Elamur Research⌕
☰
home / papers / 2018 / Compilation of Corpora for the Study of the Information Structure–Prosody Interface

Compilation of Corpora for the Study of the Information Structure–Prosody Interface

Paper · 2018 · Alicia Burga, Mónica Domínguez, Mireia Farrús et al.

Summary

Linguistic theory holds that the split of a sentence into theme (what is talked about) and rheme (what is said about it) is reflected in intonation. Empirical support for this claim is thin. Existing studies rarely use more than a couple of speakers, usually measure only F0, and rely on a binary theme/rheme division that is not formalized for computation. The authors offer a reproducible procedure for building corpora that align hierarchical thematicity with automatically measured prosody. Thematicity follows Mel'čuk's Meaning-Text Theory, where specifiers exist and thematicity can nest inside embedded propositions. Text annotated under the Bohnet et al. (2013) guidelines is converted into Praat TextGrid tiers. The audio is processed in Praat on the Web, where modular scripts attach acoustic and structural features to each span and export a csv for statistics or classifiers. The procedure is demonstrated on a small read-speech corpus of American English and checked through classification experiments. The authors present it as the first English resource pairing prosody with hierarchical thematicity, useful for more expressive, communication-aware speech synthesis.

Key points

  • Thematicity has three span types: theme, rheme and specifier (context-setting material such as "he says"). It is defined per proposition, so embedded clauses carry their own level-indexed spans, for example T1(P2) or SP1.
  • The pipeline has six modules:
    • converting thematicity-annotated txt into a TextGrid;
    • building Praat pitch and intensity objects from the wav;
    • annotating prosodic and linguistic features (modules 3–6);
    • exporting the results to csv.
  • Every level of the hierarchy is represented by a pair of tiers, the first marking proposition spans (e.g. L1P) and the second holding the thematicity labels (e.g. L1T). The main proposition P1 is marked explicitly because relative measures are computed against it.
  • Prosodic parameters cover F0, intensity and rhythm. Absolute values include mean/std/min/max, duration and speech rate in words or syllables per second. Relative values are z-scores and the minmax-normalized time position of the F0 peak and the intensity minimum. Linguistic features are word count, number of embedded spans, and the label string at the L1P level.
  • Data and module code are released at github.com/TalnUPF/compilationISpros. The platform builds on Praat on the Web and an automatic prosody tagger by the same group.

Interesting ideas

  • I1. Hierarchy is encoded as a pair of parallel tiers per nesting level, one for proposition boundaries and one for information-structure labels, which keeps embedded structure in a flat, time-aligned format that standard tools can read.
  • I2. The specifier is a third category beside theme and rheme, capturing material that sets the context rather than being topic or comment.
  • I3. Positions of events inside a span are scaled to 0–1 by the span's start and end (minmax), making them comparable across spans of different lengths.
  • I4. Every measure is stored both raw and z-normalized against the whole sample, so features stay interpretable yet comparable across speakers or sources.
  • I5. The pipeline is modular and ends in a plain per-span csv, so any statistics package or classifier can test a hypothesis without touching the annotation code.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (frames): theme/rheme/specifier spans for embedded clauses could be stored as level-indexed layers over the UD tree, so each frame knows its informational role within its own proposition.
  • I2 (frames): treating reporting clauses like "he says" as specifiers would help separate the speech-act source from the asserted content when attributing claims.
  • I3 (frames): the relative position of a head or focus word within its clause could become a length-independent feature for guessing information structure.
  • I4 (learn): reporting per-genre statistics both raw and corpus-normalized would make evaluation drift across text types easier to spot.
  • I5 (learn): exporting per-span frame and structure features to a flat table would allow cheap, transparent hypothesis tests and negative-control checks outside the main system.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Universitat Pompeu FabraBarcelona · Spain36 works · 11,433 citations

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Arts and Humanities · 9,296 worksLanguage and Linguistics · 9,206 worksPsychology · 798 worksExperimental and Cognitive Psychology · 463 works

🧭 Topics 1Research subjects and languages the work is about.Speech & spoken language · 233 works

Cite

Cite as Alicia Burga, Mónica Domínguez, Mireia Farrús and Leo Wanner. 2018. Compilation of Corpora for the Study of the Information Structure–Prosody Interface. Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018).

The paper itself, not this page. BibTeX from the library's record of the work.

@inproceedings{burga2018compilation,
  title = {{Compilation of Corpora for the Study of the Information Structure–Prosody Interface}},
  author = {Alicia Burga, Mónica Domínguez, Mireia Farrús and Leo Wanner},
  booktitle = {Proceedings of the Eleventh International Conference on Language Resources and Evaluation (LREC 2018)},
  year = {2018},
  doi = {10.63317/32pkapa586ap},
  url = {https://doi.org/10.63317/32pkapa586ap}
}

