A Three-Pronged Approach to Variation in Object Clitic Placement in Portuguese

Hannah Washington

0

Outline

n  The phenomenon q  Object pronoun clitics

n  The approach q  Random forests q  Mixed-effects regression q  Conditional inference tree

n  Frequency effects n  Discussion n  Conclusion

1

THE PHENOMENON

2

Object pronoun placement

3

Brazilian Portuguese

European Portuguese

Spanish

Eu te vejo. Eu vejo-te. Yo te veo. Eu vou te ver. Eu vou ver-te. Yo voy a verte.

Yo te voy a ver

Eu não te vou ver. Eu não te vou ver. Yo no te voy a ver.

Eu não vou te ver. Eu não vou ver-te. Yo no voy a verte.

Eu não vou ver-te.

Object pronoun clitics in Romance

n  Object pronouns = clitics in most Romance varieties

n  Placement is determined by finiteness of the

verb with which the clitic associates: q  Infinitives: clitic placement = postverbal

(enclisis) q  Conjugated verbs: clitic placement =

preverbal (proclisis)

4

Object pronouns: EP vs. BP n  In European Portuguese (EP):

q  Direct and indirect object (DO & IO) pronouns in EP are clitics n  me, te, lhe(s), nos, vos, o(s), a(s)

n  In Brazilian Portuguese (BP):

q  First and second person DO and IO pronouns are clitics in BP: n  me, te, nos

q  Archaic usage of third person clitics in BP: n  lhe(s)

q  Widespread use of tonic (subject) pronouns as object forms in BP: n  ela(s), ele(s), você(s)

5

“Triggered” contexts

Grammars and linguists alike say that certain preverbal operators trigger proclisis (preverbal placement) of clitic objects in EP: (Martins 1993, Barrie 2000, Cunha & Cintra 2002, Perini 2002, Luís 2009)

These proclisis “triggers” include: q  Wh words q  Complementizers / subordinating conjunctions q  Quanitifiers q  Certain adverbs q  Negation q  Focalized topics

6

Variation in EP

(1) Agora o que me interessava era ajudar aquela rapariga. ‘Then what concerned me was helping that girl.’

[RT B.Fonseca]

(2) Quer dizer que esforcei-me durante cinquenta minutos. ‘What I mean is that I pushed myself for fifty minutes.’

[PTePN-R.Ferro]

Examples from the Corpus do Português (Davies & Ferreira 2006-)

7

“Triggered” contexts BP

n  Some grammars say that the same proclisis triggers affect clitic placement in complex predicates in BP (Thomas 1969, Perini 2002)

n  Others argue that there is no effect of

triggers in BP (Galves, Moraes & Ribeiro 2005)

8

Variation in BP (3) não me podia livrar de meus fantasmas familiares ‘I could couldn’t free myself from my familiar ghosts’

[VE: OTeoV-P3T2]

(4) Os amigos sabiam que ele podia nos deixar a qualquer momento ‘The friends knew that he could leave us at any moment.’

[BCastordAmnRio12-5-1997]

(5) sentia falta de alguém que pudesse ensinar-me algo

‘I missed someone who could teach me something’ [AdC(22-5-97)]

Examples from the Corpus do Português (Davies & Ferreira 2006-)

9

…and in Spanish?

n  Is there an effect of trigger words in modern Spanish complex predicates? q  Some evidence (Davies 1995)

q  “Clitic climbing” (CC)

n  Variation in placement q  Grammaticalization (Davies 1995)

q  Persistent topics (Schwenter & Torres Cacoullos 2010)

q  Priming & animacy (Barnes, González López & Schwenter 2014)

10

Research questions

1.  What linguistic and social factors condition the selection of ‘non-normative’ clitic placement in EP and BP in triggered contexts, and are these factors the same in the two varieties of Portuguese?

2.  Does verbal frequency influence clitic placement and if so, what is the effect?

3.  How do the patterns relate to what has been found in prior work on related languages (e.g. varieties of Spanish)? Do EP and BP follow the same trajectories as these varieties with respect to object clitic placement?

