Mark C. Baker
(Rutgers University)

Comparing what? The layers of comparative syntax

Syntacticians from different generations often share a sense that there is a tendency in the field to “think smaller” than we used to. People are by in large not looking for a new principle of Universal Grammar or a sweeping new parameter, as generative research did in the 1980s. Rather, they are trying to get the right analysis of construction X in particular language Y—with a few remarks in the last section about how language Z might also have something like X. What does this impression (if it is true) mean? Is it a sign of growing maturity and realism in the field, or is it a sign of a loss of vision and ambition? Research must be done in some kind of modular fashion: on a scope in which projects can be carried out in finite time, with finite resources, and communicated to others with finite attention spans. But can we do realistically sized projects that are crafted in such a way that they have the potential to add up to something larger?  A classic American bumper sticker exhorted people: “Think globally; Act locally”. What would be a methodology for comparative syntax that takes seriously both halves of that exhortation?

Part of the answer, I still believe, is to practice “The Middle Way” laid out in Baker & McCloskey (2007) and Baker (2010), among others. There should be a “sweet spot” which comes from comparing (say) 5-10 languages from different families and/or areas, each studied to an intermediate degree of detail. That should help protect us on the one hand from errors of analysis (misunderstanding a construction from looking at a superficial description of it divorced from some grasp of the overall language) and on the other hand from errors of overgeneralization (wrongly taking something that is strikingly true in one or two languages to be universal).

However, I am increasingly aware of the fact that comparative syntax needs to be pursued in a layered fashion, if we are going to address the questions that I take to be central. To make this concrete, I reflect some on the different kinds of comparison that I found myself undertaking in my recent larger-scale exercise in comparative syntax (Baker 2024, to appear). My primary goal was to show that the same kinds of syntactic structures underlie indexical shift, logophoric pronouns, and upward C-agreement. All three kinds of constructions involve a null (“ghostly”) DP operator licensed by a C-like head in the periphery of an embedded clause which undergoes obligatory control by an argument of the matrix verb. That ghostly DP then either triggers agreement on C or binds a pronoun with matching features inside CP (or both). This hypothesis was intended as an extended counterargument to a view of Anand (2006) and Deal (2020), which attributes indexical shift and logophoric pronouns to two quite different mechanisms (and makes no connection to C-agreement). At the heart of the work is a less-common kind of comparative syntax: a close comparison of logophoric pronouns in Ibibio with both indexical shift in Magahi and C-agreement in Lubukusu. However, to succeed at this level, it also felt necessary to compare at a lower level: to compare logophoric pronouns in Ibibio with logophoric pronouns in Ewe, Yoruba, Edo, and Baatonum; to compare indexical shift in Magahi with indexical shift in Zazaki, Uyghur, Amharic, and Nez Perce; to compare C-agreement in Lubukusu with C-agreement in Kinande, Ibibio, and Kipsigis. In this, a kind of “Middle Way” comparative syntax of each individual construction supports a kind of “Middle Way” comparison across the constructions. (And the project had other layers of comparison too: lower level, higher level, and orthogonal.) In doing this, we should distinguish a low standard of comparison, which arguably makes an allegedly unified analysis too easy, from a gold-standard of comparison, which should be very compelling.

Finally, there is a role for keeping a vision of the great thing we are doing as we do the gritty work “in the trenches”. What is the value of (for example) showing that the same structures underlie C-agreement, logophoricity, and indexical shift? An answer: it contributes to understanding how complex structures can be built out of simple elements—in different ways in different languages, but ways that are often formally equivalent. And that is a nontrivial piece of showing how human language is generative and compositional (all human languages, as far as I know). And that is presumably a big piece of showing how human thought is generative, compositional, and unbounded, given that most human sentences express thoughts, and different sentences express different thoughts. I do not know of many things that humans can study that are nobler or more important than that, both in broad outline and in fine detail. It is worth remembering this ourselves and communicating it to others.