Language change and linguistic theory

The case of archaic Indo-European conjunction


Many archaic Indo‐European languages exhibit a system of dual conjunction in which they possess both a head‐initial exponent (e.g., Latin et) and an enclitic exponent (e.g., Latin ⸗que). Mitrović (2014) and Mitrović & Sauerland (2016) argue that these two types of conjunctions instantiate the universal lexical categories J and μ. Several syntactic, semantic, and morphological properties are argued to result from this categorial distinction. For instance, J conjunctions are claimed to lack additive readings (i.e., ‘too, also’). Diachronically, head‐initial conjunctions are predicted to originate from combinations of J and μ heads (Mitrović & Sauerland 2016: 489). A closer look at the data reveals that neither of these predictions is borne out. The empirical motivation for the J/μ distinction is in fact paltry. I therefore offer a new history of Indo‐European conjunction, in which I demonstrate first that the earliest attested Indo‐European languages do not have this double system of conjunction. It is rather an innovation that resulted from the recruitment of new conjunctions across the family. These new conjunctions developed primarily from additive focus operators, and not from combinations of J and μ heads. Empirical issues aside, the analysis of Mitrović (2014) and Mitrović & Sauerland (2016) raises deeper questions about the relationship between linguistic theory and language change. I argue that some of the properties of natural language that Mitrović (2014) and Mitrović & Sauerland (2016) assign to Universal Grammar are better analysed as epiphenomena of language change.

Transactions of the Philological Society 117:1–34