11

Prior studies

12

Portuguese Spanish Verbal frequency (binary) Verbal frequency Topicality Topicality Object function (acc/dat/refl) Relative animacy Register Gramaticalization

Dialect Priming

Davies 1997, Andrade 2010a-c

Myhill 1988, Davies 1995, Schwenter & Torres Cacoullos 2010, Schwenter & Torres

Cacoullos 2014a, Schwenter & Torres Cacoullos 2014b, Barnes, González López

& Schwenter 2014

THE APPROACH

13

Tools for variationists n  Basic frequencies n  Multivariate analysis (e.g. Varbrul)

q  correlation analyses between independent variables and the dependent variable

n  Mixed-effects regression q  models built based on results of random forest q  conditioning factors at a macro level

n  Conditional inference trees (CITs) q  interactions between variables q  conditioning factors at a micro level

14

Corpus

n  Corpus do Português (Davies & Ferreira 2006-, http://www.corpusdoportugues.org)

n  Data from EP and BP q  Written & transcribed oral sources q  1900s-early 2000s

n  Following three proclisis triggers: q  Que (subordinating conjunction / complementizer, and “o que”) q  Não (sentential negation) q  Talvez (epistemic adverb)

15

Data n  EP “personal pronouns”: 1554 tokens

q  me, te, nos, lhe, lhes q  Simple & multi-verb (complex) predicates

n  EP 3rd person DOs: 1000 tokens q  o, a, os, as q  Simple & multi-verb (complex) predicates

n  BP: 810 tokens q  me, te, nos, lhe, lhes q  3rd DOs (o, a, os, as) excluded because of typical tonic

or null expression for 3rd DOs q  Multi-verb (complex) predicates only

16

EP personal pronouns

17

Random forest: ranking of independent variables affecting personal clitic pronoun placement EP

sort(clitics.cforest.varimp)

Mood

Tense_

Frequency_lexical

SubjectExpl

Pronoun_

TypeofPron_

Mode_

Subject_

DocType_

Frequency_aux

VerbForm

ConstructionType_

0.00 0.02 0.04 0.06 0.08

Random forest: ranking of independent variables affecting clitic placement of anaphoric DOs in EP

sort(EPDOs.cforest.varimp)

RD

BRD

CombinedTP_

CombinedRD_

TP

BTP

Prior_clitic_

Dist_prior_clitic_

Mood_

Obj_Animacy_

Prior_mention_SO_

Form_prior_ref_

Tense_

Subject_

DocType_

Subj_Animacy_

Verb_conj_frequency_in_corpus

Lexical_verb_freq

ConstructionType_

0.00 0.05 0.10 0.15 0.20 0.25

EP 3rd DOs

BP enclisis

18

Random forest: ranking of independent variables conditioning enclisis (V V CL) in BP

sort(BP_VVCL.cforest.varimp)

Subj_Animacy_

LexicalVerbFreq

Trigger

Mood

TypeofPron_

DocType_

Obj_Animacy_

DistPriorClitic_

Subject_

Pronoun_

Mode

ConstructionType

Exactyr_binary

SubjectExpl

Tense_

Prior_clitic_

ModalVerbFreq

0.000 0.002 0.004 0.006 0.008

Random forest: ranking of independent variables conditioning triggered proclisis (CL V V) in BP

sort(BP_CLVV.cforest.varimp)

Tense_

Mood

SubjectExpl

DocType_

LexicalVerbFreq

Prior_clitic_

Trigger

Obj_Animacy_

Subject_

DistPriorClitic_

ConstructionType

Pronoun_

Subj_Animacy_

ModalVerbFreq

Mode

TypeofPron_

Exactyr_binary

0.00 0.01 0.02 0.03

BP proclisis

Mixed effects regression: EP – personal pronouns

19

Estimate Std Error z Value p-value N/total N (%) (Intercept) -0.2360 0.2020 -1.168 0.2427 Modal 2.3843 0.2491 9.573 < 0.001 *** 242/316 (77%) Movement 0.5595 0.2362 2.369 < 0.01 * 40/71 (56% Simple/Perf/Prog -2.9532 0.2436 -12.125 < 0.001 *** 75/1108(7%) Fiction 0.0378 0.2060 0.183 n.s. 268/1089 (25%) Oral 1.3074 0.2315 5.648 < 0.001 *** 65/228 (29%) News/Acad -1.3452 0.2820 -4.769 < 0.001 *** 31/237 (24%) 1 Subj 0.6097 0.1226 4.973 < 0.001 *** 183/411 (45%) 2 & 3 Subj -0.6097 0.2216 -4.973 < 0.001 *** 181/1143 (16%) Me 0.3726 0.1180 3.159 < 0.01 ** 160/556 (29%) Other clitic form (te, nos, lhe(s))

