Persistence and Entrenchment in Language Change: The Spanish Past Subjunctive

PERSISTENCE AND ENTRENCHMENT IN LANGUAGE CHANGE: THE SPANISH PAST SUBJUNCTIVE!Malte Rosemeyer (Albert-Ludwigs-Universität Freiburg)!Scott Schwenter (The Ohio State University)!Measuring recency and frequency in synchronic and diachronic corpus data, 18/02/2015!

!

QUESTION ADDRESSED IN THIS TALK!

•  What diachronic evidence is there for the conserving effect of persistence? Does persistence cause conservation or is it caused by conservation?!

•  How does the conserving effect of persistence in language change interact with the conserving effect of frequency?!

!

2

1.  PERSISTENCE AND ENTRENCH-MENT!

2.  –RA VS. –SE!3.  DATA!4.  DESCRIPTIVE!5.  INFERENTIAL!6.  CONCLUSION!

AN EXAMPLE: SPANISH AUXILIARY SELECTION!

•  Old Spanish: variation between haber and ser!(1) las creaturas de que avié fablado (c. 1275, GE I)!

the creatures of which had talked !(2) su padre era muerto (c. 1275, GE I)!

his father was dead/died!(3) vío cómo eran ya bueltos (c. 1275, GE I)!

saw how were already returned!!

3




•  After 1425: Loss of ser!(1) las creaturas de que avié fablado (c. 1275, GE I)!

the creatures of which had talked !(4) su marido había muerto (1525-1529, CBE)!

his father had died!(5) á la ida habían ido á la parte de poniente!

at the way-there had gone to the direction of west!y á la vuelta habían vuelto por la de levante!and at the way-back had returned by that of east!(c. 1550, CEC)!

!4




5

Development of auxiliary selection

Pro

porti

on s

er +

Ptc

P

1270-1299

1300-1349

1350-1399

1400-1449

1450-1499

1500-1549

1550-1599

1600-1649

1650-1699

0 %

20 %

40 %

60 %

80 %

100 %



CONSERVATION OF SER!

•  Two conserving effects in the history of Spanish auxiliary selection Rosemeyer 2014!

1.  Entrenchment by high token frequency!2.  Persistence!

!

6



ENTRENCHMENT IN UBL!

•  Linguistic forms with high absolute frequency undergo a process of entrenchment LANGACKER 1987: 59, NEWELL 1990, ELLIS 1996, BYBEE/HOPPER 2001, BYBEE 2006, 2010!

•  Entrenchment leads to the dissociation of the high-frequency syntagms from the original construction (ser + PtcP)!

•  When the original construction is affected by a replacement process, these entrenched syntagms are less affected, and conserved as a result!

!7



ENTRENCHMENT AND USE OF SER + PTCP!

8 Rosemeyer 2014: 209!



PERSISTENCE!

•  Persistence: the use of one type of linguistic element in a discourse raises the probability of the repetition of its use in later discourse Szmrecsanyi 2005, Gries 2005!

•  Persistence was found to be a conserving effect of ser-selection in Rosemeyer (2011, 2014):!1.  The loss of ser + PtcP occurs at a slower rate in

persistence contexts!2.  In classical Spanish (after 1525), ser + PtcP is

used less conservatively in persistence contexts!! 9



SLOWER RATE OF LOSS OF SER + PTCP!

10

1275 1325 1375 1425 1475 1525 1575 1625

20%

40%

60%

80%

100%Presence of persisting BE + PPAbsence of persisting BE + PP

Time

Per

cent

age

of B

E +

PP

Rosemeyer (2011)!



SER + PTCP IS USED LESS CONSERVATIVELY!

•  Relevance of reflexivity:!!

11



SER + PTCP IS USED LESS CONSERVATIVELY!

•  Use of ser + PtcP in Classical Spanish (after 1525):!!

12

PERSISTENCE = NONE OR HABER PERSISTENCE = SER

Reflexive = F Reflexive = T Reflexive = F Reflexive = T

haber 63.8% (1233/1932) 98.7% (464/470) 33.5% (59/176) 85.2 % (23/27)

ser 36.2% (699/1932) 1.3% (6/470) 66.5% (117/176) 14.8% (4/27)

TOTAL 1932 470 176 27



ENTRENCHMENT VS. PERSISTENCE!

•  Entrenchment is based on repetition, persistence is based on activation!

•  Entrenchment has a longer “temporal depth” than activation!

•  Entrenchment leads to irregularity, persistence leads to the neutralization of irregularity!

!

13



-RA VS. -SE!

