What kind of knowledge must speakers possess in order to produce and understand utterances? How is this linguistic knowledge organized, and how does it interact with other types of information during sentence processing? Although human language is complex and ambiguous, speakers usually process language very rapidly and effortlessly, which suggests that language relies on probabilistic information. My research focuses on how sentence structure is linked to sentence meaning, with particular interest on unbounded dependency constructions. I use corpora and controlled psycholinguistic experimentation to determine what is the division of labor between syntax, semantics, pragmatics, and cognition in such phenomena. I have also focused on other topics, involving coordination, ellipsis, extraposition, and linearization. More broadly, I am interested in grammatical theory, specially in formally explicit models of language that are consistent with what is known about human cognition and computational tractability. I have specialized in constraint-based grammatical frameworks like HPSG and SBCG because their surface-driven nature is compatible with psycholinguistic models of language comprehension and production (see here and here for more discussion, and see here for an example), and because the formal explicitness of such surface-oriented theories allows the implementation of efficient large-scale computational grammars, useful not only for improving language processing technology and benefiting society at large (e.g. translation and question answering systems), but also for research purposes such as grammar comparison, consistency checking, and hypothesis testing. Before coming to UB, I was a visiting researcher at CSLI (2004–2007), and prior to that I was an assistant researcher at CLUL.