Artificial Language Learning

We hypothesise that it is essential that preschoolers have acquired sufficient implicit knowledge of morphological and syntactic properties of their language prior to being instructed to read and write. Written language codes grammatical structures more clearly and consistently than spoken language, so learning to read and write will often lead children to detect – more or less implicitly – which elements of the language are grammatically relevant and how they are being combined to form phrases and sentences. Eventually, this will allow them to process increasingly complex written texts. The declinational system for nouns is key to understanding complex sentences and texts, as it is involved in encoding grammatically thematic roles and anaphoric relations. German declination is notoriously difficult, most likely due to the largely arbitrary assignment of grammatical gender to nouns: many German nouns possess no unambiguous phonological or semantic cues as to their gender. Such largely arbitrary grammatical categories are very hard to acquire. However, in an artificial language learning study with adult participants, Taraban (2004) has shown that syntactic cues in the morphosyntactic context of a noun are sufficient for the acquisition of semantically and phonologically unmarked gender-like subclasses, provided that the attention of the learners is drawn towards the relevant syntactic context. One way of achieving this is to present the linguistic input in a systematically structured, or blocked, fashion. A further remarkable feature of the German gender-case paradigm is that for example definite articles possess identical forms for various grammatical functions, what makes the acquisition of gender more difficult. Bebout (2013) replicated this finding with a different artificial language featuring a number of overlapping forms, as seen in German, too: For the case of definite articles, for instance, the total of 12 combinations of gender and case in the singular is covered by only six phonological forms; there are even fewer forms for the system of indefinite articles. Bebout demonstrated that it is possible to acquire artificial phonologically unmarked gender-subclasses despite such form collation phenomena in the syntactic marker system. In addition, she explored the influence of rhyme and melody on gender-like category acquisition and found out that especially a sung and rhymed version of the input induces substantial learning.
Objectives
In the present study, we intended to replicate and extend these findings reviewed above from adult learners to preschool children, so as to explore how the linguistic input must be presented for these children to be effective cues to grammatical gender. In the companion project on children's books and songs, we have shown that an optimized input presentation can be implemented in children's books and songs and is an effective means of fostering the acquisition of grammar in preschoolers.

Method

We assessed how well preschool children aged 4;6 to 6;6 acquired gender-like noun subclasses in an artificial language, when  
1) relevant syntactic cues were presented such that all relevant forms pertaining to one pseudonoun were presented in immediate succession (grouped presentation) as compared to in a random order, and when 
2) the input featured definite articles only as syntactic cues towards gender-like subclasses in the input (non-inflected condition) or definite articles plus adjective inflections (inflected condition). In the inflected condition, the syntactic cues were unambiguous but also more complex than in the uninflected condition, which featured fewer, and ambiguous, markers. 
In six individual sessions, the children trained the artificial language through interactive board games with varying linguistic tasks. 

Results

The children’s performance at test suggests that a grouped presentation is an efficient means of input optimization. Adding extra markers hinders the acquisition of gender-like subclasses but this can be compensated by highly noticeable additional markers that direct learner's attention towards paradigm cues.

References

Bebout, J. (2013). Can language play promote morphosyntactic learning? A Study about the impact of rhyme, melody and input structuring on gender-like category acquisition. Unpublished dissertation, Ruhr-Universität Bochum.
Taraban, R. (2004). Drawing learners’ attention to syntactic context aids gender- like category induction. Journal of Memory and Language, 51, pp. 202-216.