Preposition Stranding in British English: ?Up with how much constraints do you have to put?

Preposition Stranding in British English:?Up with how much constraints do you have to put?

CoGETI Forschungsnetzwerk Constraintbasierte Grammatik:Non-Canonical Structures workshop University of Göttingen, 06.07.-07.07.2006

Thomas Hoffmann

(University of Regensburg)

1. Introduction

(1) About what will I talk?(2) What will I talk about?

1. Introduction

(1) About what will I talk?(2) What will I talk about?

(1) displacement of P about (“pied-piping”) (2) P about “in-situ” without complement (“stranded”)

1. Introduction

Preposition stranding as in (2) looks like normal long-distance filler-slot structure, but:

Not all languages allow P stranding, cf. e.g.:

(3) *Das Thema, das ich über sprechen werde(4) The topic which I will talked about

Which factors affect P stranding/pied-piping in E?

Can all stranded data be captured by a general construction/constraint? [which e.g. licenses SLASH-ed COMP-lists for P]

2. Stranding and Pied-Piping in English

In English stranding occurs in four structures

in which …:

i. Strandingi I’ve heard ofi. [preposing]

ii. Whati is he talking abouti? [open interrogative]

iii. What a great topici he talked abouti! [exclamative]

iv. the structure [whichi he talked abouti]. [wh-relative]

(cf. Pullum and Huddleston 2002: 627)



in which pied piping is an alternative option:

i. Of strandingi I’ve heardi. [preposing]

ii. About whati is he talkingi? [open interrogative]

iii. About what a great topici he talkedi! [exclamative]

iv. the structure [about whichi he talkedi ]. [wh-relative]




in which pied piping is not possible:

v. the structure [(thati) he talked abouti]. [non-wh relative]

vi. the same stuffi as [I talked abouti]. [comparative]

vii. His talki was easy [to find fault withi]. [hollow clause]

viii. Strandingi has been talked abouti enough]. [passive]


3. Roadmap: What to Expect

1. P placement across clause types (corpus)

2. Categorical RC data (corpus)

3. Magnitude Estimation experiments

4. Variable RC data (corpus)

5. Conclusion

4. Corpus Data

• Corpus used:

International Corpus of English ICE-GB (educated Present-day BE, written & spoken)

(tagged for Pstranded / parsed “P+Wh“ search)

• Analysis tool:

GOLDVARB computer programme (logistic regression; Robinson et al. 2001) relative influence of various contextual factors (weights: <0.5 = inhibiting factors; >0.5 = favouring)

Pstrand/pied-piped token tested for

1. Clause Type

2. displaced element (who, what, NP, etc.)

3. XP contained in (V / N, e.g. entrance to sth. / Adj, e.g. afraid of sth.)

4. level of formality

5. X-PP relationship (Vprepositional, PPLoc_Adjunct, PPMan_Adjunct …)

(e.g. Bergh, G. & A. Seppänen. 2000; Hoffmann 2005; Trotta 2000)

4. P placement across clause types

4.1 Categorical stranding contexts

1. Which PP types occur in categorical stranding contexts?

Type Token %

Passive 97 85

Hollow 14 12

Comparison 3 3

Sum 114

4.1 Categorical stranding contexts

0

20

40

60

80

100prepositionalV

complementPP

V-X-P-idioms

affectedLoc

instrument

accompaniment

Figure 1: Categorical stranding context by PP type (%)

Note: P stranding in passive tokens only with lexically specified stored /

associated V-P combinations

4.1 Categorical stranding contexts: Passive


(5) Prepositional Verb:Maybe his absence is is not properly dealt with

<ICE-GB:S1B-044 #60:2:B> (6) Complement PP:

King 's Canterbury is being spoken of very

highly at the moment <ICE-GB:S1A-054 #88:1:B> (7) V-X-P idiom:

it 'll be taken care of <ICE-GB:S2A-028 #60:2:A> (8) Affected location:

One of the benches had been sat upon

<ICE-GB:W2F-005 #97:1>

Note: P stranding in passive tokens only with lexically specified stored /

associated V-P combinations

features of Pstranded in passive sentences combination of:

general Pstranded constraint [which licenses SLASH-ed COMP-lists for P]

general passive construction [affected arguments as Subj]


Preliminaries: several categorical data excluded, e.g.:

• all categorical stranding contexts [cf. above]

• all that/-RCs [cf. later]

• idomatic constructions:What 's it like <ICE-GB:S1A-019 #53:1:B>

• non-finite RCs [cf. Sag 1997]

• all Manner, Degree, Respect PPs [cf. later]

4.2 Variable stranding contexts:


Type Stranded Pied piped

WH-RC N 69 439 508

% 14 86

Free RC N 136 2 138

% 99 1

DirectQ N 103 5 108

% 95 5

Indir Q N 66 7 73

% 90 10

Cleft N 8 49 57

% 14 86

Sum 382 502 884

Footnote: ? pied piped free RC data?

(9) This has tended to obscure to what extent Beckett 's early writings possess a coherent , though dislocated rhetoric of their own ...

<ICE-GB:W2A-004 #22:1>

= obscure the extent to which ...

[!But: specific PP type (degree); cf. later!]


