Elamur Research⌕
☰
home / papers / 2019 / From Insanely Jealous to Insanely Delicious: Computational Models for the Semantic Bleaching of English Intensifiers

From Insanely Jealous to Insanely Delicious: Computational Models for the Semantic Bleaching of English Intensifiers

Paper · 2019 · Yiwei Luo, Dan Jurafsky and Beth Levin

Summary

The paper presents what the authors call the first computational treatment of semantic bleaching, the process by which a word loses parts of its original meaning and becomes more abstract. The test case is English de-adjectival "-ly" adverbs that begin as manner adverbs and drift toward pure intensifiers, as in awfully behaved → awfully nice. Using HistWords embeddings (W2V and SVD) and the English fiction part of Google Books n-grams and Google Syntactic ngrams, aggregated by decade from 1850 onward, they turn three textbook symptoms of bleaching into measurable quantities:

  • weakening ties to the root meaning;
  • growing closeness to very;
  • wider collocational range.

They compare 250 intensifiers with a frequency-matched control set of adverbs. The intensifiers move significantly more than the controls. A second study asks why bleaching starts. Adverbs that tend to modify adjectives semantically close to themselves (insanely jealous) bleach faster afterwards. This supports a reanalysis account built on bridging contexts. The work matters because it treats bleaching as a gradual, quantifiable process, separate from generic semantic shift.

Key points

  • Data:
    • Google Books English fiction n-grams, with HistWords embeddings trained on that same portion.
    • Google Syntactic ngrams, used for adverb–adjective pairs.
    • Both W2V and SVD variants were tested.
    • Analysis is by decade, restricted to 1850 onward because earlier data are sparse.
  • Word sets:
    • 250 intensifiers, drawn from Bolinger's (1972) root-adjective classes and extended with WordNet synonyms.
    • 178 frequency-matched control adverbs sampled from the BNC.
  • Metrics:
    • SimVery: cosine similarity to very.
    • SimLex: mean similarity to root-meaning lemmas taken from WordNet and the OED.
    • TypeDiv: count of distinct adjective types modified.
    • Breadth: negative odds-weighted mean pairwise similarity among the top 50 modified adjectives, ranked by log odds.
  • Study 1 result: SimVery, SimLex and TypeDiv rise or fall significantly more for intensifiers than for controls. SimLex separates the two classes even without a control set, so the authors recommend it as the most general bleaching measure.
  • Study 2 result:
    • SimAdjMod is the odds-weighted similarity of an adverb to the adjectives it modifies. It correlates positively with the decade-to-decade rate of bleaching on all three metrics.
    • Intensifiers modify more mutually similar adjectives than control adverbs do.
    • Intensifiers modify adjectives, relative to verbs, more often than controls do (t = 4.4, p < 1e-04).

Interesting ideas

  • I1. Corpus growth inflates diachronic metrics, so the authors judge change only by comparing target words against a frequency-matched control group, never by raw trends alone.
  • I2. Before averaging similarities, words used as semantic anchors are first checked to be stable over time against determiners, numerals and pronouns, which removes noise from moving reference points.
  • I3. Bleaching is defined as distance from the word's own root-meaning lemma set, not as closeness to a target prototype, which lets the same measure detect semantic loss for many kinds of words.
  • I4. Productivity is measured as how semantically spread out a word's collocates are, using weighted pairwise similarity, rather than only by counting distinct types.
  • I5. A "bridging context" arises when an adverb premodifies an adjective close to it in meaning, and such contexts are where a manner reading turns into an intensity reading.

For Mova

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

  • I1 (learn): Any metric Mova tracks across data versions or model iterations should be compared against a matched control set of items, so that drift caused by data size is not mistaken for real improvement.
  • I2 (world): When compiling ontology or value anchors from WordNet/FrameNet, Mova could keep only anchor terms whose behaviour is stable across sources, making derived similarity scores more trustworthy.
  • I3 (frames): Comparing a modifier against its root lemma set could signal when an adverb like terribly or insanely carries no manner role and should be read only as degree.
  • I4 (world): Compiled verb-class and modifier knowledge could store a collocate-spread score, telling the system how freely a word combines and how literally to interpret it.
  • I5 (frames): When an advmod adverb is semantically close to its adjectival head (clearly obvious), prag could prefer an emphasis/degree interpretation and skip creating a separate manner frame element.