References

29 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Baumann, S. (2012). The Intonation of Givenness. Evidence from German. Max Niemeyer Verlag, Berlin, Boston. find
  2. Beckman, M. E. and Pierrehumbert, J. (1986). Intonational Structure in Japanese and English. Phonology Yearbook, 3:255–310. ↗
  3. Bock, J. K., Mazzella, J. R., and Bock, K. (1983). Intonational marking of given and new information: Some consequences for comprehension. Memory & Cognition, 11(1):64–76. ↗
  4. Boersma, P. and Weenink, D. (2017). Praat: doing phonetics by computer [Computer program], http://www.praat.org/, version 6.0.14. ↗
  5. Bohnet, B., Burga, A., and Wanner, L. (2013). Towards the annotation of penn treebank with information structure. In Proceedings of the Sixth International Joint Conference on Natural Language Processing, pages 1250– 1256, Nagoya, Japan.
  6. Brown, G. (1983). Prosodic structure and the given/new distinction. In A. Cutler et al., editors, Prosody: Models and Measurements, pages 67–77. Springer, Berlin, Heidelberg. ↗
  7. Büring, D. (2003). On d-trees, beans, and b-accents. Linguistics & Philosophy, 26(5):511–545. find
  8. Calhoun, S. (2010). The centrality of metrical structure in signalling information structure: A probabilistic perspective. Language, 1(86):1–42. find
  9. Chafe, W. L. and Li, C. N. (1976). Givenness, contrastiveness, definiteness, subjects, topics, and point of view in subject and topic. Subject and Topic, pages 25–55. find
  10. Charniak, E. and al., E. (2000). BLLIP 1987-89 WSJ Corpus Release 1 LDC2000T43. Linguistic Data Consortium, Philadelphia. find
  11. Clark, H. H. and Haviland, S. E. (1977). Comprehension and the given-new contract. Discourse production and comprehension. Discourse processes: Advances in research and theory, 1:1–40. find
  12. Domı́nguez, M., Farrús, M., Burga, A., and Wanner, L. (2014). Towards Automatic Extraction of Prosodic Patterns for Speech Synthesis. In Proceedings of the 7th International Conference on Speech Prosody, pages 1105– 1109, Dublin, Ireland. find
  13. Domı́nguez, M., Farrús, M., Burga, A., and Wanner, L. (2016a). Using hierarchical information structure for prosody prediction in content-to-speech applications. In Proceedings of the 8th International Conference on Speech Prosody, pages 1019–1023, Boston, USA. find
  14. Domı́nguez, M., Farrús, M., and Wanner, L. (2016b). An Automatic Prosody Tagger for Spontaneous Speech. In Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics, pages 377– 387, Osaka, Japan. find
  15. Domı́nguez, M., Latorre, I., Farrús, M., Codina, J., and Wanner, L. (2016c). Praat on the Web: An Upgrade of Praat for Semi-Automatic Speech Annotation. In Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: System Demonstrations, pages 218–222, Osaka, Japan. find
  16. Domı́nguez, M., Farrús, M., and Wanner, L. (2017). A Thematicity-based Prosody Enrichment Tool for CTS. In Accepted to show and tell demonstrations at INTER-SPEECH17, Stockholm, Sweden. find
  17. Féry, C. and Kügler, F. (2008). Pitch accent scaling on given, new and focused constituents in German. Journal of Phonetics, 36(4):680–703. ↗
  18. Féry, C. (2013). Focus as prosodic alignment. Natural Language & Linguistic Theory, 31(3):683–734. ↗
  19. Fowler, C. A. and Housum, J. (1987). Talkers’ signaling of ”new” and ”old” words in speech and listeners’ perception and use of the distinction. Journal of Memory and Language, 26(5):489–504. ↗
  20. Haji-Abdolhosseini, M. and Müller, S. (2003). Constraint-Based Approach to Information Structure and Prosody Correspondence. In Proceedings of the 10th International Conference on Head-Driven Phrase Structure Grammar, pages 143–162. CSLI Publications. find
  21. Hall, M., Frank, E., Holmes, G., Pfahringer, B., Reutemann, P., and Witten, I. H. (2009). The WEKA Data Mining Software: An Update. SIGKDD Explorations, 11(1). ↗
  22. Mel’čuk, I. A. (2001). Communicative Organization in Natural Language: The semantic-communicative structure of sentences. Benjamins, Amsterdam, Philadephia. find
  23. Meurers, D., Ziai, R., Ott, N., and Kopp, J. (2011). Evaluating Answers to Reading Comprehension Questions in Context: Results for German and the Role of Information Structure. In Proceedings of the TextInfer 2011 Workshop on Textual Entailment, TIWTE ’11, pages 1– 9, Stroudsburg, PA, USA. Association for Computational Linguistics. ↗
  24. Schwarzschild, R. (1999). Givenness, avoidf and other constraints on the placement of accent. Natural Language Semantics, 7(1):141–177. ↗
  25. Silverman, K., Beckman, M., Pitrelli, J., Ostendorf, M., Wightman, C., Price, P., Pierrehumbert, J., and Hirschberg, J. (1992). TOBI: A Standard for Labeling English Prosody. In 2nd International Conference on Spoken Language Processing (ICSLP 92), pages 867– 870, Banff, Canada, October. find
  26. Steedman, M. (2000). Information structure and the syntax-phonology interface. Linguistic inquiry, 31(4):649–689, Fall. ↗
  27. van Donselaar, W. and Lentz, J. (1994). The function of sentence accents and given/new information in speech processing: different strategies for normal-hearing and hearing-impaired listeners? Language and Speech, 37(4):375–391. ↗
  28. Vanrell, M., Mascaró, I., Torres-Tamarit, F., and Prieto, P. (2013). Intonation as an Encoder of Speaker Certainty: Information and Confirmation Yes-No Questions in Catalan. Language and Speech, 56(2):163–190. find
  29. Xu, Y. (1999). Effects of tone and focus on the formation and alignment of f0 contours. Journal of Phonetics, 27(1):55–105. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/burga-2018-compilation-corpora-study-information-structure/