2018 Sat Session A 0930

Saturday, November 3, 2018 | Session A, East Balcony | 9:30am

Distributional Regularity of Suffixes Facilitates Gender Acquisition: A Contrastive Study of Two Closely Related Languages
T. Ivanova-Sullivan, I. Sekerina

Research on various languages and populations has highlighted the facilitatory role of transparency in gender acquisition [2, 3, 4, 6, 7, 8, 9], but few studies investigated it in closely related languages, e.g., Russian and Polish [2]. Janssen (2016) found that gender assignment was more accurate in Polish children than Russian children because of fixed stress position and higher morphophonological regularity in Polish.

The current project is the first contrastive study of gender acquisition in two other closely related languages, Russian and Bulgarian. Russian differs from Bulgarian because its vowel reduction and palatalization of final consonants impact the form-function mapping of transparent vs. opaque gender suffixes. Particularly, opacity is present in inanimate nouns in all three genders in Russian whereas in Bulgarian, it exists only in FEM. We hypothesize that (1) opacity will delay acquisition of gender in both languages, but (2) gender assignment in Russian will be more accurate because of the distributional regularity of opaque nouns.

We report the results of an untimed gender production task with three factors, Language  x Gender x Transparency. Russian (N=23, age range 3;8-6;11) and Bulgarian monolingual children (N=22, age range 3;10-6;8) saw two pictures side-by-side with identical objects that differ only in color or size. The participants had to produce an adjective-noun combination in singular to describe each picture (Fig. 1).

Our findings confirmed (2) that overall, the Russian children were more accurate (92.6%) than the Bulgarians (77.9%). A mixed logistic regression analysis showed that there was an effect of Transparency, with better production of the transparent MASC and FEM nouns (96.7%) than the opaque ones (88.5%). Russians also outperformed Bulgarians in FEM opaque (94.5% vs. 49.5%). In the latter group, there was also an effect of Transparency in FEM nouns with greater accuracy with transparent (91.5%) than with opaque (49.5%) FEM nouns (cf. Table 1 for examples and results).

In answering (1), we found that opacity indeed affects acquisition of gender in both languages but in different ways. The systematicity and pervasiveness of opacity in Russian that cut across all three genders make children more aware of its role in the input. Conversely, in Bulgarian, opacity is only present in FEM making it a difficult-to-notice exception that might be acquired on an item-by-item basis. The frequency of occurrence of such types of exceptions in the input was shown to affect the acquisition of both natural and artificial languages [1, 5].

Our study corroborates previous accounts of form-based gender acquisition [6,7,8], but at the same time raises the question regarding the function of opaque gender cues in the input. We propose that in Bulgarian, the facilitatory role of transparent gender cues in other Slavic languages, such as Russian [3,7,8] and Polish [2], is diminished by the distributional asymmetry of opaque nouns present only in FEM. Thus, we argue that both the degree of transparency of the gender system and the distribution of transparent vs. opaque nouns over genders should be taken into account in investigations of gender acquisition cross-linguistically.