Elamur Research⌕
☰
home / papers / 2021 / A Corpus-Based Approach to the Lemmatisation of Old English Superlative Adverbs

A Corpus-Based Approach to the Lemmatisation of Old English Superlative Adverbs

Paper · 2021 · Yosra Hamdoun Bghiyel

Summary

Old English has no fully lemmatised corpus or dictionary. Heavy dialectal spelling variation and the lack of a written standard make it hard to link inflected forms to headwords. This pilot study, part of the Nerthus Project, tackles one small, well-defined slice of the problem: adverbs inflected for the superlative. Earlier Nerthus work lemmatised verbs, so this is the first attempt on a non-verbal class.

Every token tagged as a superlative adverb was pulled automatically from the York-Toronto-Helsinki parsed corpora of Old English prose (YCOE) and poetry. A human then mapped each form to one of 1,755 adverbial lemmas supplied by the Nerthus lexical database. The lemmatised list was cross-checked against the Dictionary of Old English (DOE) and Seelig's 1930 inventory of comparative and superlative forms. This confirmed uncertain assignments and added forms that the YCOE does not attest.

The output is a lemma-by-lemma inventory that links each spelling variant to the corpus texts where it occurs. The author presents the workflow as reusable for the still-unlemmatised nouns and adjectives.

Key points

  • Sources of forms: the YCOE treebank of Old English prose texts compiled at York, Toronto and Helsinki (Taylor et al. 2003), together with its parsed poetry counterpart (Pintzuk and Plug 2001).
  • Lemma inventory: 1,755 adverbs from Nerthus, about 5% of the Old English lexicon, compared with roughly half for nouns and about 20% each for verbs and adjectives.
  • Validation sources: the DOE, which covers only letters A–I, and Seelig (1930).
  • Superlative marking: the usual ending is -ost, with variants -ast, -est and -ust. Some forms take double suffixation with -m- (e.g., inmest, ytemest). Others show only root-vowel mutation or use suppletion (wel – bet – betst).
  • Variation in the results: a single lemma can collect many spellings. Nēah, for example, maps to neahst, nehst, next, nyhst, nexð and others. Some entries tagged as superlative look comparative or positive (raðer, nytweorðlicor, wel), which shows the tagging is noisy.

Interesting ideas

  • I1. Extraction from treebank tags is automated, but lemma assignment is done by a human choosing from a fixed, closed lemma list, so every decision lands on a known headword rather than an invented one.
  • I2. Results from one source are checked against two independent references, which both confirms doubtful cases and surfaces forms the main corpus misses.
  • I3. Each attested form is stored with the IDs of the texts it appears in, so every lemma assignment can be traced back to evidence.
  • I4. Degree morphology is split into explicit mechanisms: suffix variants, double suffixation, vowel mutation and suppletion. Each mechanism needs a different lemmatisation treatment.
  • I5. The method was tested first on a small, bounded class before being scaled to larger classes.

For Mova

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

  • I1 (tokens): the en lemmatiser could restrict outputs to a curated lexicon and send unmatched forms to a review queue instead of guessing a new lemma.
  • I2 (learn): this matches the witness-agreement idea, where lemma or tag decisions count as silver only if independent lexical resources agree, and disagreements become candidates for repair rules.
  • I3 (tokens): attaching source sentence IDs to lexicon entries and irregular-form mappings would keep lemma decisions explainable and auditable.
  • I4 (tokens): English degree forms could be handled the same way, with regular -er/-est rules kept apart from suppletive tables (well/better/best, bad/worse/worst, far/further/furthest).
  • I5 (learn): new annotation or repair pipelines could be piloted on one closed category, such as degree adverbs, to tune the workflow before running it on open classes.

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

🎓Universidad de La RiojaLogroño · Spain13 works · 74 citations

📚 Shelves 5 Computer Science · 9,086 works Artificial Intelligence · 8,889 works Arts and Humanities · 9,296 works Language and Linguistics · 9,206 works Literature and Literary Theory · 340 works

🧭 Topics 3 English · 1,373 works Morphology · 277 works Historical linguistics & language change · 159 works

Cite

Cite as Yosra Hamdoun Bghiyel. 2021. A Corpus-Based Approach to the Lemmatisation of Old English Superlative Adverbs. Journal of English Studies 19.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18172/jes.4525

