Difference between revisions of "Metaphor Identification and Corpus Research"
(Add sourced introduction to metaphor identification, corpus retrieval and annotation) |
(Separate the fields in the worked research record into table rows) |
||
| Line 18: | Line 18: | ||
|- | |- | ||
| Passage and context || A committee discussion stopped making progress; no collision was described. | | Passage and context || A committee discussion stopped making progress; no collision was described. | ||
| + | |- | ||
| Expression examined || “hit a wall” | | Expression examined || “hit a wall” | ||
| + | |- | ||
| Contextual interpretation || Encountered an obstacle to continuing the discussion. | | Contextual interpretation || Encountered an obstacle to continuing the discussion. | ||
| + | |- | ||
| Comparison proposed || Physical collision and obstruction used to describe a difficulty in discussion. | | Comparison proposed || Physical collision and obstruction used to describe a difficulty in discussion. | ||
| + | |- | ||
| Uncertainty to retain || A longer passage might extend, reject or joke about the image. | | Uncertainty to retain || A longer passage might extend, reject or joke about the image. | ||
| + | |- | ||
| Source record || In a real study: document, speaker if known, date, location in the text and surrounding passage. | | Source record || In a real study: document, speaker if known, date, location in the text and surrounding passage. | ||
|} | |} | ||
Latest revision as of 15:23, 5 October 2026
Metaphor identification is the task of deciding which expressions are being used metaphorically in a particular context. Corpus research studies collections of language produced in speech or writing. Bringing the two together allows researchers to examine patterns across texts, speakers and genres, while retaining the context needed to interpret individual uses.[1]
Finding words and identifying uses
A search for a word retrieves its occurrences, not a ready-made set of metaphors. The same word can have different uses. Stefanowitsch describes three useful routes into metaphor research: searching vocabulary associated with a source domain, searching words associated with a target topic, and searching expressions that explicitly signal comparison or figurative language. Each route makes some examples easier to find and others easier to miss.[1]
For example, a search for stuck might retrieve a vehicle in mud, a person unable to make a decision, and a quotation of somebody else's complaint. Searching only for stuck would miss descriptions involving a barrier, a loop or a blocked exit. These are constructed examples illustrating a retrieval problem, not findings about how frequently people use the expressions.
Context and annotation
Annotation means attaching a recorded judgement to a passage. A useful scheme specifies the unit being classified, the possible labels, and the rules for difficult cases. Stefanowitsch explains that amendments to a scheme may require earlier material to be checked again. He also recommends assessing whether different annotators can apply the same rules reliably.[2]
Zawisławska's Synamet project illustrates how much more can be recorded than a yes-or-no metaphor label. Annotators read complete Polish blog entries and recorded topics, metaphorical units, grammatical information, source and target frames, and atypical or borderline cases. Its procedure was designed for sensory metaphor and should be understood in relation to that purpose.[3]
A small worked record
The following is an illustrative record for the invented sentence “The discussion hit a wall.” It is not a validated identification procedure.
| Field | Example entry |
|---|---|
| Passage and context | A committee discussion stopped making progress; no collision was described. |
| Expression examined | “hit a wall” |
| Contextual interpretation | Encountered an obstacle to continuing the discussion. |
| Comparison proposed | Physical collision and obstruction used to describe a difficulty in discussion. |
| Uncertainty to retain | A longer passage might extend, reject or joke about the image. |
| Source record | In a real study: document, speaker if known, date, location in the text and surrounding passage. |
Reading a corpus claim
Before treating a pattern as typical, check what was collected and how examples were found. A collection of perfume reviews answers a different question from a collection of everyday conversations. A count of matching words also differs from a count of uses judged metaphorical after reading their contexts.
Useful questions include: which texts were eligible, which were sampled, what counted as one example, how disagreements were handled, and what the frequency was measured against? These questions connect retrieval and annotation to the interpretation of the result. The corpus and method should support the scope of the claim.[2][1]
Related pages
- Linguistic References
- Metaphor Categories
- Sensory Metaphor
- Metaphor in Economics
- Metaphor Across Languages and Cultures
- Metaphor Research Library
Sources
- ↑ 1.0 1.1 1.2 Anatol Stefanowitsch (2020), Corpus Linguistics: A Guide to the Methodology, chapter 11, especially pp. 397–398. Language Science Press. Free complete book.
- ↑ 2.0 2.1 Stefanowitsch (2020), Corpus Linguistics, chapter 4, pp. 122–128.
- ↑ Magdalena Zawisławska (2019), Metaphor and Senses: The Synamet Corpus. A Polish Resource for Synesthetic Metaphors, chapter 2, pp. 33–54. Peter Lang. Free complete book.