What Makes a Grammar Diffcult? Experimental Evidence
Arturs Semenuks, Aleksandrs Berdičevskis · Proceedings of the 12th International Conference on the Evolution of Language (Evolang12) · 2018
It is sometimes argued that linguistic structure can be affected by extralinguistic factors (Ladd, Roberts & Dediu, 2015).One specific hypothesis that has garnered a significant interest in the recent years links increase in the proportion of adult non-native (L2) learners in the population of a language to decrease in the morphological complexity of that language (Trudgill, 2011).Support for this view comes from qualitative typological analyses (Dahl, 2004;McWhorter, 2007;Trudgill, 2011), as well as correlational studies using both proxy (Lupyan & Dale, 2010) and direct measures of L2 proportion in the population (Bentz & Winter, 2013).This suggests that a greater L2 share in the population is associated with a higher probability of a language being analytic.The exact details of how and why increased number of adult learners in the population can lead to morphological simplification, however, are not yet fully fleshed out.One hypothesis proposes that the simplification crucially depends on the cognitive differences between children and adults (Lupyan & Dale, 2010), as adults experience more difficulties learning morphology (DeKeyser, 2005).These difficulties, in turn, lead to the simplification of morphology.McWhorter ( 2001) argues that grammars of creoles predominantly use analytic or isolating strategies, but an analytic morphosyntax by itself does not imply that a language had been affected by L2 speakers.However, his qualitative analyses do suggest that older non-creole languages have more grammatical distinctions compared to creoles (McWhorter, 2001).This begs the question of what exactly is more difficult for L2 speakers, and thus more likely to change in a contact situation: how many grammatical categories a language has or how complex the expression of these categories is (i.e.fusional is more difficult than agglutinative)?To investigate this, we designed an artificial language (AL) learning experiment in which we manipulated the number of grammatical distinctions and the way they are expressed in the AL, and measured how these differences