. when native French speakers classified the gender of mentally evoked names of pictures, no effects of these cues were obtained. Experiment 2 used an improved methodology, with participants classifying the gender of words translated from English. English-speaking learners of French were influenced strongly by lexical and sublexical cues, while French speakers showed a weaker influence. However, for both speaker groups, words whose gender was classified slowly during recognition were also classified slowly during production, and error rates were similarly correlated across tasks. The conclusion was that gender is not equally available for all words once the associated "lemma" is accessed Current models of language production may have to incorporate mechanisms allowing differential speed of access to gender information depending on a word's formal properties.

Grammatical gender in the brain: Evidence from an fMRI study on Italian

The grammatical gender of a word is a lexical-syntactic property determining agreement among different sentence parts. Recent fMRI investigations identified the areas involved in the retrieval of grammatical gender near the left Broca's area providing further evidence to confirm the preeminent syntactic role of this area. However, these studies employed categorical designs based on the controversial methodology of the cognitive subtraction of neural activations related to different tasks. In the present study we identified the neural substrates of grammatical gender assignment using an fMRI parametric study. Participants decided the grammatical gender of visually presented Italian words whose gender-to-ending regularity varied. The results showed activation in left and right fronto-temporal areas suggesting an interplay of both hemispheres in the processing of grammatical gender.

Sublexical and lexical influences on gender assignment in French

Two experiments showed that French native speakers rely on sublexical and lexical cues to allocate gender during word recognition. Sublexical cues were based on whether the word ending was typical for a particular gender rather than neutral with regard to gender. Lexical cues were based on whether the associated definite article was informative (for words beginning with a consonant) or uninformative (for words beginning with a vowel). Classification of single nouns and verification of grammatical combinations of indefinite article and noun led to longer times when both sublexical and lexical cues were uninformative compared with when one or both cues were informative. Verification of ungrammatical combinations of indefinite article and noun yielded separate effects of both cues, though only when monitoring for both semantic and syntactic unacceptability in meaningful phrases did people attend to both cues independently. It was argued that people became more cautious in their gender assignments as task requirements became "deeper": If strategic changes as a function of task demands are incorporated, the results are compatible with connectionist models proposing that gender decisions are computed from strength of past associations of the word and gender-specifying elements.

On the representation of Portuguese gender-inflected words in the mental lexicon

This study aims at verifying whether Portuguese gender-inflected nouns and adjectives are represented as full forms as suggested by Spanish data (Dominguez, Cuetos, & Segui, 1999). A series of lexical decision experiments is reported. Grammatical gender, frequency dominance, and grammatical category are manipulated and cumulative frequency is controlled. The results do not provide support for a full form representation of gender-inflected words. They suggest that grammatical category, or the nature of the inflectional process involved (lexical or syntactic), affects the way words are represented and accessed. Shorter recognition latencies were obtained for nouns drawn from Feminine dominant gender-inflected pairs than from Masculine dominant pairs whereas a tendency in the opposite direction was observed in adjectives. The effect of frequency dominance appears, nevertheless, to be restricted to feminine nouns. The data are compatible with the view that masculine nouns and adjectives are represented as gender-unmarked forms. These results are discussed in relation to current dual-access models of word recognition and to the notion of "interpretability" of lexico-syntactic features, as put forward in the Minimalist Program of Generative Linguistics.

Explaining derivational morphology as the convergence of codes

Language users have a remarkable ability to create, produce and comprehend complex words. Words such as undercut and bakery appear to be composed of units, traditionally called morphemes, that recombine in rule-like ways to form other words, such as underline and cannery. However, morphological systems are quasiregular: they are systematic and productive but admit many seemingly irregular forms. Thus, bakery is related to bake and cannery to can but what is the groce in grocery? There is no bread in sweetbreads, liver in deliver, corn in corner or ginger in gingerly. Such words exhibit partial regularities concerning the correspondences between form and meaning, the treatment of which has important implications for linguistic and psycholinguistic theories. This article describes an approach to morphological phenomena called the convergence theory, in which morphology is a graded, inter-level representation that reflects correlations among orthography, phonology and semantics.