You are here

Motivating Osthoff's Law in Latin

Anthony Yates

Motivating Osthoff's Law in Latin

Standard handbooks (e.g. Weiss (2009:125-26), Meiser (2002:75-76)) now reflect the consensus view that Osthoff's Law (OL), the traditional designation for a process of vowel shortening targeting long vowels preceding a sonorant and another consonant (cf. Osthoff 1881) evident in several ancient Indo-European language branches, was operative within the history of Latin (cf. Collinge 1985:127). It is clear, moreover, that certain attested forms require its application both before and after sound changes which can be dated with respect to one another: a very early OL is necessary to account for the (historically) paradigmatically-isolated participles parentēs ‘parents' and Kalendae ‘first of the month' (cf. parāre, calāre), while Latin-internal monophthongization feeds its operation in ŭndecim `eleven', whose resulting short vowel is guaranteed by its development into Romance (Fr. onze, Sp. once).

The same handbooks therefore assume that OL applied not once, but at multiple points both within the history of the language, though exactly how many times and at what dates remains disputed. Most explicit on this point is Weiss (2009:ibid.), who posits three separate 'rounds' of OL shortening. While descriptively adequate, this analysis implicitly assumes that these three applications of OL are independent phonological processes. A more economical approach might begin with a 'persistent rule' in the terms of Myers (1991) or a 'surface filter', as Ringe (2006:118-121) has proposed for Siever's Law in Germanic, which would continuously remain active in the grammar and, as diachronic change creates new environments, apply wherever the structural description of the rule is met. Building on these ideas, I develop an analysis of OL in terms of Optimality Theory (Prince and Smolensky 1993), arguing that OL in Latin is driven by a high-ranking constraint against superheavy (viz. trimoraic) syllables. Byrd (2010a,b) has demonstrated the important role of this constraint in Proto-Indo-European, where it drives phonological processes such as Siever's Law. I argue that Latin stably continues the PIE situation in this respect, inheriting a grammar with this high-ranking constraint which motivates OL as a means of phonological repair for superheavy syllables.

This approach has at least two advantages. First, since constraints are necessarily violable, it is possible to address the problem of surface exceptions to a more general prohibition against trimoraic syllables, such as past participles showing Lachmann's Law (e.g. āctus `driven') (cf. Jasanoff 2004) and the productive class of -ēscō fientives (cf. Keydana 2012:5), by situating the constraint against superheavy syllables with respect to other (morpho)phonological processes in the grammar that may supercede, or outrank, it. Moreover, it allows the same constraint to be linked with superficially disparate moraic reduction phenomena within the attested record of the language, including the vowel shortening evident in (e.g.) Cl. Lat. frāter < OLat. frātēr (Pl. Aul. 140), and the ‘simplification' of final long diphthongs in (e.g.) Cl. Lat. them. dat. -ō < -ōi (DVENOI; CIL I2 4.3). I further argue that the operation of these processes in final syllables is due a Latin phonological innovation in its treatment of final consonants, specifically, the loss of word-final consonant extrametricality that is maintained in Greek (cf. Steriade 1982, 1988).

Session/Panel Title

Greek and Latin Linguistics

Session/Paper Number


© 2020, Society for Classical Studies Privacy Policy