West Germanic noun-noun compounds and the morphology-syntax trade-off

What it is about

Earlier work on the Parallel Bible Corpus found a trade-off: languages that carry more information in word order carry less in word structure, and vice versa. The paper asks whether English noun-noun compounds, written as separate words, are part of why English sits far from Dutch and German on this plane. The method is a word-pasting procedure: the most frequent noun-noun pairs in English Bible translations are merged into single words in growing steps. After each step, entropy estimates give the information in word order and in word structure. Pasting does shrink the gap between English and the other two languages, but the effect is small and the results are statistically inconclusive. The authors read this as a sign that orthographic word boundaries matter, not as a settled finding.

Note: the local text file holds only page markers, so this card follows the abstract in the library card, not the body.

Ideas

• I1. en/src/tok.rs: add a counted option that joins noun-noun pairs when the pair is frequent in a training corpus and the lexicon (en/src/dict.rs) has no entry for the joined form. Keep the counts in BTreeMap order and run twice before any claim.
• I2. Negative control: paste random noun pairs of the same frequency band instead of the most frequent ones. If the effect depends on frequency, the frequent-pair run should differ from the random-pair run; if not, the effect is an artefact.
• I3. Use the order-versus-structure entropy pair as a per-text diagnostic in tools/roundtrip, to see whether tree-to-text loss sits in word order or in word spelling.

For Mova

The chapter is 18. Compounding, conversion and the minor processes (compounds and their spelling) and 19. The lexicon: vocabulary, multiword expressions and register (multi-word units). The code is en/src/tok.rs for the joining option and en/src/lin.rs for the order tables. The measure: entropy of word order before and after joining, on the same Bible sample, with the random-pair control. The paper's gap is small and inconclusive, so expect a null result unless the control is clean.

Cite

The paper itself, not this card. BibTeX from the library's record of the work. DOI 10.18653/v1/2025.sigmorphon-main.2

@inproceedings{mosteiro2025west,
  title = {{West Germanic noun-noun compounds and the morphology-syntax trade-off}},
  author = {Mosteiro, Pablo and Blasi, Damián and Paperno, Denis},
  booktitle = {Proceedings of the 22nd SIGMORPHON workshop on Computational Morphology, Phonology, and Phonetics},
  year = {2025},
  doi = {10.18653/v1/2025.sigmorphon-main.2},
  url = {https://doi.org/10.18653/v1/2025.sigmorphon-main.2}
}

This page in Markdown → https://md.mova.elamur.ai/free/en/docs/papers/mosteiro-2025-west-germanic-noun-noun-compounds/