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

🎓Stanford UniversityStanford · United States310 works · 284,480 citations

📚 Shelves 6 Computer Science · 9,086 works Artificial Intelligence · 8,889 works Arts and Humanities · 9,296 works Language and Linguistics · 9,206 works Psychology · 798 works Experimental and Cognitive Psychology · 463 works

🧭 Topics 1 English · 1,373 works

Cite

Cite as Yiwei Luo, Dan Jurafsky and Beth Levin. 2019. From Insanely Jealous to Insanely Delicious: Computational Models for the Semantic Bleaching of English Intensifiers. Proceedings of the 1st International Workshop on Computational Approaches to Historical Language Change.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/w19-4701

@inproceedings{luo2019insanely,
  title = {{From Insanely Jealous to Insanely Delicious: Computational Models for the Semantic Bleaching of English Intensifiers}},
  author = {Yiwei Luo, Dan Jurafsky and Beth Levin},
  booktitle = {Proceedings of the 1st International Workshop on Computational Approaches to Historical Language Change},
  year = {2019},
  doi = {10.18653/v1/w19-4701},
  url = {https://doi.org/10.18653/v1/w19-4701}
}

References

34 references, 6 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. William L. Hamilton, Jure Leskovec, and Dan Jurafsky. 2016b. Diachronic word embeddings reveal statistical laws of semantic change. In Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1489–1501, Berlin, Germany. Association for Computational Linguistics. ↗ our card
  2. Andrea Beltrama and M Ryan Bochnak. 2015. Intensification without degrees cross-linguistically. Natural Language & Linguistic Theory, 33(3):843–879. ↗
  3. Andrea Beltrama and Laura Staum Casasanto. 2017. Totally tall sounds totally younger: Intensification at the socio-semantics interface. Journal of Sociolinguistics, 21(2):154–182. ↗
  4. Paul J Hopper and Elizabeth Closs Traugott. 2003. Grammaticalization. Cambridge: Cambridge University Press. find
  5. Dwight Bolinger. 1972. Degree Words. Paris: Mouton. find
  6. Vivek Kulkarni, Rami Al-Rfou, Bryan Perozzi, and Steven Skiena. 2015. Statistically significant detection of linguistic change. In Proceedings of the 24th International Conference on World Wide Web, pages 625–635. International World Wide Web Conferences Steering Committee. ↗
  7. Michel Bréal. 1964. Semantics: Studies in the science of meaning. New York: Dover. find
  8. Joan Bybee. 1995. Regular morphology and the lexicon. Language and Cognitive Processes, 10(5):425–455. ↗
  9. Yuri Lin, Jean-Baptiste Michel, Erez Aiden Lieberman, Jon Orwant, Will Brockman, and Slav Petrov. 2012. Syntactic annotations for the google books ngram corpus. In Proceedings of the ACL 2012 System Demonstrations, pages 169–174, Jeju Island, Korea. Association for Computational Linguistics. find
  10. Joan L Bybee, Revere Dale Perkins, and William Pagliuca. 1994. The evolution of grammar: Tense, aspect, and modality in the languages of the world, volume 196. The University of Chicago Press: Chicago. find
  11. Mark Davies. 2010-. The Corpus of Historical American English (COHA): 400 million words, 1810-2009. Available online at https://corpus.byu.edu/coha/. ↗
  12. Gunter Lorenz. 2002. Really worthwhile or not really significant? A corpus-based approach to the delexicalization and grammaticalization of intensifiers in Modern English. New reflections on grammaticalization, 49:143. ↗
  13. Haim Dubossarsky, Daphna Weinshall, and Eitan Grossman. 2017. Outta control: Laws of semantic change and inherent biases in word representation models. In Proceedings of the 2017 Conference on Empirical Methods in Natural Language Processing, pages 1136–1145, Copenhagen, Denmark. Association for Computational Linguistics. ↗ our card
  14. Yiwei Luo and Yang Xu. 2018. Stability in the temporal dynamics of word meanings. In CogSci. find
  15. Ronald Macaulay. 2006. Pure grammaticalization: The development of a teenage intensifier. Language Variation and Change, 18(3):267–283. ↗
