Research

My dissertation investigates the interplay of phonotactic and morphological learning by focusing on two types of morphophonological interface phenomena: morphological derived environment effects (MDEEs) and morpheme-bounded phonotactics. I am conducting a series of Artificial Grammar Learning (AGL) experiments on these two types of pattern in which I vary the difficulty of morphological learning. The preliminary findings show that learning MDEEs is much more dependent on the difficulty of morphological learning that learning of morpheme-bounded phonotactics. I am also using computational learning to test hypotheses on how such a difference may arrive in the learning process. 

Sequences of [t] followed by a sibilant in Hebrew are overall underattested yet only actively repaired via metathesis in very specific morphological environments. We are interested in the role of morphology in constraining the well-formedness of such sequences and the application of the metathesis process in Hebrew speakers' grammars. 

We presented the results of a phonotactics judgement study at MFM (slides). We find a uniform preference for sibilant-[t] sequences both within-morpheme and in familiar between-morpheme contexts. However, we find a split of preference in non-familiar between-morpheme contexts (see figure below), which we argue to arise from difference in how speakers assign weight to general and specific markedness constraints in their grammar. We are currently using formal modelling to explore how such division in grammars might arise. We are also running a wug test on the same pattern to explore the relationship between phonotactic knowledge and repair. 

We are using a variety of methods (including meta-analysis, corpus study, and AGL experiments) to investigate the acquisition of non-concatenative morphology, which is abundant in the input of children learning Semitic languages but which consists of relating non-adjacent segments. Non-local phonotactic dependencies have been shown to be acquired rather late in languages with little to no non-concatenative morphology. 

As a first step, we have done a meta-analysis of priming studies to identify the status of non-concatenative morphology in the adult grammar. We found that root priming is robust across Semitic languages and reflects morphological priming effects that are independent from phonological and semantic effects, while the evidence of template priming is less conclusive. Read our paper in Mental Lexicon here.  We are currently running a series of priming experiments to address one open question on template priming, namely whether the difference in verbal and nominal template priming is due to a true word class effect or driven by type frequency differences. 

I'm conducting fieldwork on the dialect of K'iche' spoken in Cantel, Guatemala. This dialect features frequent deletion of stressless vowels. One topic of investigation is the opaque interaction of syncope with possessive allomorphy. The rich verbal morphology of the language also provides good testing grounds for how syncope is constrained by morphological structure.