Upload
others
View
4
Download
0
Embed Size (px)
Citation preview
Fillmore Fest , Berkeley, July 31 2009
TRANSITIVITY, SUBJECTIVITY AND DISCOURSE:
A Corpus Analysis Of Spanish Argument Structure
José M. García-Miguel Victoria Vázquez RozasUniversidade de Vigo Univ. de Santiago de Compostela
[email protected] [email protected]
Fillmore Fest , Berkeley, July 31 2009
The aim of the paper
• Are Argument Structure Constructions (ASC) related to text types (genres)?
• What is the role of semantic and grammatical parameters, such as verb class and grammatical person in the link between ASC and discourse?
Discourse and constructions
• Usage moulds grammatical patterns:– research should be based on contextualized
utterances– frequency of use is expected to be relevant for
ASC • Frequency of specific constructions has proven to
be related to differences of genre (Biber)
Syntactic and semantic annotation of corpora is needed.
Corpus: ARTHUS
Genre Words % Clauses %
Essay 257,718 17.78% 20,013 12.59%
Narrative 538,906 37.19% 72,425 45.56%
Spoken 273,070 18.85% 25,143 15.82%
Press 166,804 11.51% 13,247 8.33%
Theater 212,507 14.66% 28,126 17.69%
Total 1,449,005 158,954
ARTHUS = ARchivo de Textos Hispánicos de la Universidad de SantiagoTexts in Spanish published between 1980 and 1991
ADESSE: Alternancias de Diátesis y Esquemas Sintactico-Semánticos del Español
http://adesse.uvigo.es/• A database with the (manual) syntactic and
semantic annotation of the clauses of ARTHUS:– Syntactic relations – Syntactic categories – Semantic-roles– Semantic classification of verbs– etc…
• ADESSE is an expanded version of BDS (Base de datos sintácticos del español actual, http://www.bds.usc.es/ )
(part of) a record in ADESSE
PRED ARGUMENTSTRANSFER Receiver Donor PossessumREGALAR 'give (as a present)'
Arg1 Arg0 Arg2
Active IObj Subj DObj3sg.dative [le] 3pl [-ron]a NP [el novio]
NP [un automóvil]
[Past] Animate Animate Concrete thing
Al novio le regalaron un automóvil convertible [CRO: 44, 1] To-the bridegroom him gave-they a car convertible 'They gave the bridegroom a convertible car as gift'
Syntactic-semantic parameters and sample for this study
• Syntactic Transitivity
• Process type (Verb semantic class)
• Grammatical person of the first argument
Total number of clauses considered: 128,149
NB: Nonfinite clauses and clauses with no subject have been excluded
Syntactic Patterns in BDS/ADESSE
Subj V DObj (SD ) 41,747 32.6%Subj V Obl (S O ) 29,951 23.4%Subj V (S ) 27,791 21.7%Subj V DObj Obl (SD O ) 8,168 6.4%Subj V Compl (SC ) 6,683 5.2%Subj V DObj IObj (SDI ) 5,733 4.5%Subj V IObj (S I ) 5,485 4.3%Subj V IObj Obl (S IO ) 1,267 1.0%Subj V Compl IObj (SCI ) 998 0.8%Subj V Compl IObj Obl (SDIO ) 229 0.2%Subj V Compl Obl (SC O ) 93 0.1%
128,145
Some examples1. Subj V DObj
El comisario abrió la bocaThe captain opened his mouth
2. Subj V ComplPlácida Linero pensó que había pasado el peligroP.L. thought that the danger was over
3. Subj V Obl Estamos en casa de papáWe are at dad’s homeTú vas a seguir calladitoYou are going to keep quiet
4. Subj V IObjNo me importan las calificacionesI don’t care about grades
Transitive and Intransitive clauses
TransitivityIntransitive(S )
72,272(56.4%)
Transitive(SD )
55,877(43.6%)
TOTAL 128,149
<Subj V Compl *> clauses have been counted as intransitive constructions (cf. Halliday 1985; Biber 1999; Verhagen 2001; Thompson 2002, etc)
Transitivity
Transitive Intransitive
Conversational English(Thompson & Hopper 2001)
27% 73%
Spoken & written English (Nijmegen corpus: Oostdijk & de Haan 1994)
30,2% 69,8%
Spoken Spanish (Venezuela) (Bentivoglio 1992) 34% 66%
ARTHUS 43,3% 56,7%
Transitivity
• The different percentage of intransitive clauses in ARTHUS and in the other corpora might be due to the different makeup of the samples
• ARTHUS is mostly composed of literary texts
• The difference may be determined by text type
Are there significant differences in transitivity depending on the type of text?
