Croatian adjectives have two forms in the masculine gender: the Long (L) form and the Short (S) form. The main distributional difference is that the Short adjective can be in predicative position and the Long one cannot, while both can be in attributive position. This difference between attributive and predicative can be related to a variety of other cross-linguistic distributions concerning adjectives (Alexiadou 2001). It has been stated (Aljović 2002, Trenkić 2004) for (Serbo-)Croatian that the two forms mark a distinction in definiteness or specificity with the long one being [+DEF/+SPEC] and the Short one [-DEF/-SPEC].
A survey on 32 adults was conducted in order to obtain more information about the distribution of the two forms in general; to find out whether it is definiteness or specificity that is being marked by the Long form; and to check whether one of the forms (the Long one) can function as a subject of a sentence in the absence of a noun. The results of the statistical analysis show that the predicative/attributive distinction is not as strict as described in the previous literature (Silić and Pranjković 2007); and that the Long form is related to specificity but does not express it.
I propose an analysis that builds on cross-linguistic parallelisms described in Alexiadou 2001 and I propose that Croatian Long and Short distributional patterns are caused by the same factors as Noun Raising in Romance and Determiner Spreading in Greek, even though we find that this is not as strict as in those languages. However, it is only with expanding our cross-linguistic analysis to more languages that we can fully understand the nature of what these subtle differences of adjectives mark.