Alternative Lexicalizations of Discourse Connectives in Czech
Summary
This study looks at how Czech marks discourse relations with expressions that work like connectives but fall outside the standard connective classes. Following the PDTB notion of alternative lexicalizations (AltLex), the author searched the PDT 2.0 data for items annotators had flagged as AltLex. She cleaned the list and described the resulting expressions lexico-syntactically and semantically, comparing them with the English AltLex's reported by Prasad et al. for PDTB 2.0.
Key findings:
- Czech admits more syntactic classes for "classic" connectives than PDTB does, so a larger share of Czech AltLex's fall in admitted classes. Those items should probably be re-annotated as connectives.
- Most AltLex's form part of the clause structure. Prepositional phrases are the most frequent form.
- The expressions range from frozen to free combinations, with most lying near the free end.
- Like English AltLex's, they combine a relation signal with an anaphoric link to the first argument. That link may be obligatory, optional or impossible to express on the surface.
The author shows the current PDT annotation is inconsistent and incomplete. This argues for treating discourse markers as an open-ended class and informs future PDT releases.
Key points
- 306 tokens were flagged as AltLex in 43,955 PDT sentences. Removing misinterpreted cases left 261 tokens of 94 types.
- Consistency check on "speaking" (řečeno) expressions: 53 occurrences, 23 used as discourse markers. Of these, 3 were tagged AltLex, 7 tagged as connectives and 13 left unannotated.
- Tokens from syntactically admitted classes make up 26% in PDT versus 14.7% of 624 AltLex tokens in PDTB. By type, PDT has 37 admitted (39%) and 57 free (61%).
- By type, 78 (83%) are integrated into the clause and 16 (17%) are disjuncts or text-structuring items. Prepositional phrases form the largest group, with 33 types.
- Anaphoric reference by type: 41% optional, 31% obligatorily explicit, 28% impossible to express on the surface. Czech subject drop adds an optional-reference class not found in English.
Cited by
- Reflections on the Penn Discourse TreeBank, Comparable Corpora, and Complementary Annotation
- The Centre and Periphery of Discourse Connectives
Cite
@inproceedings{rysova2012alternative,
title = {{Alternative Lexicalizations of Discourse Connectives in Czech}},
author = {Rysová, Magdaléna},
booktitle = {Proceedings of the Eighth International Conference on Language Resources and Evaluation (LREC'12)},
year = {2012},
doi = {10.63317/25m37tzcwz99},
url = {https://doi.org/10.63317/25m37tzcwz99}
}References
- Biber, D. and Conrad, S. (1999). Lexical bundles in conversation and academic prose. In Out of Corpora: Studies in Honour of Stig Johansson, ed. by Hasselgard, H. and Oksefjell, S. Amsterdam: Rodopi. find
- Forbes-Riley, K., Webber, B. and Joshi, A. (2006). Computing discourse semantics: The predicate-argument semantics of discourse connectives in D-LTAG. Journal of Semantics, 23:55–106. find
- Halliday, M. A. K. and Hasan, R. (1976). Cohesion in English. London: Longman. find
- Knott, A. (1996). A Data-Driven Methodology for Motivating a Set of Coherence Relations. Ph.D. thesis, University of Edinburgh, Edinburgh. find
- Martin, J. R. (1992). English text: System and structure. Benjamins, Amsterdam. find
- Mladová, L., Zikánová, Š., Bedřichová, Z., Mírovský, J., Jínová, P., Zdeňková, J., Rysová, M., Hajičová, E. (2011). Manual for Annotation of Textual Relations (of Discourse) in the Prague Dependency Treebank. Praha: UFAL MFF. To appear. find
- Prasad, R. et al. (2010). Realization of Discourse Relations by Other Means: Alternative Lexicalizations. In COLING (Posters). 6.1 find our card
- Data sources Hajič, J. et al. (2006). Prague Dependency Treebank 2.0. Philadelphia: Linguistic Data Consortium. find
This page in Markdown → https://md.research.elamur.ai/free/en/papers/rysova-2012-alternative-lexicalizations-discourse-connectives-czech/