Transitivity and textual genre
Total IntransitiveEssay 14,266 7,922 56%Narrative 58,393 32,556 56%Spoken 21,663 12,989 60%Press 10,159 5,615 55%Theater 23,668 13,190 56%
128,149 72,272 56%
Chi-square=135.5 (p < .001)
Intransitive clauses and genre in BDS/ADESSE
Transitivity and conversation• Thompson and Hopper (2001:53):
"The low Transitivity in our conversational data is to a considerable extent determined by the kinds of things we are doing when we talk with friends and acquaintances. We do not seem to talk much about events, let alone actions […] but rather, our talk is mostly about 'how things are from our perspective'"
"Our data show that we describe states, reveal our attitudes, ascribe properties to people and situations, and give our assessments of situations and behaviour."
Transitivity and conversation
If Thompson and Hopper are right, we should expect a higher frequency of certain verb classes (verbs that ascribe properties or give assessments) in conversation than in any other genre.
Transitivity and spoken discourse
• Are relational verbs and assessment verbs more frequent in spoken discourse than in written discourse ?
• Are relational verbs and assessment verbs less transitive than other verb classes?
To answer these questions, we need to view the semantic classification of the verbs in our corpus as related to genre.
ADESSE verb classes
TOP-LEVEL CLASSES Exs CLAUSES1 MENTAL 11 Feeling Gustar ‘like’
29.965 (24,1%)
12 Perception Ver ‘see’13 Cognition Saber ‘know’
2 RELATIONAL 21 Attributive Ser ‘be’ 25.084(20,2%)22 Possession Tener ‘have’
3 ‘MATERIAL’ 31 "Space" Ir ‘go’45.783
(36,9%)32 Change Abrir ‘open’33 Other facts Tocar 'touch'34 Behaviour Reír ‘laugh’
4 COMMUNICATION Decir ‘say’ 12.725(10,3%)
5 EXISTENTIAL Haber ‘exist’’ 6.827(5,5%)
6 MODULATION Hacer ‘make 3.723(3%)
TOTAL
a hierarchical conceptual/ontological classification (70 classes) based in lexical relations of synonymy and hyponymy, not aspectual or primarily syntactic
Text type and verb class (ADESSE)
Verb class Essay (14,266)
Narrative (58,393)
Spoken (21,663)
Press (10,159)
Theater (23,668)
Total (128,149)
MENTAL 20.3% 24.1% 29.3% 15.5% 24.1% 23.9%
RELATIONAL 27.8% 17.8% 26.9% 21.8% 18.4% 20.9%
MATERIAL 30.6% 37.6% 26.5% 35.4% 38.9% 35.0%
VERBAL 8.9% 11.1% 8.6% 15.1% 9.0% 10.4%
EXISTENTIAL 7.8% 6.4% 7.1% 8.6% 6.8% 6.9%
MODULATION 4.7% 3.0% 1.6% 3.6% 2.8% 2.9%
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
Essay Narrative Oral Press Theatre
1-MENTAL 2-RELATIONAL 3-MATERIAL 4-VERBAL
Main verb classes and text typesMore MATERIAL processes
More VERBAL processes
More RELATIONAL processes
More MENTAL processes
Transitivity and process type [verb class]
Class Trans Intr Total%
Intrans1 MENTAL 14.514 16.058 30.572 53%2 RELATIONAL 12.673 14.050 26.723 53%3 MATERIAL 19.568 25.330 44.898 56%4 VERBAL 4.621 8.671 13.292 65%5 EXISTENTIAL 2.242 6.643 8.885 75%6 CAUSATIVE 2.259 1.520 3.779 40%
total 55.877 72.272 128.149 56%
The hypothesis that relational and mental processes, per se, imply less transitivity must be rejected.
Transitivity, verb class and text type
% of INTRANSITIVEClass ARTHUS SPOKEN
1 MENTAL 53% 67%2 RELATIONAL 53% 48%3 MATERIAL 56% 61%4 VERBAL 65% 58%5 EXISTENTIAL 75% 81%6 CAUSATIVE 40% 39%
total 56% 60%
There is a significant increase of intransitive constructions with mental verbs in the spoken subcorpus.
