Model-based clustering and first language acquisition

Language has been traditionally considered as a qualitative phenomenon that mainly requires hermeneutical methodologies in order to be studied, yet in recent decades - thanks to advances in data storage, processing and visualization - there has been a growing and fertile interest in analysing language by relying on statistics and quantitative methods. In light of these reasons, we think it is worthwhile to try to explore databases made up of transcripted infant spoken language in order to verify whether and how underlying patterns and recurrent sequences of learning stages work during acquisition. So, we think that model-based clustering method via the Expectation-Maximization (EM) algorithm can be useful to evaluate the development of linguistic structures over time in a reliable way.