14

•  Past subjunctive (examples from CdE)!(6) Habría incluso que vender el aire contaminado !

would.have even that sell the air contaminated!que respiran los españoles a alguien que fuera capaz!that breathe the Spaniards to someone that were capable! de hacer que se pueda, por fin, respirar en Madrid!of do that refl can finally breathe in Madrid!

(7) Creo que tengo un lugar merecido, pero tampoco !I.believe that I.have a place deserved but also.not!tanto como si yo fuese Garcilaso o Góngora.!so.much as if I was Garcilaso or Góngora!



DIACHRONIC DEVELOPMENT!

15

1200-1299

1300-1399

1400-1499

1500-1599

1600-1699

1700-1799

1800-1899

1900-2000

02000

4000

6000

8000

Freq

uenc

y pe

r mill

ion

Diachronic development of -ra vs. -se

-ra-se1.  PERSISTENCE

AND ENTRENCH-MENT!


SYNCHRONIC STUDY: DATA EXTRACTION!

16

•  Corpus del español (Davies 2002), 20th century section, spoken & written data, 20.5 million tokens, POS tagged!

•  Identification of –se forms with n=>10; identification of the 32 verb lemmas; sorting according to lemma frequency in the CdE!

•  Extraction of all the 20th century tokens of –ra and –se-forms for two lemmas each of maximal, high, low and minimal frequency > eight lemmas total!

•  Randomization of ser, haber, dar to 500 occurrences!



DATA OVERVIEW!

17

Verb Lemma frequency n -ra n -se n total

ser 18709.7 436 64 500

haber 9848.1 431 69 500

ver 2292.70 384 48 432

dar 2011.65 464 31 495

salir 716.95 211 12 223

sentir 657.50 101 14 115

caer 322.05 93 11 104

acabar 317.15 90 10 100



ENTRENCHMENT: TYPE-TOKEN RATIO!

18

•  The finite forms fuese and hubiese (3sg) make up about 50 % of all the verb forms that have more than 10 –se tokens!

•  Looking at the first 1000 –ra forms and first 1000 –se forms:!-  –ra forms: 33.9 types per token (33930/1000)!-  –se forms: 4.4 types per token (4424/1000)!



ENTRENCHMENT: NUMBER MORPHOLOGY!

19

SINGULAR PLURAL

-ra 88.5% (1694/1915) 93.1% (516/554)

-se 11.5% (221/1915) 6.9% (38/554)

TOTAL 1915 554 X-squared = 9.5(1)p <.01**



ENTRENCHMENT: LEMMA FREQUENCY !

20

LOWER FREQUENCY (all other verbs)

HIGH FREQUENCY (ser & haber)

-ra 91.4% (1343/1469) 86.7% (867/1000)

-se 8.6% (126/1469) 13.3% (133/1000)

TOTAL 1469 1000

X-squared = 13.6(1)p <.001***



LEMMA FREQUENCY VS. NUMBER MORPHOLOGY!

21



-ra 91.4% (1343/1469) 86.7% (867/1000)

-se 8.6% (126/1469) 13.3% (133/1000)

TOTAL 1469 1000

X-squared = 13.6(1)p <.001***



Singular Plural Singular Plural

-ra 90.4% (990/1095) 94.4% (353/374) 85.9% (704/820) 90.6% (163/180)

-se 9.6% (105/1095) 5.6% (21/374) 14.1% (116/820) 9.4% (17/180)

TOTAL 1095 374 820 180

X-squared = 5.1(1)p <.05*

X-squared = 2.4(1)p > .05 (n.s.)



PERSISTENCE: CODING!

22

•  Extraction of all of the preceding context in the –ra and –se tokens from the CdE, annotation for preceding –ra and –se tokens!

•  Result: variable RECENT representing the last preceding –ra or–se tokens, with three levels: none, se, ra!

•  Coding for the distance in words between the target token and the prime: variables DIST_TO_LAST_SE and DIST_TO_LAST_RA!



PERSISTENCE: VARIABLE RECENT!

23

RECENT = NONE RECENT = RA RECENT = SE

-ra 90.4% (1483/1641) 93.5% (676/723) 48.6% (51/105)

-se 9.6% (158/1641) 6.5% (47/723) 51.4% (54/105)

TOTAL 1641 723 105

ra: 1.03 x more frequent in recent=ra contexts !se: 5.44 x more frequent in recent=se contexts !

X-squared = 157.7(1) p <.001***

X-squared = 5.8 (1)p < .05*



PERSISTENCE: DISTANCE TO PRECEDING -SE!

24



recent = se

distance.in.words

subj$Dependent[x]

0 10 20 30 40 50 60 70 80 90

rase

0.0

0.2

0.4

0.6

0.8

1.0

PERSISTENCE: DISTANCE TO PRECEDING -RA!

