The variable agreement of presentational haber in Dominican Spanish Jeroen Claes (Universiteit Antwerpen) [email protected]
Nov 19, 2014
The variable agreement of presentational haber in Dominican Spanish
Jeroen Claes(Universiteit Antwerpen)[email protected]
2
Outline
• Introduction• Research questions• Theoretic framework• Hypotheses• Methods• Results• Conclusions
3
1. Introduction
• Impersonal, subjectless verb in normative Spanish.- Había niños en el parque?
‘Was-there children in the park?’• NP argument is a direct object:
- Sí, losOBJ había.‘Yes, thereOBJ, PLUR was.’
• Default 3rd person singular verb-agreement.• In many varieties, optional number-agreement is
observed.- Había/habían niños en el parque.
‘There were children in the park.’
4
1. Introduction
• Variation has been around for at least 200 years (Fontanella de Weinberg, 1992).
• Change in progress, favored by: (e.g. Díaz-Campos 2003)- Human-reference NPs.- Certain verb-tenses.- Lower socioeconomic status.- Male gender.
5
2. Research questions
• What is the linguistic distribution of the pluralization of presentational haber in the Spanish of Santo Domingo, the Dominican Republic?
• What is the social distribution of the pluralization of presentational haber in Santo Domingo, the Dominican Republic?
• How can these distributions be explained in a psychologically and sociolinguistically adequate manner?
6
3. Cognitive Construction Grammar
• Usage-based.
• Every meaningful aspect of language can be modeled
with constructions (form-meaning pairs).
• Broad generalizations (e.g. transitivity) and
idiosyncratic patterns (e.g. words, idioms) are
captured with the same ease.
7
3. Cognitive Construction Grammar
• Constructions determine argument-structure: - Which and how many argument roles.- How these are mapped onto syntactic
functions.- How information is distributed over the
arguments. • Verbs can combine with multiple argument-
structure constructions.
8
4.1 Main hypothesis
• In Santo Domingo, the Dominican Republic, the pluralization of presentational haber corresponds to a slowly advancing ongoing language change from below that consists in the replacement of PRES-1 by PRES-2, which only differ in regards to the syntactic function of the NP (PRES-1: object; PRES-2: subject) and the social and stylistic connotations that are expressed by their relative frequencies.
9
4.2 Corollary hypotheses
• Preference for unmarked coding (Langacker, 1991: 298):
H1: Speakers will tend to code subject-like NPs as subjects, using PRES-2.
10
4.2 Corollary hypotheses
• Usage-based memory model (e.g. Langacker, 1987: 59-60):- Forms that occur mainly in one pattern are stored as
partially filled instances of that schema.- This discourages speakers to use competing constructions
to express similar conceptualizations.
H2: The tenses for which the form of haber had a high token frequency in the PRES-1 pattern, but occurred only sporadically in other
constructions before PRES-2 began its advancement will disfavor PRES-2 .
11
4.2 Corollary hypotheses
• Language users tend to recycle structures (Goldberg, 2006: 120-125; Labov, 1994: Chap. 20):
H3: There will be priming effects at the argument-structure level.
• Linguistic change from below:H4: The variation will conform to the
Principles of Linguistic Change Labov (2001) formulates for changes from below.
12
5. Methods
• Recordings of 24 native speakers, residents of the Greater Santo Domingo Area.- Rougly 28 hours of speech/250,000 words.- Fieldwork took place in April-May, 2011.
• Stratified by: - Age (25-35 years; 55+ years).- Academic achievement (University vs. No university).- Gender (Male, Female).
• Post-stratified by: - Social class (Academic achievement, Housing, Profession).
13
5. Methods
• Three sections/speech styles:- +/- 30-minute sociolinguistic interview
• Included questions with the variable to test for comprehension priming.
- Reading task• 35 decision contexts.• 20 trials, 15 fillers.
- Questionnaire task.• 45 decision contexts.• 32 trials, 13 fillers.
14
5. Methods
• Mixed-effect logistic regression with Johnson’s (2009) Rbrul:- VARBRUL-style factor weight output:
• 0-0.5: factor disfavors variant• 0.5-1: factor favors variant
- Fixed effects:• Animacy, Definiteness/specificity, Distribution of the verb-forms
in the sixteenth century, Production priming, comprehension priming, academic achievement, age, gender, social class, interview section.