@article{bghiyel2021corpus,
  title = {{A Corpus-Based Approach to the Lemmatisation of Old English Superlative Adverbs}},
  author = {Yosra Hamdoun Bghiyel},
  journal = {Journal of English Studies 19},
  year = {2021},
  doi = {10.18172/jes.4525},
  url = {https://doi.org/10.18172/jes.4525}
}

References

29 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Bosworth, J. and T. N. Toller. 1973 (1898). An Anglo-Saxon Dictionary. Oxford: Oxford University Press. find
  2. Burkhanov, I. 1998. Lexicography. A Dictionary of basic terminology. Wydawn: Wyzszej Szkoly Pedagogicznej w Rzeszowie.122. find
  3. Campbell, A. 1959. Old English Grammar. Oxford: Oxford University Press. find
  4. Claridge, C. 2008. “Historical corpora”. Corpus Linguistics. An international Handbook. Eds. A. Lüdeling, M. Kytö and T. McEnery. Berlin: Mouton de Gruyter. 242-254. find
  5. Faab, G. 2017. “Lexicography and corpus linguistics”. The Routledge Handbook of Lexicography. Ed. P. A. Fuertes-Olivera. Abingdon: Routledge. Accessed 15 May 2019, Routledge Handbooks Online. 123-425. find
  6. Fulk, R. D. 2018. A Comparative Grammar of the Early Germanic Languages. Amsterdam/Philadelphia: John Benjamins Publishing Company. 237-240. ↗
  7. García Fernández, L. 2018. “Preterite-present verb lemmas from a corpus of Old English”. Verbs, Clauses and Constructions: Functional and Typological Approaches. Eds. P. Guerrero Medina, R. Torre Alonso and R. Vea Escarza. Newcastle: Cambridge Scholars Publishing. 59-76. find
  8. Journal of English Studies, vol. 19 (2021) 97-121 find
  9. Hall, J. R. C. 1984 (1960). A Concise Anglo-Saxon Dictionary. With a supplement by H. D. Meritt. Toronto: University of Toronto Press. find
  10. Healey, A. diPaolo et al. (ed.). 2018. Dictionary of Old English: A to I online. Toronto: Dictionary of Old English Project, Centre for Medieval Studies, University of Toronto. find
  11. Healey, A. diPaolo with John Price Wilkin and Xin Xiang. 2009. Dictionary of Old English Web Corpus. Toronto: Dictionary of Old English Project, Centre for Medieval Studies, University of Toronto. find
  12. Maíz Villalta, G. 2010. The Formation of Adverbs in Old English: A Functional View. Master’s Dissertation, Department of Modern Languages, University of La Rioja. find
  13. Martín Arista, J. 2010. “Building a Lexical Database of Old English: issues and landmarks”. Current Projects in Historical Lexicography. Ed. John Considine. Newcastle: Cambridge Scholars Publishing. 1-33. find
  14. Martín Arista, J., L. García Fernández, M. Lacalle Palacios, A. E. Ojanguren López and E. Ruiz Narbona. 2016. Nerthus V3. Online Lexical Database of Old English. Nerthus Project. Universidad de La Rioja [www.Nerthusproject.com] find
  15. Metola Rodríguez, D. 2015. Lemmatisation of Old English Strong Verbs on a Lexical Database. PhD Dissertation, Department of Modern Languages, University of La Rioja. find
  16. Metola Rodríguez, D. 2017. “Strong Verb Lemmas from a Corpus of Old English. Advances and issues”. Revista de Lingüística y Lenguas Aplicadas 12: 65-76. ↗
  17. Mitchell, B. and F. Robinson. 1985 (1964). A Guide to Old English. Oxford:Blackwell. find
  18. Pintzuk, S. and L. Plug (eds.). 2001. The York-Helsinki Parsed Corpus of Old English Poetry. Department of Language and Linguistic Science, University of York. find