-0.3726

0.1180 -3.159 < 0.01 ** 204/998 (20%)

Mixed effects regression: EP – personal pronouns

Factor Effect on enclisis

Verb

al

Con

stru

ctio

n Modal/Movement

FAVORS

Simple/Perf/Prog DISFAVORS D

ocum

ent

Type

Fiction n.s.

Oral FAVORS

News/Acad DISFAVORS

Subj

ect

pers

on

1 Subj FAVORS

2 & 3 Subj DISFAVORS

Obj

ect f

orm

Me FAVORS

Other clitic form (te, nos, lhe(s))

DISFAVORS

20

Mixed effects regression: EP – 3rd DOs

21

Estimate Std Error z Value p-value N/total N (%) (Intercept) -0.2349 0.2063 -1.139 0.2548 Modal/Movement/Progressive

2.9040 0.2985 9.729 < 0.001 *** 319/385 (83%)

Simple/Perf -2.9040 0.2985 -9.729 < 0.001 *** 33/615 (5%) Fiction 0.0524 0.1961 0.267 n.s. 239/607 (39%) Oral 0.7139 0.2575 2.772 < 0.01 ** 47/122 (39%) News/Acad -0.7663 0.2250 -3.406 < 0.001 *** 66/271 (24%) 1 & 2 Subj 0.3283 0.1469 2.236 < 0.05 * 106/237 (45%) 3 Subj -0.3283 0.1469 2.236 < 0.01 ** 246/763 (32%)

Mixed effects regression: EP – 3rd DOs

22

Factor Effect on enclisis

Verb

al

Con

stru

ctio

n Modal/Movement/Progressive

FAVORS

Simple/Perf DISFAVORS

Doc

umen

t Typ

e Fiction n.s.

Oral FAVORS

News/Acad DISFAVORS

Subj

ect

pers

on 1 & 2 Subj FAVORS

3 Subj DISFAVORS

Interim summary: EP data

n  Variables favoring “non-normative” enclisis in triggered environments in EP q  Complex predicates (querer, poder, dever,

etc. + inf) q  Oral registers q  1st person subjects (+2nd for 3rd DOs) q  1st person sing object (me)

23

Mixed effects regression: BP – V V CL

24

Estimate Std Error z Value p-value N/total N (%) (Intercept) -2.6019 0.2976 -8.744 <2e-16 *** Pre-1990 0.5936 0.2150 2.760 < 0.01 ** 92/420 (22%) Post-1990 -0.5936 0.2150 -2.760 < 0.01 ** 39/390 (10%) Prior enclitic 0.8420 0.2046 4.115 < 0.001 *** 69/228 (30%) Prior proclitic -0.2526 0.1959 -1.290 n.s. 54/451 (12%) No prior clitic -0.5894 0.3004 -1.962 < 0.05 * 8/131 (6%)

Mixed effects regression: BP – V V CL

25

Factor Effect on enclisis

Year

Pre-1990 FAVORS

Post-1990 DISFAVORS

Prim

ing

Prior enclitic FAVORS

Prior proclitic n.s.

Not available DISFAVORS

Mixed effects regression: BP – CL V V

26

Estimate Std Error z Value p-value N/total N (%) (Intercept) -2.1135 0.3315 -6.375 1.83e-10 *** Pre-1990 1.3584 0.2651 5.123 < 0.001 ** 154/420 (37%) Post-1990 -1.3584 0.2651 -5.124 < 0.001 *** 40/390 (10%) Human subj -0.6616 0.1612 -4.105 < 0.001 *** 132/644 (29%) Nonhuman subj 0.6616 0.1612 4.104 < 0.001 *** 62/166 (37%) Direct object -0.4718 0.2030 -2.324 < 0.05 * 39/235 (17%) Indirect object 0.8482 0.1755 4.834 < 0.001 *** 130/371 (35%) Reflexive pron -0.3764 0.2203 -1.709 n.s 25/204 (12%)

Mixed effects regression: BP – CL V V

27

Factor Effect on enclisis

Year

Pre-1990 FAVORS

Post-1990 DISFAVORS Su

bjec

t A

nim

acy Human subject DISFAVORS

Nonhuman subject FAVORS

Obj

ect f

unct

ion Direct object DISFAVORS

Indirect object FAVORS

Reflexive n.s.