- Random intercepts:• Speakers• Lemmas of NPs’ heads
15
6. Results
46.7% N= 859
53.3% N=100
2
Plural haberSingular haber
16
6.1 Resemblance to prototypical subject
• Which objective factors can model object/subjecthood?- Best-known set (agent patient & topic focus)
cannot be used.- Animacy:
• Animates vs. inanimates (Du Bois, 1987)- Definiteness/specificity:
• Definite > Specific indefinite > indefinite (Langacker, 1991)
17
6.1 Animacy
Animates
Inanimate
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
.60
.40
.60
.40
Nouns Speakers
18
6.2 Degree of entrenchment of the verb-form in PRES-1
• Analysis of 10,000 tokens of 3rd-person haber in a sixteenth-century Latin-American corpus (CORDE):- Hay: practically exclusive to the PRES-1 pattern.
• Strongest cognitive representation: PRES-1 + hay.- Hubo: occurs primarily in PRES-1.
• Strongest cognitive representation: PRES-1 + hubo.- Había, haya, habrá and hubiera: used in 3 constructions.
• Strongest cognitive representation: independent node.- Habría & composed tenses: very infrequent.
• Strongest cognitive representation: independent node.
19
6.2 Degree of entrenchment of the verb-form in PRES-1
Habían, hubieran, hayan, habrán: mainly outside of presentational expressions
Habrían, composed tenses and verbal periphrases: infrequent
Hayn and hubieron: mainly in presentational expressions
0 0.2 0.4 0.6 0.8
.70
.68
.17
.73
.71
.13
Nouns Speakers
20
6.3 Priming
• Distance to the prime (in clauses)- 0-20 clauses- 21+ clauses
• Formal similarity to the prime- Same/different construction.- Same/different Tense, Mood, Aspect morphology
(TMA).
21
6.3 Priming: Production priming
PRES-2, identical verb-form
PRES-2, different verb-form
No earlier use/last use 21+ clauses removed
PRES-1, identical verb-form
PRES-1, different verb-form
0 0.2 0.4 0.6 0.8
.63
.58
.47
.45
.37
.64
.54
.47
.46
.38
Nouns Speakers
22
6.3 Production priming in language change
0.00%20.00%40.00%60.00%80.00%
4.80%
69.90%55.60%
3.80%
57.40%41.40%
6.70%
71.40% 64.40%
UnprimedPRES-1PRES-2
23
6.3 Priming: Comprehension priming
PRES-2
PRES-1
No earlier use/last use 21+ clauses removed
0 0.2 0.4 0.6 0.8
.62
.44
.43
.63
.45
.42
Nouns Speakers
24
6.4 Social class
Middle class
Lower class
Upper class
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
61
.46
.44
.60
.47
.42
Nouns Speakers
25
6.5 Gender
Note: p=0.007
Female
Male
40.00% 42.00% 44.00% 46.00% 48.00% 50.00% 52.00%
49.90%
43.60%
26
6.5 Gender (and age)
First generation Second generation0.00%
10.00%
20.00%
30.00%
40.00%
50.00%
60.00%
38.10%
48.80%52.40%
47.60%
Male Female
27
7. Discussion & conclusions
• Hypothesis 1 = confirmed- The most salient feature of protoypical subjects favors
PRES-2.
• Hypothesis 2 = confirmed- Those tense-forms that occurred mainly in presentational
clauses disfavor PRES-2.
• Hypothesis 3 = confirmed- Long-lasting priming effects at argument-structure level.
• Phenomenon = argument-structure variation
28
7. Discussion & conclusions
• Three general principles of language use constrain the
alternations:
• H1: The preference for unmarked coding encourages the use of
PRES-2 with NPs that approach the subject prototype.
• H2: Statistical preemption discourages the use of PRES-2 for
conceptualizations that match entrenched instances of PRES-1.
• H3: Priming extends the use of PRES-2 to conceptualizations that
match entrenched instances of PRES-1.