  19. Seelig, F. 1930. Die Komparation der Adjektiva und Adverbien im Altenglischen. Heidelberg: Winter. find
  20. Smith, J. J. 2009. Old English. A Linguistic Introduction. Cambridge: Cambridge University Press. find
  21. Sweet, H. 1976 (1896). The Student’s Dictionary of Anglo-Saxon. Cambridge: Cambridge University Press. find
  22. Taylor, A., A. Warner, S. Pintzuk and F. Beths (eds.) 2003. The York-Toronto-Helsinki Parsed Corpus of Old English Prose. Department of Language and Linguistic Science, University of York. find
  23. Tío Sáenz, M. 2019. “A semi-automatic lemmatisation procedure: Old English class 1 and class 2 weak verbs”. Electronic Lexicography of Old English 1: 67-76. find
  24. Journal of English Studies, vol. 19 (2021) 97-121 find
  25. Andgietfullīce ‘sensibly, clearly, plainly, distinctly, andgitfullicast [Coboeth.o.02], andgitfullicost [Coprefcura.O2]. and intelligibly’: Beorhte ‘brightly, brilliantly, splendidly; clearly, lucidly, distinctly’: beorhtost [Coboeth.o.02, Conicoda, Coverhom], beortost [Coboeth.o.02], biorhtost [Coverhom], biorhtust [Coverhom]. Beorhtlīce ‘splendidly’: gebeorhlicost [Coverhom]. Bet ‘better, of manner better‘: best [Colaw2cn.O3], betesð [CocuraC], betest [CocuraC], betst [Coaelholm.o3, Coaelive.O3, Coapollo.O3, Coboeth.O2, Cobyrhtf. O3, Cochdrul, Cochrond, CocuraC, Codicts.O34, Codocu3.O3, Coinspolx, Colaw2cn. O3, Colaw5atr.O3, Colaw6atr.O3, Conicodc, Coorosiu.o2, Coverhom, Cowulf.O34]. Clǣne ‘clean, clearly, fully, purely’ Superlative: clænost [Cowulf.O34]. Ēaðe ‘easily, lightly, soon; willingly, readily’: eaðost [Cowulf.O34,Colaece.O2], eaðust [Cowulf.O34], eðest [Coalcuin], iðesð [CocuraC], yðest [coeuphr]. Ēaðelīce ‘easily (BT)’: eaðelicost [coherbar], eaðelicust [Cowsgosp.O3], yðelicor [Coalex.o23, Cogregdc.o24], eðelicost [Coboeth.o.02]. Eallmǣst ‘nearly all, almost, for the most part’: eallmæst [Cochronc, Cochrond]. Endemest ‘equally, likewise, at the same time, together, unanimously; fully, entirely; in procession’: ændemest [Coboeth.o.02], endemest [Coboeth.o.02]. Ende-nēxt ‘last, final’: endenexð [Coaelholm.o3]. Fægre ‘fairly, elegantly, beautifully; pleasantly, softly, gently, kindly; well, justly; early’: fægerost [Coverhom]. Fæste ‘fast, firmly, securely; straitly, strictly; heavily (sleep); speedily’: fæstost [Coaelive.O3]. Fæstlice ‘irmly, constantly, fast, quickly’: fæsðlicost [CocuraC]. Journal of English Studies, vol. 19 (2021) 97-121 find
  26. Feorr ‘far, far away, distant, remote; far back (in time); further, besides, moreover’: feorst [Cogregdc.O24], firrest [Coorosiu.o2], fyrrest [Coboeth.o.02]. Forhæfendlīce ‘continently’: forhæfendlicust [Cochdrul]. Fullīce ‘entirely, fully, perfectly, completely’: fullecost [CocuraC], fullicost [CocuraC]. Fyrmest ‘at first, most, very well, best’: fyrmest [Coaelive.O3, Coboeth.o.02, Cocathom, Cochronc, Cochrond, CocuraC, Codocu3.O3, Cogenesic, Coherbar, Coinspolx, Colaece.O2,Colaw2cn.O3, Colaw6atr.O3, Colwgeat, Coorosiu.o2, Cootest.03, Coverhom, Cowulf.O34], fyrmust [Coherbar]. Fyrst ‘at first’: fyrst [Cochad .024]. Gearwe ‘well, effectually, sufficiently, thoroughly, entirely; quickly; near’: gearost [Cogregdc.O24]. Gehende ‘near, at home; closely, in detail’: gehendost [Coaelive.O3, Cocathom1, Contempo.O3]. Gelīce ‘like as if’: geliccost [Coverhom], gelicost [Coalcuin, Coboeth.o.02, CocuraC, Codicts, Coleofri.O4, Cogregdc.O24, Colaece.O2, Coorosiu.o2, Coverhom], gelicust [Cowsgosp.O3]. Gelōmlīce ‘often, repeatedly (Sweet)’: gelomlicost [Coorosiu.o2]. Georne ‘eagerly, zealously, earnestly, gladly; well, carefully, completely, exactly; quickly’: geornast [Coinspolx], geornest [Coalcuin], geornost [CocuraC, Coinspold.O34, Coinspolx, Colaw1cn.O3, Colaw2cn.O3, Colaw5atr.O3, Colaw6atr.O3, Cowulf.O34]. Geornlīce ‘earnestly, diligently, zealously, strenuously, carefully, willingly’: geornlicast [Coinspolx], geornlicest [Cowulf.O34]. Gesundiglīce ‘prosperously (BT)’: gesundlicost [Coboeth.o.02]. Grimlīce ‘fiercely, severely, cruelly (BT)’: grimlicost [Comart3.O23]. Hāte ‘hotly, fervidly’: hatost [Colacnu.O23, Colaece.O2], hattost [Coverhom], hatust [Colacnu.o23]. Hēalīce ‘highly, aloft; in or to high position or rank, loftily; intensely, very’: healicast [Cobenrul.O3]. Hearde ‘hard, hardly, firmly, very severely, strictly, vehemently; exceedingly, greatly; painfully, grievously’: heardost [Coboeth.o.02, Coorosiu.o2]. HlŪde ‘loudly, aloud’: hluddost [Coverhom]. Hraðe ‘hastily, quickly, promptly, readily, immediately, soon; too soon’: hraðost [Coboeth.o.02, Cocathom1, Coherbar], raðer [Cowulf.O34], raðosð [CocuraC], Journal of English Studies, vol. 19 (2021) 97-121 raðost [Cochronc, Cochrond, Cogenesic, Coinspolx,Colaw2cn.O3, Colwstan1.O3, Coorosiu.o2, Cootest.O3, Cosevensl, Cowulf.O34], raðust [Cowsgosp.O3]. find
  27. Inn ‘in, into, inwards, within, inside of; inwardly’: inmest [CocuraC]. Lāðe ‘inimically, in detestation’: laðost [Colaw6atr.O3]. Lange ‘long, a long time, far’: længast [Codocu3.O3], længest [Cogregdc.O24], længst [Cochrond]. Late ‘late; slowly; at last; lately’: latost [Cobenrul.O3, Coboeth.o.02, Cogregdc. O24]. Leng ‘longer’: lengest [Cochronc, Codicts, Coorosiu.o2], lengst [Colsigewz.O34]. Līðe ‘in a gentle, soft, calm, mild; gracious, kind, agreeable, sweet manner’: licost [Coorosiu.o2] Mǣrlīce ‘gloriously, splendidly; excellently’: mærlicost [coaelive.o3, cowulf.O34]. Mǣst ‘most, chiefly, especially’: mæst [Cochronc, Cochrond], meast [Cochrond]. Nēah ‘near, nigh; about, almost, nearly, lately’: neaxst [Coboeth.o.02], neahst [Coboeth.o.02, Comart3.O23], nehst [Cochronc, Cogregdc.O24, Colaw1cn.O3], nehste [Coboeth.o.02], nest [Coelofri.04], nexð [Cochronc], next [Coboeth.o.02], nihst [Coorosiu.o2], nyhst [Comart3.O23, Coorosiu.o2]. Nearolīce ‘narrowly, closely, briefly, accurately; strictly, stringently, oppressively; evilly’: nearwlicast [Coinspolx]. Niðer ‘below, beneath, down, downwards’: nieðemesð [CocuraC], nyðemyst [Contempo.o3]. Nytwierðlīce ‘in a ‘useful, profitable, manner’: nyttwyrðlicost [CocuraC], nytweorðlicor [CocuraC]. Oft ‘above, on high; to or on the other side; from side to side, across; beyond, above (quantity)’: oftosð [Coboeth.o.02, CocuraC], oftost [Coaelive, Coboeth.o.02, Cocathom2, Cochrond, CocuraC, Coinspold.O34, Coinspolx, Colaece.O2, Colaw6atr. O3, Colwsigexa.O34, Coorosiu.o2, Coprefcura, Coquadru.O23, Coverhom, Covinsal, Cowulf.O34], oftust [Coherbar, Coverhom]. Onlice ‘in a like, resembling, similar manner’: onlicost [Coorosiu.o2]. Orenlīce ‘openly, manifestly, plainly, clearly, unreservedly’: [Coverhom]. orenlicost Rihte ‘right, due, straight (of direction, as in right on, due east), outright; precisely, exactly, just; rightly, duly, well, correctly, truly, properly, fairly, justly; directly, immediately’: rihtost [Coinspold], ryhtosð [CocuraC]. Journal of English Studies, vol. 19 (2021) 97-121 find
