Subject Case and Finiteness: A Corpus-Based Study on Microvariation and Historical Development of Modal Constructions in Croatian
Part 2: The Historical Development

The second part of the contribution is dedicated to the question of how the considerable microvariation of the encoding of subject and finiteness in Croatian modal constructions described in the first part came about. More precisely, we analyse the changes a) in the inventory of modal predicates and b) in the distribution of the constellation types attested in modern Croatian language use. For this specific purpose, we built a corpus resource for the history of Croatian CroDi (available on-line: http://hu.berlin/crodi). In the first section, we describe the principles of corpus compilation, the corpus architecture and the texts included. Currently, CroDi comprises texts ranging from the 16th to the 18th century, which thus form separate subcorpora for each century. We include texts written in Štokavian, Čakavian and Kajkavian repre- senting a major variety of genres.
Based on the analysis of all modal constructions found in CroDi (2.985 sentences) we detect certain changes in the modal system of Croatian across the centuries: 1) there are several lexical changes; some modal elements disappear (like uzmožan), others are innovations (e. g. morati, valjati) and a third class of elements undergoes semantic and mor- pho-syntactic changes. We specifically address the modal trebati which in modern usage is one of the most frequent ones, but is not attested in our historical corpus. There is evidence that it developed from the nominal form treba/triba not earlier than during the 18th century; 2) The distribution of the modals across parts of speech shows a certain conti- nuity, as from the very beginning verbs clearly dominate and nouns are subsequently getting out of use; 3) Concerning the development of the constellation types we show that the type Ljudi moraju raditi prevails already in the 16th century and becomes even more frequent later. The type Ljudima valja raditi is attested in the 16th century and remains in use with relatively low frequency. It is shown that the type Ljudi mora da su radili is not yet attested in CroDi, which leads us to the hypothesis that it must be an innovation not earlier than of the 19th century.
We have come to the conclusion that the encoding of the subject in Croatian is relatively stable, but we see changes in tense and agreement marking.