Interim summary: BP

n  Contexts favoring “non-normative” enclisis in triggered environments in BP q  Pre-1990 time period q  Prior enclitics (V CL or V V CL)

n  Contexts favoring “normative” triggered

proclisis in BP q  Pre-1990 time period q  Nonhuman subjects q  Indirect objects

28

EP personal pronouns Conditional inference tree for placement of personal clitics in EP

ConstructionType_p < 0.001

1

modal, movement perfect, simple

Tensep < 0.001

2

future, present past

TypeofPron_p = 0.003

3

directobj, reflexive indirectobj

DocType_p = 0.047

4

fiction interview, news

Node 5 (n = 92)

10

0

0.2

0.4

0.6

0.8

1 Node 6 (n = 43)

10

0

0.2

0.4

0.6

0.8

1 Node 7 (n = 81)

10

0

0.2

0.4

0.6

0.8

1 Node 8 (n = 171)

10

0

0.2

0.4

0.6

0.8

1

Mode_p < 0.001

9

oral written

Node 10 (n = 195)

10

0

0.2

0.4

0.6

0.8

1

Subject_p < 0.001

11

1 23

Tensep = 0.017

12

future, present past

Node 13 (n = 94)

10

0

0.2

0.4

0.6

0.8

1 Node 14 (n = 78)

10

0

0.2

0.4

0.6

0.8

1

Pronoun_p < 0.001

15

lhe me

Node 16 (n = 570)

10

0

0.2

0.4

0.6

0.8

1 Node 17 (n = 230)

10

0

0.2

0.4

0.6

0.8

1

29

EP 3rd person DOs

Conditional Inference Tree for Anaphoric DOs in EP

ConstructionType_p < 0.001

1

simple modal

Subject_p < 0.001

2

1_2 3

DocType_p < 0.001

3

{acad/news, fiction} oral

Node 4 (n = 91)

10

0

0.2

0.4

0.6

0.8

1 Node 5 (n = 40)

10

0

0.2

0.4

0.6

0.8

1 Node 6 (n = 484)

10

0

0.2

0.4

0.6

0.8

1 Node 7 (n = 385)

10

0

0.2

0.4

0.6

0.8

1

30

Summary - EP

31

CONSTRUCTION TYPE

SUBJECT PERSON ANIMACY

Most important factor for both personal and 3rd DOs

Important in the grammars for simple predicates for personal and 3rd DOs

Important in the grammars for simple predicates for personal and 3rd DOs

Multi-verb predicates are the site of extremely high rates of non-normative behavior

1st person subjects show increased rate of non-normative enclisis, 2nd person subjects for 3rd person DOs

Clitic animacy < subject animacy = more enclisis (cf. Myhill 1988) Human subjects are entry point of analogical placement (enclisis)

BP – enclisis (V V CL)

Conditional Inference Tree for Enclisis in BP

Prior_clitic_p < 0.001

1

enclitic_diff, enclitic_same not_avail, proclitic_diff, proclitic_same

ModalVerbFreqp = 0.023

2

≤ 325.62 > 325.62

Node 3 (n = 26)

10

0

0.2

0.4

0.6

0.8

1 Node 4 (n = 202)

10

0

0.2

0.4

0.6

0.8

1

Exactyr_binaryp = 0.007

5

post pre

Node 6 (n = 310)

10

0

0.2

0.4

0.6

0.8

1 Node 7 (n = 272)

10

0

0.2

0.4

0.6

0.8

1

32

BP – proclisis (CL V V) Conditional Inference Tree for Triggered Proclisis in BP

Exactyr_binaryp < 0.001

1

post pre

Prior_clitic_p < 0.001

2

EN.diff, PRO.diff EN.same, not_avail, PRO.same

Subj_Animacy_p = 0.006

3

human nonhuman

Node 4 (n = 148)