25



recent = ra

distance.in.words

subj$Dependent[x]

0 10 20 30 40 50 60 80 100

rase

0.0

0.2

0.4

0.6

0.8

1.0

INTERACTION PERSISTENCE & NUMBER!

26

RECENT = NONE OR RA RECENT = SE

Singular Plural Singular Plural

-ra 90.3% (1656/1834) 94.9% (503/530) 46.9% (38/81) 54.2% (13/24)

-se 9.7% (178/1834) 5.1% (27/530) 53.1% (43/81) 45.8% (11/24)

TOTAL 1834 530 81 24

X-squared = 10.5(1)p <.01**

X-squared = 0.2(1)p > .05 (n.s.)



CONDITIONAL INFERENCE TREE!

27



•  Dependent variable: SE (vs. RA)!•  Inclusion of the following predictor variables:!

-  RECENT (NONE/RA/SE)!-  IDENTITY (BETWEEN PRIME LEMMA AND TARGET LEMMA) (T/F)!-  HIGH TARGET FREQUENCY (T/F)!-  TARGET PERSON (1/2/3)!-  TARGET NUMBER (SG/PL)!-  DISTANCE TO LAST RA (IN WORDS)!-  DISTANCE TO LAST SE (IN WORDS)!

28



recentp < 0.001

1

{none, ra} se

Target_FreqHIGHp = 0.002

2

≤ 0 > 0

Target_Numberp = 0.027

3

s p

Node 4 (n = 1048)

sera

0

0.2

0.4

0.6

0.8

1 Node 5 (n = 359)

sera

0

0.2

0.4

0.6

0.8

1 Node 6 (n = 957)

sera

0

0.2

0.4

0.6

0.8

1

dist_to_last_sep = 0.027

7

≤ 18 > 18

Node 8 (n = 49)

sera

0

0.2

0.4

0.6

0.8

1 Node 9 (n = 56)

sera

0

0.2

0.4

0.6

0.8

1

SUMMARY OF RESULTS!

29



•  entrenchment important for the use of –se:!-  lower TTR, restriction to singular morphology!-  occurs more frequently with high-frequency verbs!-  for high-frequency verbs, higher TTR!

•  persistence has an influence on –se / –ra use:!-  –se more likely with preceding –se!-  much stronger persistence effect of preceding –se!-  persistence leads to a higher TTR!

DISCUSSION!

30

•  Productivity is not only sensitive to paradigmatic, but also syntagmatic factors!

•  Both entrenchment and persistence have been shown to be correlated with a rise in productivity!

•  However, this relationship is not of a causal nature for entrenchment: it is due to the fact that –se is more frequent with frequent verbs that its TTR is higher > correlation!

•  For persistence, the relationship is of a causal nature: persistence causes a rise in the productivity of –se !



REFERENCES!

31

•  Bybee, Joan L. and Paul J. Hopper (2001): "Introduction to frequency and the emergence of linguistic structure", in: Joan L. Bybee and Paul J. Hopper, eds.: Frequency and the Emergence of Linguistic Structure. Amsterdam, Philadelphia: Benjamins, 1-26.

•  Bybee, Joan L. (2006): "From usage to grammar: the mind’s response to repetition", in: Language 82 (4), 711-733.

•  Bybee, Joan L. (2010): Language, Usage, and Cognition. Cambridge, New York: Cambridge University Press.

•  Davies, Mark (2002): Corpus del español (100 million words, 1200s-1900s). Available online at http://www.corpusdelespanol.org. Last access: 17 February 2015.

•  Ellis, Nick (1996): "Sequencing in SLA: phonological memory, chunking and points of order", in: Studies in Second Language Acquisition 18, 91-126.

•  Gries, Stefan Th. (2005): "Syntactic priming: a corpus-based approach", in: Journal of Psycholinguistic Research 34 (4), 365-399.

•  Langacker, Ronald W. (1987): Foundations of Cognitive Grammar. Theoretical Prerequisites. Stanford: Stanford University Press.

•  Newell, Allen (1990): Unified Theories of Cognition. Cambridge: MIT Press.

•  Rosemeyer, Malte (2011): "Persistence and analogy in the history of auxiliary selection in Spanish", Talk given at the Workshop "System, Usage, and Society", Freiburg, 12 November 2011.

•  Rosemeyer, Malte (2014): Auxiliary Selection in Spanish. Gradience, Gradualness, and Conservation. Amsterdam, Philadelphia: Benjamins.

•  Szmrecsanyi, Benedikt (2005): "Language users as creatures of habit: a corpus-based analysis of persistence in spoken English", in: Corpus Linguistics and Linguistic Theory 1 (1), 113-150.!



Persistence and Entrenchment in Language Change: The Spanish Past Subjunctive

Documents