A usage-based approach
• Large-scale frequency distributions arise from particular discourse events in which speakers choose what they say following the goals of discourse and cultural conventions
• Now, we will look at the most frequent tokens / exemplars to see how they fit in the discourse goals of different genres
More frequent verb forms in the corpus
es '(s)he/it
is' 2310tiene '(s)he/it
has' 1114
creo 'I think' 1043sé 'I know' 954era '(s)he/it
was' 884
tengo 'I have' 699dijo '(s)he
said' 606
tenía '(s)he/it
had' 579está '(s)he/it
is' 474
me gusta 'I like' 450
More frequent verb forms in the SPOKEN subcorpus
N Verb
typees (s)he/it
is 738 Relational
creo I think 734 Mentaltiene (s)he/it
has 496 Relational
tengo (s)he/it
has 373 Relationalme gusta 'I like' 353 Mental
sé 'I know' 347 Mentalme parece 'It
seems
to
me' 202 Mental
tienen they
have 184 Relationalpasa It
happens 145 Existencial
va (s)he
goes 126 Material
More frequent MENTAL verb forms in the spoken subcorpus
N%
of
verb
usescreo 'I think' 734 81%
me gusta 'I like' 353 47%sé 'I know' 347 56%
me parece 'It
seems
to
me' 202 66%veo 'I see' 95 13%
me acuerdo 'I remember' 90 67%opinas 'you
think
about' 84 72%
conozco 'I know' 76 27%crees 'you
think' 76 8%
sabes 'you
know' 61 10%recuerdo 'I remember' 60 59%sabe 'he knows' 59 10%pienso 'I think' 58 25%
9. yo creo que si llegaran a decirlo, los americanos es que creo que hacen el más espantoso de los ridículos si no dicen nada [MADRID: 048, 10] I think that if they come to say it, Americans … I think they make a fool of themselves if they say nothing
10. Enc.- ¿Pero lo... lo supieron así interpretar o...? Inf. A.- Mirá, yo no sé. [BAIRES: 069, 35]Interviewer: But they knew how to interpret it like that or…? Interviewed: Look, I don’t know.
11. yo me gusta que los chiquillos sepan por lo menos nociones de música, [MADRID: 210, 20]Me, I like the kids to know at least some notions of music
12. me gusta el rojo y me gusta el naranja y los amarillos. Son colores alegres [MADRID: 070, 13]I like red and I like orange and yellow. They are bright colors
• Expressions such as (no) creo, (no) sé or (no) me acuerdo do not describe the mental state of the speaker, but rather express his or her epistemic evaluation of the relevant fragment of the discourse in which they occur
• Their grammaticalization as discourse markers has been related to the progressive subjectivization of their meaning (Traugott 1989), which would also explain the weakening of their valency pattern (less transitive) as a consequence of the loss of their capacity to encode events
More frequent verb forms in the PRESS subcorpus
es '(s)he
is' 128 Relacionaltiene '(s)he
has' 87 Relacional
explicó '(s)he
explained 55 Verbaldijo '(s)he
said' 52 Verbal
afirmó '(s)he
stated 48 Verbalseñaló '(s)he
pointed
out' 44 Verbal
se encuentra '(s)he
is' 44 Relationalconsidera '(s)he
considers' 44 Mental [verbal]
aseguró '(s)he
assured' 43 Verbalestá '(s)he
is' 43 Relacional
13. Leguina explicó que "la venta de droga está penada en España y en todo el mundo, y es un delito y los delitos los cometen los delincuentes" ([1VOZ: 19, 2, 1, 008])Leguina explained that “drug sale is punished in Spain and all around the world, and it is a crime and crimes are committed by criminals”
14. La organización Yihad Islámica afirmó ayer en Beirut que uno de sus comandos penetró en el norte de Israel (1VOZ: 07, 3, 1, 024)The Islamic Jihad organization stated yesterday in Beirut that one of its cells entered the north of Israel
15. El ministerio de Sanidad considera que la prevención es el instrumento más efectivo para la lucha contra el SIDA [3VOZ: 69, 2, 1, 009]The Ministry of Health considers that prevention is the most effective instrument to fight AIDS
Speech Act Participants as subject and textual genre
SAP as Subject and Verb Class
A general conclusion
• The specific communicative function of each textual genre becomes apparent in the frequency and distribution of highly subjective constructions.