10

0

0.2

0.4

0.6

0.8

1 Node 5 (n = 59)

10

0

0.2

0.4

0.6

0.8

1 Node 6 (n = 183)

10

0

0.2

0.4

0.6

0.8

1

TypeofPron_p < 0.001

7

DO, reflexive IO

Node 8 (n = 200)

10

0

0.2

0.4

0.6

0.8

1 Node 9 (n = 220)

10

0

0.2

0.4

0.6

0.8

1

33

Summary - BP

34

PRIMING DOCUMENT YEAR ANIMACY

Most important factor for enclisis

Most important factor for triggered proclisis; relevant for enclisis

Relevant for triggered proclisis

Low frequency verbs + enclitic prime = very high rates of enclisis

More enclisis and (“normative”) triggered proclisis in older documents, interacting with lack of priming / clitic animacy Both are older placement options

Noncanonical (nonhuman) subjects maintain normative grammar in modern BP IOs (human) maintain normative placement more in older usage

FREQUENCY EFFECTS

35

Verb frequency

36

Verb Frequency per million

words in Corpus do Port. Estar a ‘to be Xing’ 4289.53 Ir ‘to be going to X’ 2854.18 Poder ‘to be able to X’ 2992.59 Querer ‘to want to X’ 1402.66 Dever ‘should X’ 1072.97 Conseguir ‘to manage to X’ 549.46 Resolver ‘to decide to X’ 170.92

Verbal frequency - EP

0 1000 2000 3000 4000

7075

8085

9095

100

Plot of Enclisis in Triggered Contexts by Verbal Frequency: 3rd Person Anaphoric DOs

frequency of verb (per million words)

encl

isis

rate

vir

ir

poder

dever

saber

tentar

conseguir

querer

procurar

esperar

desejarjulgarpretenderintentarcostumar continuar começar

voltar

estardecidir

chegar

tencionarousarcessar

0 1000 2000 3000 4000

2040

6080

100

Plot of Enclisis Rate in Triggered Contexts by Verbal Frequency: Personal Pronouns

frequency of verb (per million words)

encl

isis

rate

ameaçar

conseguir

convircostumardecidirdesejar

dever

entenderesperar

estar

ir

lograrousar permitir

poder

precisarpretenderprocurarprometer

querer

resolver

saber

sentirtentar

vir

Rate of enclisis of personal pronouns in EP in complex predicates (modals, “movement” verbs) following proclisis triggers by verbal frequency. Multiple R-squared = 0.914.

Rate of enclisis of third person DOs in EP in complex predicates (modals, “movement” verbs, progressives) following proclisis triggers by verbal frequency. Multiple R-squared = 0.08833.

37

Verbal frequency - BP

0 1000 2000 3000 4000 5000

020

4060

80100

Plot of Enclisis in Triggered Contexts by Verbal Frequency: BP Personal Pronouns

frequency of verb (per million words)

encl

isis

rate

adorar

andar

cansardecessarde

chegar

começar

conseguir

continuara

decidir

deixarde

desejar

dever

estar

haverde

ir

pararde

passar

pensar

planejar

poder

precisarpretender

procurar

querer

resolver

saber

tender

tentarterde

terque

tornara

tratar

vir

0 1000 2000 3000 4000 5000

010

2030

4050

Plot of Triggered Proclisis in Triggered Contexts by Verbal Frequency: BP Personal Pronouns

frequency of verb (per million words)

proc

lisis

rate

(CL

V V

)

adorar andarcansardecessarde chegarcomeçar

conseguir

continuaradecidirdeixardedesejar

dever

estar

haverde

ir

pararde passarpensarplanejar

poder

precisar

pretender

procurar

querer

resolver

saber

tender

tentar

terdeterquetornara tratar

vir

0 1000 2000 3000 4000 50000

20

40

60

80

100

Plot of Untriggered Proclisis in Triggered Contexts by Verbal Frequency: BP Personal Pronouns

frequency of verb (per million words)

pro

clis

is r

ate

(V

CL

V)

adorar

andar

cansardecessarde

chegar

começar

conseguir

continuara

decidir

deixarde

desejardever

estarhaverdeir

pararde

passar

pensar

planejar

poder

precisar

pretenderprocurar

querer

resolver

saber

tender

tentar

terde

terque

tornara

tratar

vir

Rate of enclisis (V V CL) by governing verb frequency. Multiple R-squared = 0.06216.

