Doreen Georgi
(University of Potsdam)

Unifying everything? On the importance of recognizing Hidden Variability

A central aim of formal linguistics has been to describe humans’ general abstract linguistic knowledge as well as the observed variation between languages with a restricted (i.e., small and thus necessarily abstract) set of rules / constraints, whose interaction should ideally describe all and only the attested strings of a given language. In syntax (and other linguistic subfields), this has led to a certain “urge to unify” constructions, that is, to assign the same abstract analysis to strings that look very similar on the surface – and sometimes even to strings that do not look that similar. While analytical unification is certainly a conceptually desirable goal, it should not become a dogma but must always be supported by empirical evidence.This seems obvious, but my impression is that the empirical part is sometimes neglected for the sake of formal “fanciness” and the aim to have one and the same analysis of construction X for all language. In this panel discussion I would like to remind us that even two completely identical surface strings / constructions do not always have the same underlying syntax (or interpretation in semantics), a phenomenon that I will refer to as Hidden Variability. We need to keep the option of Hidden Variability in mind as an analytical option, and prioritize empirical evidence for or against the uniformity of two strings over conceptual elegance.

That Hidden Variability exists is well established in cross-linguistic comparison. For instance, it is generally accepted that languages can differ in the structure of their relative clauses (e.g., raising vs. head-external) or that not all VSO orders are created equal (in some languages they are derived by pure V-fronting, while in others they result from remnant predicate fronting) – claims based on solid empirical evidence. However, the possibility of Hidden Variability is often neglected when it comes to language-internal comparison. That is, a given string may be assigned different underlying structures even within a single language. A few claims for within-language Hidden variability can be found in the literature, e.g., Sichel (2014) on different structures available for headed relatives in Hebrew, Barros and Vicente (2011) on two co-existing derivations for Right Node Raising in English (see also Belk et al. 2023), den Dikken et al. (2000) on subtypes of specificational pseudoclefts in English, and Sande et al. (2019) on the non-uniformity of SAuxOV orders in West African languages.

I will provide more examples for language-internal Hidden variability from my own (co-authored) research (involving object-sharing serial verb constructions, non-structural case, resumption), and discuss examples where the desire to unify everything is misguided. The existence of Hidden variability potentially has further welcome consequences: it may explain divergent results for the same diagnostics applied to surface-identical/-similar strings within languages and within and across individuals, as well as apparently mixed properties of some constructions. Furthermore, the emergence of surface-stable patterns (with different underlying structures) may also reveal language-external causes of language change (e.g., processing ease or language contact). 

Based on these insights, I will advocate an approach to comparative syntax along the lines of Baker and McCloskey (2007) (“Middle Way”): instead of aiming for a large-scale typology (comparing hundreds of languages, which is necessarily more surface-oriented), we compare a given construction in a smaller set of languages (<10) in detail, applying the same diagnostics across all languages, so that we can be sure that we have properly understood its structure in each language before we start comparing languages.

References:

Baker, Mark C. and Jim McCloskey (2007): “On the relationship of typology to theoretical syntax.” Linguistic Typology 11(1), 285–296.

Barros, Matthew and Luis Vicente (2011): “Right Node Raising requires both ellipsis and multidominance.” University of Pennsylvania Working Papers in Linguistics 17(1), 1–19.

Belk, Zoe, Ad Neeleman and Joy Philip (2023): “What Divides, and What Unites, Right-Node Raising.” Linguistic Inquiry 54(4), 685–728.

den Dikken, Marcel, André Meinunger and Chris Wilder (2000): “Pseudoclefts and Ellipsis.” Studia Linguistica 54(1), 41–89.

Sande, Hannah, Nico Baier and Peter Jenks (2019): The syntactic diversity of SAuxOV in West Africa. In: E. Clem, P. Jenks and H. Sande (eds.), Theory and description in African Linguistics: Selected papers from the 47th Annual Conference on African Linguistics. Berlin: Language Science Press, pp. 667–701.

Sichel, Ivy (2014): “Resumptive Pronouns and Competition.” Linguistic Inquiry 45(4), 655–693.