  16. Nicholas Evans and David Wilkins. 2000. In the mind’s ear: The semantic extensions of perception verbs in Australian languages. Language, 76:546– 592. find
  17. George Miller. 1998. WordNet: An Electronic Lexical Database. MIT press. find
  18. Sunny Mitra, Ritwik Mitra, Martin Riedl, Chris Biemann, Animesh Mukherjee, and Pawan Goyal. 2014. That’s sick dude!: Automatic identification of word sense change across different timescales. In Proceedings of the 52nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1020–1029, Baltimore, Maryland. Association for Computational Linguistics. ↗
  19. Lea Frermann and Mirella Lapata. 2016. A Bayesian model of diachronic meaning change. Transactions of the Association for Computational Linguistics, 4:31–45. ↗ our card
  20. Yoav Goldberg and Jon Orwant. 2013. A dataset of syntactic-ngrams over time from a very large corpus of English books. In *SEM. our card
  21. Kristina Gulordava and Marco Baroni. 2011. A distributional similarity approach to the detection of semantic change in the Google books ngram corpus. In Proceedings of the GEMS 2011 Workshop on GEometrical Models of Natural Language Semantics, pages 67–71, Edinburgh, UK. Association for Computational Linguistics. our card
  22. Marcin Morzycki. 2008. Adverbial modification of adjectives: Evaluatives and a little beyond. Event structures in linguistic form and interpretation, 5:103. ↗
  23. Rick Nouwen. 2011. Degree modifiers and monotonicity. In Vagueness and language use, pages 146–164. Springer, Palgrave. ↗
  24. William L. Hamilton, Jure Leskovec, and Dan Jurafsky. 2016a. Cultural shift or linguistic drift? comparing two computational measures of semantic change. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 2116–2121, Austin, Texas. Association for Computational Linguistics. our card
  25. Mark Pagel, Quentin D Atkinson, and Andrew Meade. 2007. Frequency of word-use predicts rates of lexical evolution throughout Indo-European history. Nature, 449(7163):717. ↗
  26. A Carita Paradis. 1997. Degree modifiers of adjectives in spoken British English, volume 92. Lund: Lund University Press. Full set of 250 intensifiers abnormally abundantly acutely alarmingly annoyingly atrociously beautifully breathtakingly clearly comically conspicuously criminally decadently deeply delectably depressingly disastrously dismayingly disturbingly dramatically embarrassingly enormously epically exceedingly excessively extensively extremely fiercely frightfully fundamentally gorgeously grossly heavily honestly hugely immoderately improperly indecently inexcusably insanely intolerably lavishly literally madly majorly Hans Peters. 1994. Degree adverbs in Early Modern English, volume 13. Berlin: Walter de Gruyter. find
  27. Christian Ramiro, Mahesh Srinivasan, Barbara C. Malt, and Yang Xu. 2018. Algorithms in the historical emergence of word senses. Proceedings of the National Academy of Sciences, 115(10):2323–2328. ↗
  28. Alex Rosenfeld and Katrin Erk. 2018. Deep neural models of semantic shift. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Papers), pages 474–484, New Orleans, Louisiana. Association for Computational Linguistics. find
  29. Skipper Seabold and Josef Perktold. 2010. Statsmodels: Econometric and statistical modeling with python. In 9th Python in Science Conference. find
  30. John Simpson, Edmund SC Weiner, et al. Oxford English Dictionary online. Accessed October 2018. find
  31. Gustaf Stern. 1931. Meaning and change of meaning. Bloomington: Indiana University Press. find
  32. Eve Sweetser. 1989. From etymology to pragmatics: Metaphorical and cultural aspects of semantic structure, volume 54. Cambridge: Cambridge University Press. find
  33. Sali A Tagliamonte. 2008. So different and pretty cool! Recycling intensifiers in Toronto, Canada. English Language & Linguistics, 12(2):361–394. find
  34. Yang Xu and Charles Kemp. 2015. A computational evaluation of two laws of semantic change. In CogSci. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/luo-2019-insanely-jealous-insanely-delicious-computational/