In what follows we describe how prefixation is annotated in the database

The general guideline that we followed in determining if a verb is prefixed or not is ‘if it can be taken to be a prefix, then it is a prefix’ where the starting point were existing lists of prefixes in BCS and Slovenian (Babić 1986, Klajn 2002 for BCMS, Toporišič 2000 for Slovenian).

To this end, the first step was to determine if a verb that seems to be prefixed has an unprefixed pair, e.g. Slovenian prebrati ‘to finish reading’ has the pair brati ‘to read’, the BCS pročitati ‘to finish reading’ has the pair čitati ‘to read’.

If the seemingly prefixed verb does not have an unprefixed pair, we have checked if a verb with the same root, but a different prefix exists. If it does, the verb was taken to be prefixed and annotated as such. Examples of this kind are, for example, verbs with the root četi:za-četi ‘start’, po-četi ‘to do’, na-četi ‘start’, pri-četi ‘start’, za-četi ‘start’ (even though četi does not exist).

In general, the annotation shows if the verb is prefixed and, if it is, which prefixes are present on the verb (see below). The column Prefixed_verb shows if a verb has prefix or prefixes (1) or not (0).

This information is relevant for the column Simplex_verb too - in this colomn either a prefix or any type of an affix, it is marked with a 0.

Prefixes are marked in 4 additional columns (Prefixes (list), 1, 2 and 3). In the column Prefixes (list), we show all the prefixes that the verb contains, separated by a “+”. The next three columns show specific prefixes, as well as the position of each individual prefix with respect to the root, where column 1 refers to the position closest to the root, column 2 the second one, and column 3 the prefix that is most distant from the root. This means that the order of prefixes in the columns is reversed compared to their linear order in the verb itself (and in column Prefixes (list)). If there is no prefix in the respective position, we annotate this with 0. More examples are given below.

Note that this annotation only marks the linear position of the prefix and that we are making no claims about the syntactic position of each prefix (whether it is lexical, intermediate, super-lexical etc.).

Some verbs cannot be annotated as prefixed using the criteria described above, however, they seems to contain a prefix. This is marked in the Potentially prefixed column. Verbs that are potentially prefixed are annotated with 1 in this column.

For instance, in BCS, the verb zataškati ‘to cover up.pfv’ seems to include the prefix za- (which seems to retain the same meaning as in some other verbs with za-), but the base taškati is not attested independently nor with other prefixes.

Also marked in the Potentially prefixed column are verbs that seem prefixed, but no unprefixed pair or a pair with a different prefix exist, however there exist a noun that has the same base. Consider the example predsedovati 'to chair.ipfv'in Slovenian - the perfective pair *predsedeti is not attested, nor the unprefixed version sedovati (note that the imperfective sedeti 'to sit' exists). However, the noun predsednik 'president/chair' exists.

Finally, verbs that could be taken as derived form a preposition are marked with a 1 in the Potentially prefixesd column. There is, however, a difference in marking this in the Roots column.

The following table gives an overview. Note that verbs that are annotated with 1 in the Potentially prefixed are not taken to be Simplex (i.e., they are annotated with a 0 in the Simplex verb column).

In some verbs it is unclear whether the item in question is a single prefix or if it could be further separated. One such example is Slovenian izpod- where prefixes iz- and pod- also exist (as do prepositions iz, pod and izpod). In the database we decompose these prefixes. note however that this increases the number of multiply prefixed verbs.