Rate of triggered proclisis (CL V V) by governing verb frequency. Multiple R-squared = 0.189.

Rate of untriggered proclisis (V CL V) by governing verb frequency. Multiple R-squared = 0.00582.

38

Verbal frequency in EP

n  Lower frequency governing verb = more enclisis q  Personal pronouns – Very strong negative

correlation n  Near categorical enclisis with low frequency

q  3rd person DOs – Weak negative correlation n  Placement more dependent on verbal

construction type

39

Verbal frequency in BP

n  Similar correlations between frequency and placement are also seen in BP q  Proclisis (CL V V) – strongest correlation q  Proclisis (V CL C) – no correlation q  Enclisis (V V CL) – weak negative correlation

40

DISCUSSION

41

Discussion

n  Differences in conditioning factors in EP and BP q  EP: construction type, frequency, animacy, document

type (register) q  BP: priming, year, animacy, frequency

n  Change in progress? q  EP à greater enclisis, relatively stable variation over time q  BP à greater clitic placement between two verbs (V CL V) q  Abandonment of constraints based on presence of trigger in

favor of consistent placement throughout the system

42

Verbal frequency n  Analogical behavior in EP

q  Low frequency forms affected first (cf. Bybee 2002, Bybee 2007)

q  High frequency results in grammaticalization (CC) in Mexican Spanish (cf. Schwenter & Torres Cacoullos 2014a)

n  Low frequency + an enclitic prime produce more enclisis in BP q  Low frequency governing verbs result in either 0% or

100% enclisis based on the presence or absence of a prime

43

Priming n  Enclisis primes enclisis in BP

q  Obsolescing forms show stronger effect of priming cross-linguistically (cf. Schwenter 2015a, Schwenter 2015b)

q  Follows patterns seen with: n  (en)clitics in Asturian Spanish (Barnes, González López &

Schwenter 2014)

n  -se/-ra subjunctive in Spanish (Schwenter 2014)

n  nós/a gente subjects in BP (Beaton 2013, Schwenter 2015b)

n  ser/haber perfects (Rosemeyer 2015)

n  present/present progressive in Spanish (Torres Cacoullos 2015)

44

Animacy n  Nonhuman subjects correlate with ‘normative’

proclisis in EP and BP q  Canonical subjects invite analogical change q  Non-canonical subjects resist change

45

Conclusions

46

BP EP General Priming as key factor in choosing obsolescing form

Different grammars based on sentence/phrase structure and social context

Gradience in grammar

Near complete change to clitic placement between verbs

Stable variation (or very slow change in progress) based on verb phrase complexity

EP and BP speakers use different grammars for object pronoun placement

Nonhuman subjects maintain older placement patterns

Multi-verb predicates (+low frequency), canonical subjects, as entry point for analogical behavior

Like BP, diglossic situation in EP with respect to prescriptive vs. descriptive grammars

BP Priming as key factor in choosing obsolescing form

Near complete change to clitic placement between verbs

Nonhuman subjects maintain older placement patterns

EP Different grammars based on sentence/phrase structure and social context

Stable variation (or very slow change in progress) based on verb phrase complexity

Multi-verb predicates (+low frequency), canonical subjects, as entry point for analogical behavior

General Gradience in grammar

EP and BP speakers use different grammars for object pronoun placement

Like BP, diglossic situation in EP with respect to prescriptive vs. descriptive grammars

Future directions

n  Proclisis priming in BP: separation of different proclisis primes and their effect on proclisis targets (CL V, CL V V, and V CL V as primes)

n  Consideration of reflexive/inherent/passive ‘se’

n  Clitic cluster behavior (as a reflex of animacy?)

n  Investigation of Caribbean Spanish (Guy 2014)

47

THANK YOU! OBRIGADA! ¡GRACIAS!