29
7. Discussion & conclusions
• Hypothesis 4 = partially confirmed- Social class pattern conforms to Labov’s (2001)
Principles of Linguistic Change.• The variation expresses social class identity for the entire speech
community.
- Gender is not statistically significant for all speakers.• Only for younger Dominicans the variation expresses
gender identity- Social meaning changes over time (Eckert 2008).
- Relatively early stage of an ongoing change from below (Labov, 2001: 307-309).
- Slowly progressing change.
30
ReferencesDU BOIS, J. (1987). The discourse basis of ergativity. Language. LXIII
(4), 805-855.ECKERT, P. (2008). Variation and the indexical field. Journal of sociolinguistics, XII (4), 453-476. Fontanella de Weinberg, M.B. (1992). Variación sincrónica y diacrónica de las construcciones con haber en el español americano. Boletín de filología,XXXIII, 35-46. GOLDBERG, A. (2006). Constructions at work. Oxford: Oxford University Press.GOLDBERG, A. (1995). Constructions. Chicago: Chicago University Press. JOHNSON, D. E. (2009). Getting off the GoldVarb standard: Introducing Rbrul for mixed-effects variable rule analysis . Language and Linguistics Compass , III (1), 359-383.
31
ReferencesLABOV, W. (2001). Principles of Linguistic Change. Vol. 2. Oxford:
Blackwell. LABOV, W. (1994). Principles of Linguistic Change. Vol. 1. Oxford:
Blackwell. LABOV, W. (1972). Sociolinguistic Patterns. Philadelphia: UPenn
Press. LAKOFF, G. (1987). Women, fire, and dangerous things. Chicago:
Chicago University Press. LANGACKER, R. (1991). Foundations of Cognitive Grammar. Vol.2.
Stanford: Stanford University Press. LANGACKER, R. (1987). Foundations of Cognitive Grammar. Vol.1.
Stanford: Stanford University Press.
32
Description of the variants
• Syntax:
• PRES-1: <[Locative] haber [Object]>
• PRES-2: <[Locative] haber [Subject]>
• The constructions do not specify the linear ordering of the
arguments.
• Boldface square brackets indicate profiled, omissible
arguments.
• The locative is an argument, not an adjunct (Lakoff 1987).
33
Description of the variants
• Semantics:
• POINTING-OUT Idealized Cognitive Model (Lakoff
1987).
• Argument roles:
• NP argument: zero.
• Locative: location.
34
Description of the variants
• Pragmatics:
• Hearer-New NP argument (Lakoff 1987).
• Social connotations:
• This kind of meaning can be modeled quite
straightforwardly in Cognitive Construction Grammar.
35
Social connotations in CCG
• Only the frequencies of otherwise ‘meaningless’ alternations can signal social meaning directly.
• Constructions that capture ‘meaningless’ alternations connect abstractions of observed frequencies (probabilities) to social meanings.
• Central social meaning: subgroup membership (Silverstein 2003).
• Metonymy can account for the variety of interpretations that Eckert (2008) points out.
• Extensions can be extended multiple times more, which leads to the fluid ‘indexical field’ proposed by Eckert (2008).
36
Social connotations in CCG
• With time and repetition, some of these extended meanings can become conventionalized (e.g. stylistic appropriateness can be considered a conventionalized extension of social class; Silverstein 2003).
• The context of the usage event will activate or background potential meanings (Langacker 1987).
37
Where did this come from?
• Occasional confusion (online constructional blends caused by analogy [Desagulier, 2005)) has always existed in Spanish.- E avién allí muchos engeños e muchas armas
‘And there, there were a lot of deceits and a lot of weapons.’ (13th century; Moreno-Bernal (1978: 290-291)
• Not a change: occasional glitches caused by analogy.
38
Where did this come from?
• Actuation:- Latin America/Canary Islands:
• Language/dialect contact through colonization. Large Population of L2/D2 speakers; formation of a new variety.
• Greater opportunities for social mobility.
- Catalan Language Area:• Language/dialect contact during industrialization process
of the 19th century.• Pluralization in Catalan Pluralization in Spanish.• Rapid expansion during 19th century.
39