Abstract Heinz and Idsardi (2013) draw attention to a profound computational difference between syntax and phonology: phonology only requires regular computations over strings (Johnson 1972; Kaplan and Kay 1994), whereas syntax involves non-regular computations over strings (Chomsky 1956; Huybregts 1984; Shieber 1985; Radzinski 1991; Michaelis and Kracht 1997; Kobele 2006). We offer an alternative picture that is synthesized from several recent works in computational linguistics and closes the apparent chasm between phonology and syntax. The computational complexity of a linguistic domain can be measured along two axes: the nature of the structural representations, and the power of the computations that manipulate these structures. The previously observed complexity differences between syntax and phonology (and morphology) can be recast entirely in terms of the data structures rather than the power of the computations. Phonology operates over string-like structures, whereas syntax uses trees of unbounded size, which grant it increased expressivity. But the dependencies we find in both involve similar computations that fall into the formal class tier-based strictly local. Among other things, this entails that the linguistic notion of relativized locality plays a crucial role across language modules.
@Misc{GrafEtAl18Meertenstalk,
author = {Graf, Thomas and
Aks\"{e}nova, Al\"{e}na and
Baek, Hyunah and
De Santo, Aniello and
Dolatian, Hossep and
Moradi, Sedigheh and
Rawski, Jon and
Yang, Suji and
Heinz, Jeffrey},
title = {Tiers and Relativized Locality Across Language Modules},
year = {2018},
note = {Slides of a talk given at the 1-day workshop {P}arallels Between Phonology and Syntax, {J}uly 9, {M}eertens {I}nstituut, {A}msterdam, {N}etherlands}
}