Comparison of the acquisition of two transparent gender systems: Italian and Croatian

This study compares the acquisition of the gender values in Italian and Croatian monolingual children. Both Italian (Chini, 1995; Gudmundson, 2010) and Croatian (Kovačević, Palmović, & Hržica, 2009) are considered to have transparent gender systems, i.e. that the gender of the noun is evident from its form. All languages diverge from transparency to some extent (Audring, 2014) and thus this study focuses on the different degrees of transparency between the two languages placing them on a continuum on gender transparency (Kupisch, Geiß, Mitrofanova, & Westergaard, 2018). The criteria taken into consideration here are the presence of a gender marked article, which Italian has (Chini, 1995) and Croatian does not; and the syncretism between genders in the extended nominal paradigm. Additionally, Croatian has a three gender values (masculine, feminine, and neuter) with the neuter having low frequency in child-directed speech (Kovačević et al., 2009), while Italian has a two-way distinction (masculine and feminine) and the two genders are quite equally represented in the language (Costa, Kovačić, Franck, & Caramazza, 2003). According to these criteria, Italian is more transparent than Croatian, and thus we expect it to be acquired earlier by the children.

We have conducted an adjective elicitation task on a total of 60 monolingual Italian and Croatian children divided into two age groups (Italian=3;0 and 3;10, Croatian= 2;10 and 4;2) with 15 participants in each group.

The results were obtained by conducting a linear regression with Jamovi (Love, Dropmann, & Selker, 2018) on the data contained in figure 1. The statistical analyses have revealed that the gender in Italian noun-adjective agreement is mastered already by the youngest child (age=2;6) as there were no statistical differences between the younger and older group; there was also no difference between the masculine and feminine gender values. Conversely, there was a significant difference of the error ratio between the two age groups in the Croatian children (p-value=0.003). Additionally, two stages of acquisition can be identified for the Croatian children: the first stage consists of similar error rates with F and M, but significantly higher error rates with N; the second stage consists of F being at ceiling and the error rates with M and N being similar, but with N significantly improved.

What the results suggest is that the degree of transparency of the gender system matters. We cannot look only at a manifestation of gender in isolation, but at the full agreement, paradigm to make more accurate predictions of how a gender system might be acquired. Transparency is not a binary value but has to be placed on a continuum and the full paradigms of the agreeing elements have to be taken into consideration to assess how transparent a gender system is. Frequent and clear cues contribute greatly to a fast mastery of the gender system (i.e. the Italian article), while low frequency in the input (Croatian N) hinders this process. 



Audring, J. (2014). Gender as a complex feature. Language Sciences, 43, 5-17.

Chini, M. (1995). Genere grammaticale e acquisizione: aspetti della morfologia nominale in italiano L2 (Vol. 14): Franco Angeli.

Costa, A., Kovačić, D., Franck, J., & Caramazza, A. (2003). On the autonomy of the grammatical gender systems of the two languages of a bilingual. Bilingualism: Language and Cognition, 6(3), 181-200. doi:10.1017/S1366728903001123

Gudmundson, A. (2010). L'acquisizione del genere grammaticale in italiano L2: Quali fattori possono influenzare il grado di accuratezza. Department of French, Italian and Classical Languages,

Kovačević, M., Palmović, M., & Hržica, G. (2009). The acquisition of case, number and gender in Croatian. Development of nominal inflection in first language acquisition: A cross-linguistic perspective, 153-177.

Kupisch, T., Geiß, M., Mitrofanova, N., & Westergaard, M. (2018). Gender cues in L1 Russian children acquiring German as an early L2. . Paper presented at the EuroSLA.

Love, J., Dropmann, D., & Selker, R. (2018). Jamovi project (Version jamovi version 0.9). Retrieved from