  28. Rihtlīce ‘justly, uprightly, virtuously; properly, rightly, regularly; correctly, precisely’: rihtlicost [Cobyrhtf, Cocathom1, Codocu3.O3, Coinspold.O34, Coinspolx, Contempo.O3]. Scortlīce ‘shortly, briefly, soon’: sceortlicost [Coaelholm.o3]. Sēl ‘better, more effectually, rather, sooner, in preference’: selest [Cobenrul.O3, Coboeth.o.02, Coherbar, Colaece.O2, Coverhom], selost [Coaelive.O3, Colaw1cn. O3, Colaw2cn.O3, Colaw6atr.O3, Colsigewz.O34, Colwsigexa.O35, Contempo.O3, Coquadru.O23], selust [coherbar], soelest [Codocu2.O12]. Seldor ‘more seldom, less frequently’: seldost [Coboeth.o.02]. Sīð ‘late, afterwards’: siðestan [Colawine.Ox2], suiðusð [CocuraC], suiðust [CocuraC] Smalum ‘little by little’: smalost [Colaece.O2]. Smēalīce ‘closely, thoroughly, accurately; subtlely’: smealicost [Coboeth.o.02]. Stīðlīce ‘forcibly’: stiðlucest [Coneot]. Sweotollīce ‘clearly, precisely, plainly, visibly, openly’: sweotelicost [Coorosiu. o2], sweotolost [Coboeth.o.02], swiotelecost [Coverhom], swiotolusð [CocuraC]. Swīðe ‘very much, exceedingly, severely, violently, fiercely’: swiðast [Coalex. o23, Comarvel.o23], swiðest [Coaelholm.o3, cochad.024], swiðosð [ CocuraC], swidost [cobenrul], swiðost [Coaelholm, Coaelive.O3, Cobenrul, Coboeth.O.02, Cocathom1, Cocathom2, Cochdrul, Cochronc, Codicts, Coepigen.O3, Coeuphr, Cogenesic, Cogregdc.O24, Coherbar, Colaece.O2, Comarvel.O24, Comary, Coorosiu.O2, Cootest, Coprefcath1.O5, Copreflives, Cosevensl, Coverhoml], swiðusð [ CocuraC], swiðust [coaelholm, cobenrul.o3, Coboeth.o.02, cocathom1, cogregdC.o24, coherbar, Coverhom], swiðast [Colacnu.o23], swiðost [Coaelive. O3,Cobenrul,Coboeth.O.02, Cocathom1, Cochronc, Cogregdc.O24, Coherbar, Colacnu.O23, Colaece.O2, Comart3.O23, Coorosiu.O2, Cowulf.O34], swiðust [Cobenrul, Coboeth.O.02, Colaece.O2], swyðost [Coaelholm, Coaelive.O3, Coalex. O23, Cocathom2, Cochronc, Cogregdh.O23, Coinspolx, Colaw2cn.O3, Comargat, Cosevensl, Cochrond, Coherbar, Coinspold.O34, Colaw6atr.O3, Cootest, Cowulf. O34], swyðust [Cobenrul], swyðast [Cowulf.o34]. Teartlīce ‘sharply, severely (BT)’: teartlicost [cobenrul]. Tēonlīce ‘in a manner that causes harm or trouble, grievously, miserably (BT)’: teonlycost [conicodA]. Tulge ‘strongly, firmly, well’: tylgest [Cochad.024]. Ðwēorlīce ‘insolently’: ðwyrlicost [cogregdH.o23]. Journal of English Studies, vol. 19 (2021) 97-121 find
  29. Ufor ‘higher, further away, further up; later, posterior, subsequent’: ufemest [coaelive.o3], yfemest [Coboeth.o.02]. Undēore ‘cheap’: undeorest [cobenrul]. Ungerǣdlīce ‘sharply, roughly, violently’: ungeredelicost [Coboeth.o.02]. Ūt ‘out; without, outside’: ytemest [cobenrul, cowsgosp.o3]. Wǣrlīce ‘truly’: wærlicast [cowulf.o34]. Wel ‘well, abundantly; very, very easily, very much; fully, quite; nearly; indeed, to be sure’: wel [cowulf.o34]. Weorðlīce ‘with distinction, gloriously; befittingly (BT)’: wurðlicost [coapollo.o3]. Wyrs ‘worse’: wierst [CocuraC], wyrest [conicodC]. Journal of English Studies, vol. 19 (2021) 97-121 find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/bghiyel-2021-corpus-based-approach-lemmatisation-old/