Meaning in poetry: semantic annotation of verse with the Historical Thesaurus of English

Koristashevskaya, Elina (2021) Meaning in poetry: semantic annotation of verse with the Historical Thesaurus of English. PhD thesis, University of Glasgow.

Full text available as:
Download (3MB) | Preview
[img] Multimedia file (Digital Appendix)
Download (44MB)


This thesis addresses the current gap in semantic annotation of poetry by presenting the first semantic tagging system specifically designed to disambiguate senses in a diachronic corpus of poetry. The ‘ambiguity tagger’ developed for this purpose utilises the hierarchical taxonomy of the Historical Thesaurus of English (HTE; Kay 2011: 42) to assign conceptual ‘tags’ to lexical items that denote the meaning of the word in context, with multiple meanings assigned to ambiguous words. The ambiguity tagger encompasses a configurable pipeline for semantic annotation, thus presenting a more flexible alternative to existing applications (Piao et al. 2005a; Rayson 2009a; Piao et al. 2017). To train the tagger, a corpus was curated from the Oxford Book of English Verse, containing poetry from the early 16th to the late 19th century (OBEV; Quiller-Couch 1919/1999).

As the ambiguity tagger allows multiple meanings to be assigned to individual words in the corpus, without restricting the maximum number of senses, the semantic metadata produced by the tagger is unique in its breadth. Correspondingly, the analysis sections of the thesis look at different techniques for interpreting the data, using case studies from the OBEV corpus. Both macro- and micro-level approaches to analysing the data are explored, highlighting the benefits of the ambiguity tagger at different levels of critical analysis. To further explore the capabilities of semantic annotation with HTE data, this research extends the interpretative analysis of the semantic metadata gained through the ambiguity tagger by presenting a systematic approach for analysing the significant co-occurrence of concepts in the text. This process borrows the framework for identifying significantly co-occurring words (collocates) and extends this into a measure of ‘semantic collocation’, thus significantly expanding on existing research in this field (Alexander et al. 2015a; Archer & Malory 2015; 2017). By shifting the focus from lexical collocation to the significant co-occurrence of ‘meaning’ in texts, this approach reveals a pattern of previously inaccessible textual data for analysis and marks a further methodological contribution of this research.

Item Type: Thesis (PhD)
Qualification Level: Doctoral
Keywords: semantic annotation, corpus linguistics, corpus stylistics, historical thesaurus of english, semantic collocation, semantic tagging.
Subjects: P Language and Literature > P Philology. Linguistics
Colleges/Schools: College of Arts > School of Critical Studies > English Language and Linguistics
Supervisor's Name: Alexander, Professor Marc and Anderson, Professor Wendy and Dallachy, Dr. Fraser
Date of Award: 2021
Depositing User: Dr Elina Koristashevskaya
Unique ID: glathesis:2021-82230
Copyright: Copyright of this thesis is held by the author.
Date Deposited: 02 Jun 2021 07:28
Last Modified: 02 Jun 2021 07:41
Thesis DOI: 10.5525/gla.thesis.82230

Actions (login required)

View Item View Item


Downloads per month over past year