- Title
- Rho-grams and rho-sets: significant links in the web of words
- Creator
- Burrows, John
- Relation
- Digital Scholarship in the Humanities Vol. 33, Issue 4, p. 724-747
- Publisher Link
- http://dx.doi.org/10.1093/llc/fqy004
- Publisher
- Oxford University Press
- Resource Type
- journal article
- Date
- 2018
- Description
- Much of the quantitative work undertaken in stylistic analysis has to do with word frequencies-usually the relative frequencies of an appropriate set of single word-types. Those who have sought to go further by choosing word-types that tend to 'go together' have taken sequence and close proximity as their criteria. But many words display similar patterns of frequency without necessarily meeting those criteria: sets of grammatical associates, deictic features, archaisms, colloquialisms, Latinisms, and many others. Such sets, moreover, have negative corollaries, the alternatives consistently not chosen. Across a range of texts appropriate to whatever case may be in hand, both positive resemblances and direct contrasts of frequency can be identified by Spearman's method of correlation. The coefficients for many of the pairs united in this way show very high levels of statistical significance. These pairs can be gathered in sets embracing all the partners of a given member, with separate subsets for positives and negatives. When, for example, 'the' is taken as a 'headword', it yields positive and negative sets, 'THE_p' and 'THE_n'. Such 'rho-sets' can then be treated as composite variables and employed as data in much the same ways as we customarily use single-word variables. The trials undertaken (and illustrated here) suggest that this approach gives unusually accurate measures of stylistic difference, especially with short texts. Many of the sets themselves are of considerable philological interest and help to explain how the study of word frequencies can be so rich in stylistic information.
- Subject
- stylistic analysis; word frequencies; word types; grammatical associates; deictic features; archaisms; colloquialisms; Latinisms
- Identifier
- http://hdl.handle.net/1959.13/1403939
- Identifier
- uon:35240
- Identifier
- ISSN:2055-7671
- Language
- eng
- Reviewed
- Hits: 1157
- Visitors: 1157
- Downloads: 0
| Thumbnail | File | Description | Size | Format |
|---|