The paper focuses on which role is given to elided sentence participants in coreference chains, i.e. whether (and to which degree) the participants that are present only implicitly in the surface layer are involved in relations of textual and grammatical coreference. Generally, the paper introduces the methods how it is possible to examine the interplays of different language phenomena in corpus data of the Prague Dependency Tree-bank containing multilayer annotation.
It is only the aid of large corpora (several billions of words) that enables us to discover some intuitively and spontaneously followed rules of grammar. Different kinds of ellipsis and non-ellipsis (repetition of a word or a nominal phrase, which - under some conditions - can be omitted) can also be governed by such rules. The corpus findings of sentence structures as (1) Zastavila se a podívala se na hodinky or (2) Zastavila se a podívala na hodinky (She stopped and looked at her watch) have clearly shown that ellipsis as well as repetition is a (strict) rule under specific semantic and syntactic conditions.
In the stylization of spontaneous, non-prepared spoken expression in contemporary literary texts (including prose, drama, and even comics), one of the most striking syntactic elements to emerge is the one-syllable word (se, si, sem, sme, ste, mě, mi, tě, ti, bych, by…) at the beginning of an utterance or turn. Sgall and Hronek (1992) call these words enclitics or proclitics, though according to J. Toman (2002) or A. Svoboda (2002) they are not clitics. However, all of these authors consider them to be the result of word-order inversion (Se mu to nepovedlo = ''Nepovedlo se mu to'') or of processes of ellipsis (Bych si taky myslel = ''To bych si taky myslel''). Yet there are likely also other motivations, e.g. phonetic ones related to the specific techniques of spoken expression. This type is common in our research, for example, in the communication of young people engaging in internet chat, i.e. in written texts strongly influenced by spoken expression. With the help of corpora of spoken Czech and literary texts from the Czech National Corpus (SYN2000, SYN2005, SYN2010), the authors found that these one-syllable beginnings of utterances or turns are a striking and non-detachable sign of contemporary colloquial Czech, of authentic Czech dialogues - and thus not merely a myth heavily sustained by Czech authors of literary texts, who make efforts to stylize casual expression.