Summary of results
• There is an association between genre and verb class; e.g., spoken discourse - mental processes
• Spoken discourse is more subjective than other genres:– strong association with first person Subject/IndObjExp
– high frequency of first person Experiencers with mental verbs
• Press avoids expressing subjectivity explicitly:– fewer mental verbs and fewer 1st person subjects– high frequency of communication verbs in 3rd person, to
frame the validity of the facts and reported opinions
ReferencesBentivoglio, Paola (1992). “Linguistic correlations between subjects of one-argument
verbs and subjects of more-than-one-argument verbs in spoken Spanish”. In P. Hirschbühler and K. Koerner, eds., Romance Languages and Modern Linguistic Theory, Amsterdam: John Benjamins, pp.11-25.
Benveniste, É. (1958). “De la subjectivité dans le langage”, Journal de Psychologie. Reprinted in Problèmes de linguistique générale I. París: Gallimard, 1966, pp. 258- 266.
Biber, D. (1988). Variation Across Speech and Writing, Cambridge, CUP.Brinton, Laurel J. (2009). “Pathways in the development of pragmatic markers in English”.
In A. Kemenade and B. Los, eds., The Handbook of The History of English, London: Blackwell, 307-334.
Bybee, J. (2006) From usage to grammar: the mind’s response to repetition, Language 82/4, pp. 711-733.
Fillmore, Ch., Ch. Johnson & M.L. Petruck. 2003. Background to FrameNet. International Journal of Lexicography 16/3: 235–250.
Halliday, M.A.K. (1985). An introduction to functional grammar London: Arnold, (3rd edition, revised by Christian M. I. M. Matthiessen 2004 ).
Halliday, M.A.K. (1993). “Quantitative studies and probabilities in grammar”, en M. Hoey, ed., Data, description, discourse. Londres: HarperCollins Publ., pp. 1-25
Bentivoglio, Paola (1992). “Linguistic correlations between subjects of one-argument verbs and subjects of more-than-one-argument verbs in spoken Spanish”. In P. Hirschbühler and K. Koerner, eds., Romance Languages and Modern Linguistic Theory, Amsterdam: John Benjamins, pp.11-25.
Benveniste, É. (1958). “De la subjectivité dans le langage”, Journal de Psychologie. Reprinted in Problèmes de linguistique générale I. París: Gallimard, 1966, pp. 258- 266.
Biber, D. (1988). Variation Across Speech and Writing, Cambridge, CUP.Brinton, Laurel J. (2009). “Pathways in the development of pragmatic markers in English”.
In A. Kemenade and B. Los, eds., The Handbook of The History of English, London: Blackwell, 307-334.
Bybee, J. (2006) From usage to grammar: the mind’s response to repetition, Language 82/4, pp. 711-733.
Fillmore, Ch., Ch. Johnson & M.L. Petruck. 2003. Background to FrameNet. International Journal of Lexicography 16/3: 235–250.
Halliday, M.A.K. (1985). An introduction to functional grammar London: Arnold, (3rd edition, revised by Christian M. I. M. Matthiessen 2004 ).
Halliday, M.A.K. (1993). “Quantitative studies and probabilities in grammar”, en M. Hoey, ed., Data, description, discourse. Londres: HarperCollins Publ., pp. 1-25
References
Oostdijk, Nelleke and Pieter de Haan (1994). “Clause patterns in Modern British English: A corpus-based (quantitative) study”, Icame Journal, 18: 41-79.
Scheibman, J. (2002). Point of view and grammar: Structural patterns of subjectivity in American English conversation. Amsterdam: John Benjamins.
Thompson, Sandra A. (2002). ‘Object complements’ and conversation: towards a realistic account, Studies in Language, 26: 125-164.
Thompson, S. A. & P. Hopper (2001). Transitivity, clause structure, and argument structure: evidence from conversation, in J. Bybee & P. Hopper, eds, Frequency and the emergence of linguistic structure. Amsterdam: John Benjamins, pp. 27-60.
Traugott, Elizabeth C. (1989): “On the rise of epistemic meanings in English: An example of subjectification in language change”, Language, 65: 31-35.
Verhagen, A. (2001). Subordination and discourse segmentation revisited, or why matrix clauses may be more dependent than complements. In T. Sanders, J. Schilperoord & W. Spooren, eds., Text representation. Linguistic and psychological aspects, Amsterdam: John Benjamins.
Thank you for your attention