Vincent Czarnecki

Hello!

My name is Vinny Czarnecki. 

I am a PhD candidate at Rutgers University in the Linguistics Department co-advised by Adam Jardine and Troy Messick. Before my time at Rutgers, I received a BS in Pure Mathematics and an MA in Computational Linguistics at Stony Brook University. My main research interests lie at the intersection of Linguistics, Mathematics, Theoretical Computer Science, and Philosophy. 

Broadly, I use mathematical tools such as finite model theory, formal language & automata theory, abstract algebra, etc. to understand the computational structure and complexity of linguistic competence. I am not interested in using computers to do language stuff, I’m interested in using computation (abstractly construed) as a means to describe how humans do language stuff.

More specifically as a generative linguist, I am interested in delineating between what is logically possible vs. what is attested across natural languages. In separate linguistic modules, this can take many forms: in phonology this could be possible/attested phonotactic patterns, in syntax this could be possible/attested types of head movement, in semantics this could be possible/attested meanings for certain kinds of lexical items, etc. But this often entails identifying some concrete computational property and using it to directly constrain our representations or computations. Regardless of the module in question, as a mathematical linguist I see what the tools above bring to bear on our theory construction. 

As a silly little example of what this can look like, take the familiar example of English plural formation. Where if a word:

• ends in a [-voice] sound, pluralize with 's'

• ends in a [+sibilant] sound, pluralize with 'iz' 

• otherwise, pluralize with 'z'

This necessarily only relies on things that are very close-by. But there is no language where you pluralize a word based on whether the number of vowels in it is even or odd. Namely, if a word:

• has an even number of vowels in it, pluralize with 's' 

• has an odd number of vowels in it, pluralize with 'z' 

Computationally speaking, the first one is really simple and the second is much more complex! This is one particular example from phonology, but the underlying thing that sets apart possibility and attestedness in this case is not anything about voiced segments or sibilants, but rather about parity. Similarly, imagine a syntactic constraint that says "a tree is only well-formed iff it has an even number of leaves", or a lexical item that means "individual x goes to location y an even number of times". These all share computational stuff despite their obvious representational differences. The tools above can express things like this very naturally, and it can help us theorize with a nice balance of expressivity and restrictiveness. 

While we use tools like these to formally understand representations and computations within linguistic modules, my main research interest is how we can employ them to understand computations between linguistic modules. In other words: what are logically possible vs. attested mappings at the linguistic interfaces? This largely entails constraining things in the same sense as above; however, at the interfaces, there are some other problems to consider. If human language is a modular system, then there are certain properties of computation that cannot be ignored if we want this to be reflected in our theories (e.g. domain specificity tells us that computation in a given module is limited to operating over its own primitives, so rules like "raise a L tone to a H tone in any word c-commanded by a +past T head"  should intuitively feel slightly gross since it simultaneously relies on phonological and syntactic primitives). This is simply another way of constraining our computations between modules and while these questions are reasonable to ask at any interface, I have found myself consistently thinking most about the morphosyntax-phonology interface. 

Lately, this has all mostly been leading me to think about things like (in no particular order):

model-theoretic syntax, model-theoretic phonology, logical interpretations, modularity, recursion, linearization, phrasal phonology, PF operations, head movement, reduplication, infixation, spatial semantics, formal learnability, underlying representation learning, etc.

When I’m not doing linguistics related stuff, I like to play the guitar (I play guitar Long Island Post-Rock band Golden Hymns Sing 'Hurrah’). I also made the album art shown here!