48

References Andrade, Aroldo Leal de. 2010a. A subida de clíticos e o estatuto do complemento de verbos causativos do português clássico. Anais do Seta 4:84-98. Andrade, Aroldo Leal de. 2010b. A subida de clíticos em português: Um estudo sobre a variedade europeia dos séculos XVI a XX. Doctoral dissertation. Campinas, SP. Andrade, Aroldo Leal de. 2010c. The application of clitic climbing in European Portuguese and the role of register. In Claudia Borgonovo et al. (eds.), Selected Proceedings of the 12th Hispanic Linguistics Symposium, 97-108. Somerville, MA: Cascadilla Proceedings Project. Barnes, Sonia, Verónica González López and Scott Schwenter. 2014. Variable clitic position in Asturian Spanish. Paper presented at the 7th Workshop on Spanish Sociolinguistics (WSS7). Madison, WI. Barrie, Michael. 2000. Clitic Placement and Verb Movement in European Portuguese. MA Thesis, University of Manitoba. Beaton, Mary Elizabeth. 2013. The Role of Lexical Frequency and Morphosyntactic Processing: 1st Person Plural Expression in Rio de Janeiro Portuguese. Paper presented at the International Congress of Linguists, Geneva, Switzerland.

49

References (cont’d) Bybee, Joan. 2002. Word frequency and context of use in the lexical diffusion of phonetically conditioned sound change. Language Variation and Change 14(3): 261-290. Bybee, Joan. 2007. Frequency of use and the organization of language. Oxford/New York: Oxford University Press. Cunha, Celso and Lindley Cintra. 2002. Nova Gramática do Português Contemporâneo, 17th ed. Lisboa: Edições João Sá da Costa. Davies, Mark. 1995. Analyzing Syntactic Variation with Computer-Based Corpora: The Case of Modern Spanish Clitic Climbing. Hispania 78(2): 370-380. Davies, Mark. 1997. A corpus-based approach to diachronic clitic climbing in Portuguese. Hispanic Journal 17:93-111. Davies, Mark and Michael Ferreira. 2006-. Corpus do Português: 45 million words, 1300s-1900s. Available online at http://www.corpusdoportugues.org. Guy, Gregory R. 2014. Variation and Change in Latin American Spanish and Portuguese. In Patrícia Amaral and Ana Maria Carvalho (eds.), Portuguese-Spanish Interfaces: Diachrony, synchrony, and contact, 443-464. Amsterdam: John Benjamins.

50

References (cont’d) Luís, Ana R. 2009. Patterns of clitic placement: Evidence from 'mixed' clitic systems. In Patience Epps and Alexandre Arkhipov (eds.), New Challenges in Typology: Transcending the Borders and Refining the Distinctions, 11-34. Berlin: Walter de Gruyter. Myhill, John. 1988. The Grammaticalization of Auxiliaries: Spanish Clitic Climbing. Proceedings of the Fourteenth Annual Meeting of the Berkeley Linguistics Society: 352-363. Perini, Mário A. 2002. Modern Portuguese: A Reference Grammar. New Haven: Yale University Press. Rosemeyer, Malte. 2015. How usage rescues the system: Persistence as as conversation. In Aria Adli an, Marco García García, Göz Kaufmann (eds.), Variation in Languages: System- and Usage-based Approaches, 289-316. Berlin: Walter de Gruyter. Schwenter, Scott A. 2015a. Priming and morphosyntactic variation. Paper presented at the Workshop on Measuring Recency and Frequency in Diachronic and Synchronic Data. Albert-Ludwigs-Universität, Freiburg, Germany. Schwenter, Scott A. 2015b. Would you just die already? Priming and obsolescence in grammar. Arts and Humanities Inaugural Lecture Series, The Ohio State University.

51

References (cont’d) Schwenter, Scott A. and Rena Torres Cacoullos. 2010. Pragmatic Factors in the Placement of Direct Object Clitics in Spanish. Presented at OSUCHiLL 2010, Columbus, OH. Schwenter, Scott A. and Rena Torres Cacoullos. 2014a. Competing constraints on the variable placement of direct object clitics in Mexico City Spanish. Revista Española de Lingüística Aplicada 27(2):514-536. Schwenter, Scott A. and Rena Torres Cacoullos. 2014b. Variability in Spanish Clitic Placement: Animacy and Topicality. Paper presented at the 7th Workshop on Spanish Sociolinguistics (WSS7). Madison, WI. Torres Cacoullos, Rena. 2015. Gradual loss of analyzability: Diachronic priming effects. In Aria Adli an, Marco García García, Göz Kaufmann (eds.), Variation in Languages: System- and Usage-based Approaches, 265-288. Berlin: Walter de Gruyter.

52