Indeterminacy and the
interpretation patterns for modal
FACULTY OF EDUCATION AND SOCIETY DEPARTMENT OF CULTURE, LANGUAGES AND MEDIA
Interpre'ng
Modal
U0erances
in English
and Swedish
Modal interpreta-on Modal IndeterminacyUsage-‐based approach to language study
The data
The method The results Summary
You must go to the doctor.
You must have a fever.
Modal
De kan väl åka ut -ll
They MOD MOD drive-‐INF out to
Marsvinsholm när de får
Marsvinsholm when they get-‐PRES
'd?
(HM)'me
‘They can drive out to Marsvinsholm when they have 'me, can’t they?’
Modal
“The recurring problem for linguis-c analyses of the
modals has been the lack of a principled account of how we arrive at an explicit
interpreta-on of a sentence containing a modal”(Klinge 1993: 318).
“The model of language proposed by Cogni've Linguists is so
completely simple that it places the emphasis squarely on method
and data. Rather than simplifying the object of study by carving off its complexi'es with hypothe'cal modules of language structure, it lands the linguist in the midst of a chao'c phenomenon that is the nature of all socially structured
systems” (Glynn 2010: 2).
Usage-‐Based
Approach to
Language
Meaning as “a process of
sense crea'on” (Geeraerts
1993: 260)
Empirical data
Mul'factorial feature
analysis
Usage-‐Based
Approach to
Language
Study
Ca 3,000 examples of
uIerances containing must, may, måste and kan
The English-‐Swedish Parallel Corpus
Originals only
Both fic-on and non-‐fic-on Co-‐text of 5 sentences before and aPer
(a) What cogni've-‐func'onal
features are present in the
collected data?
(b) Is there a systema'c
rela'onship between the
presence of these features
and the modal
interpreta'on?
21 nominal a0ributes for a data
mining analysis:
Interpreta5on: epistemic/deon'c/dynamic/indeterminate
Adverbial par5cle: yes/no
Subject: animate/specific/generic/introductory
Control: yes/no
Person: 1st/2nd/3rd
Verb: event/state
Aspect: simple/perfec've/progressive
Time reference for modality: past/present/future
Time reference for proposi5on: ant/sim/post
Situa5on: telic/atelic
Nega5on: of modality/of proposi'on
Voice: ac've/passive
Condi5on: explicit/implicit
Type of uBerance: asser've/non-‐asser've/exclama'on
1,461 examples
1,388 examples, or 95% of the data, are correctly predicted
73 examples, or 5% of the
data, are incorrectly predicted
The Result:
Must and
Måste
Yrk 90hp
The Result:
Must and
Måste
=== Confusion Matrix ===
a b c <-‐-‐ classified as 287 32 0 | a = epistemic 32 1,102 0 | b = deon'c 3 5 0 | c = indet.
361 examples
280 examples, or 77.6% of the data, are correctly
predicted
81 examples, or 22.4% of the data incorrectly
predicted
The Results:
Yrk 90hp
=== Confusion Matrix ===
a b c <-‐-‐ classified as 289 6 5 | a = epistemic 9 15 1 | b = deon'c
32 3 0 | c = indeterminate
The Results:
1,001 examples
662 examples, or 66.1% of the data, are correctly
predicted
339 examples, or 33.9% of the data, are incorrectly predicted
The Results:
Yrk 90hp
=== Confusion Matrix ===
a b c d e <-‐-‐ classified as 27 61 0 13 0 | a = epistemic 28 266 2 77 0 | b = root poss. 0 12 23 27 0 | c = deon'c 1 89 13 344 1 | d = dynamic 1 4 1 16 2 | e = indet.
The Results:
Kan
Seman'c structures for the modals must, måste and may successfully revealed through mul'factorial analysis
No discernible pa0erning in indeterminate u0erances
Yrk 90hp
References
Geeraerts, D. 1993.Vagueness’s puzzles, polysemy’s vagaries. Cognin4ve Linguis4cs, 4, 223-‐272
Glynn, D. 2010. Corpus-‐Driven Cogni've Seman'cs. An Introduc'on to the field. In Glynn, D and K. Fischer (eds.), Corpus-‐Driven Cogni4ve
Seman4cs. Quan4ta4ve Approaches, 1-‐42. Berlin: Mouton de Gruyter Klinge, A. 1993. The English modal auxiliaries: from lexical seman'cs to
u0erance interpreta'on. Journal of Linguis4cs, 29, 315-‐357
Wärnsby, A. 2006. (De)coding Modality: The case of Must, May, Måste and Kan. In Thormählen, M. and B. Warren (eds.), Lund Studies in English 113. Lund: Lund University