Abstract / Description of output
This paper explores data-driven, stylometric methodologies in order to
see whether their findings would reflect stylistic developments from Early Modern English to the present day, on the basis of a small corpus of 13 texts representing the same register (horse manuals) and the same topic (feeding a horse) from 1565 to 2009. The corpus was POS-tagged, and then stripped of its lexical material, leaving only the tags. Sequences of tags were then input to a correspondence analysis and subjected to a procedure to create an association plot. Interpreting the results is a particular challenge of data-driven methods, which requires a mixture of quantitative and qualitative analysis, and differently-sized samples. The assumption that the results would not be driven by syntactic change was confirmed,so that the chronological progression that emerged in the data had to be explained by changing styles. Horse manuals contain at least two text types, to varying degrees: instructional/procedural “cookery book” writing and the logicalargumentative text type of scientific writing. The latter has been known to exhibit higher frequencies of passivization and a more nominal style (Halliday 2004); this is also evident in these texts. The data further suggest that the flow of given to new information is not adhered to as strictly in the earlier texts than in the later texts; in the earlier texts, pronouns are found more frequently in end-focus position, and the strategy of using the by-phrase of the “long passive” to manoeuvre a new agent into end-focus position is only found in the later texts.
see whether their findings would reflect stylistic developments from Early Modern English to the present day, on the basis of a small corpus of 13 texts representing the same register (horse manuals) and the same topic (feeding a horse) from 1565 to 2009. The corpus was POS-tagged, and then stripped of its lexical material, leaving only the tags. Sequences of tags were then input to a correspondence analysis and subjected to a procedure to create an association plot. Interpreting the results is a particular challenge of data-driven methods, which requires a mixture of quantitative and qualitative analysis, and differently-sized samples. The assumption that the results would not be driven by syntactic change was confirmed,so that the chronological progression that emerged in the data had to be explained by changing styles. Horse manuals contain at least two text types, to varying degrees: instructional/procedural “cookery book” writing and the logicalargumentative text type of scientific writing. The latter has been known to exhibit higher frequencies of passivization and a more nominal style (Halliday 2004); this is also evident in these texts. The data further suggest that the flow of given to new information is not adhered to as strictly in the earlier texts than in the later texts; in the earlier texts, pronouns are found more frequently in end-focus position, and the strategy of using the by-phrase of the “long passive” to manoeuvre a new agent into end-focus position is only found in the later texts.
Original language | English |
---|---|
Title of host publication | Context, Intent and Variation in Grammaticalization |
Editors | Hendrik De Smet, Peter Petré, Benedikt Szmrecsanyi |
Publisher | De Gruyter |
Chapter | 11 |
Pages | 275-301 |
Number of pages | 26 |
ISBN (Electronic) | 9783110753059 |
ISBN (Print) | 9783110752953 |
DOIs | |
Publication status | Published - 19 Jun 2023 |
Publication series
Name | Trends in Linguistics Studies and Monographs [TiLSM] |
---|---|
Publisher | De Gruyter |
Volume | 365 |
ISSN (Print) | 1861-4302 |
ISSN (Electronic) | 2199-3734 |
Keywords / Materials (for Non-textual outputs)
- stylistic variation
- information structure
- Early Modern English
- Late Modern English
- modification
- horse manuals