This dissertation is a fundamental rejection of the notion that is still being proffered (at least in Croatian grammars, cf. Barić 2007, Raguž 2010) that there are contextually based rules or tendencies for adjectival long-form (DL –e, -u/ G –a) (ALFA) usage in Contemporary BCS. While previous grammars may have stated the facts accurately for their synchronic period of BCS, many contemporary grammars have essentially “piggy-backed” on these prior grammars (e.g Shirokov & Gudkov 1977, Browne & Alt 2004), which viewed these ALFAs as adding a sense of “definiteness”, and subsequently have neglected the possibility of language change.
Alexander (2006) is the first to propose that ALFAs are more common in Croatian than in Bosnian or Serbian. My sociolinguistic questionnaire and analysis of the Croatian Language Repository, the Croatian National Corpus, and Wikipedia have all verified that ALFAs are more common in Croatian, although they can be heard rarely in Serbian (only for possessives and demonstratives), and slightly more frequently in Bosnian.
Furthermore, the empirical studies conducted in this investigation have informed the diachronic origin of ALFAs. It can certainly be said that ALFAs occur infrequently with descriptive adjectives, but exhibit strong patterns of variation for possessives and demonstratives. Moreover, for monosyllabic possessives and demonstratives (which are still monosyllabic after the addition of G and DL base forms –og and –om), when a preposition precedes the adjectival phrase the ALFAs are overwhelmingly preferred. I link this tendency to pronominal prepositional phrases which require long forms, e.g. o meni, but *o mi. This can be viewed as a “two-syllable” (or “two-footed”) constraint, which then spread to all prepositional phrases in the language via analogy. This “two-syllable” rule can be viewed as a restatement of “Wackernagel’s position”, which can be either prosodic or syntactic (Serbian vs. Croatian, for example), in syllabic terms, although it is possible that this process could have begun on the moric level o vās, but *o vas. In the absence of a noun this element either had already took on, or developed an anaphoric relationship to a following noun, and thus was reinterpreted as a marker of definiteness. Subsequent analogies spread ALFAs to the descriptives, although the “two-syllable” constraint would never have been violated for descriptives, given that all descriptive roots after the addition of ALFA bases would have at minimum two syllables already.
As repeated migrations of Serbs brought to Croatian dialects characteristic Serbian features, such as: the neo-Štokavian stress retraction, the da complementizer, etc., it appears they brought their ALFAs as well. Subsequently, the linguistic conformance behind the Vienna Literary Agreement of 1850 would have viewed the presence of these features in Croatian dialects as “unifying” elements.
Therefore, the statement made by Greenberg (2004) that the ALFAs were reintroduced as “puristic” elements in Croatian post-1991 can be considered an ironic twist in Croatian language policy, inasmuch as, from the diachronic perspective, ALFAs have been shown to be characteristic of 15th century Serbian, and then later Croatian.