Interestingly: Statistical analysis revealed

ClauseType * Formality interaction


Free RC / Indir Q / Direct Q: not affected by level of formality

0

10

20

30

40

50

60

70

80

90

100

F_stranded

F_piped

Q_stranded

Q_piped

I_stranded

I_piped

Informal Medium Formal

WH-RC: affected by level of formality

0

10

20

30

40

50

60

70

80

90

100

R_stranded

R_piped


Cleft-RC: affected by level of formality

0

10

20

30

40

50

60

70

80

90

100

L_stranded

L_piped


Best Goldvarb model for data: (Fit: X-square(7) = 4,006, p = 0,7784R2 = 0,99 / adjusted-multiple R2 = 0,99

Cross-validation estimate of accuracy = 0,922)

significant factors:

PP-types

Clause*Formal


with respect to pied piping:

<0.5 = inhibiting pied piping / favouring stranding

>0.5 = favouring pied piping / inhibiting stranding



PP type relationship(p = 0.000)

prepositional "X""V-X-P" idioms

subcategorized PP obligatory complement

0,169

optional complements 0,333

movement accompaniment

means/instrument cause/reason/result

0,547

position in timeaffected location

directionposition/location

0,941

Pstrand

Ppiped


Clause*Formal relationship(p = 0.000)

Free RCIndirect QDirect Q

0,028

less formal*

WH-RC/Cleft-RC0,134

more formal*

WH-RC/Cleft-RC0,904

Pstrand

Ppiped

Gries 2002: P placement affected by

1) processing effort

2) prescriptive grammar rules

Yes, but also:

3) idiosyncratic combination of both!


processing:stranding more complex than pied piping since

1) Hawkins 2004: potential processing problems

(11) Whoi did John see*i Bill talk toi

(12) To whomi did John see Bill talki

2) Stranding defers filler-gap identification beyond verbal head of clause


“Gap sites and nodes containing them that are predictable on the basis of conventionalized co-occurrence of their subcategorizers are easier to process than adjunct gaps and adjunct clauses.” (Hawkins 2004: 213)


“Gap sites and nodes containing them that are predictable on the basis of conventionalized co-occurrence of their subcategorizers are easier to process than adjunct gaps and adjunct clauses.” (Hawkins 2004: 213)

explains effect of factors in PP type:• lexically specified PPs favour stranding• stranding with adjunct PP: semantic factors

(cf. below)


In languages that have filler-gap structures for both relative clauses and wh-questions, if a gap is grammatical for a relative clause filler in an FGD of complexity n, then a gap will be grammatical for a wh-question filler in an FGD of complexity n. (Hawkins 2004: 200)


In languages that have filler-gap structures for both relative clauses and wh-questions, if a gap is grammatical for a relative clause filler in an FGD of complexity n, then a gap will be grammatical for a wh-question filler in an FGD of complexity n. (Hawkins 2004: 200)

partly explains effect of Clause*Formal:

• Free-RC/Q less complex than RC favour Pstrand

• yet: level of formality interaction effect?


Note: if only processing effect

only need for one general Pstrand construction

Yet: level of formality only associated with Cleft-/WH-RCs

!require extra Pstrand and Ppiped constructions

for these clause types!


As the ICE-GB data showed both stranding and pied piping occur mostly in relative clauses

closer look at RC data

[further constraints beyond formality?]

5. Corpus Study II: Relative clauses

1. relativizer:

all that/Ø-tokens in ICE-GB stranded

176 that+Pstranded-token

(10) a data source on that I can rely

177 Ø+Pstranded-token

(11) a data source on Ø I can rely

ICE-GB result: expected

implications: (2) = (3)? / that WH-


2. X-PP relationship:

ICE-data showed: difference between adjunct PPs

claim:

Pstranding restricted to PPs which add thematic information to predicates/events

= processing constraint: allows integration of P within VP



Categorical effect of non-θ-WH-PPAdjuncts-tokens:

a) just P+WH / no that/Ø+P in ICE-GB: e.g. manner adjunct PPs:

(12) a. the ways in which the satire is achieved <ICE-GB:S1B-014 #5:1:A>

b. the ways which/that/Ø the satire is achieved in



Categorical effect of θ-WH-PPAdjuncts-tokens:

b) just P+WH / but that/Ø+P in ICE-GB: e.g. locative PP adjuncts

(13) a. … the world that I was working in and studying in

<ICE-GB:S1A-001 #35:1B>

b. … the world in which I was working and studying


Claim: comparison of WH- vs that/Ø shows:

P can only be stranded if: PP adds thematic information to predicates/events[= can be semantically integrated by head of RC]

e.g.: manner & degree adjuncts:compare events “to other possible events of V-ing” (Ernst 2002: 59)

don’t add thematic participant Pstrand with these: systematic gap


Claim: comparison of WH- vs that/Ø shows:

P can only be stranded if: PP adds thematic information to predicates/events[= can be semantically integrated by head of RC]

e.g.: locative adjuncts:

add thematic participant WH+P with these: accidental gap


Comparison of WH- vs that/Ø good evidence, but:“negative data” problem

further corroborating evidence neededIntrospection: Magnitude Estimation study


• relative judgements (reference sentence)

• informal, restrictive RCs tested for:

P-PLACEMENT (Pstrand, Ppied-piped)RELATIVIZER (WH-, that-, Ø-)X-PP (VPrep, PPTemp/Loc_Adjunct, PPManner/Degree_Adjunct)

• tokens counterbalanced: 6 material groups a 18 tokens + 36 filler = 54 tokens

• tokens randomized (Web-Exp-software)

• N = 36 BE native speakers (sex: 18m, 18f / age: 17-64)

6. Magnitude Estimation: RC I

18 filler sentences: ungrammatical

a. That’s a tape I sent them that done I’ve myself (word order violation; original source: <ICE-GB:S1A-033 074>)

b. There was lots of activity that goes on there (subject contact clause; original source: <ICE-GB:S1A-004 #067>)

c. There are so many people who needs physiotherapy (subject-verb agreement error; original source: <ICE-GB:S1A-003 #027>)


ANOVA: significant effects

• P-PLACEMENT: F(1,33) = 4.536, p < 0.05

• RELATIVIZER: F(2,66) = 17.149, p < 0.001

• P-PLACEMENT*X-PP: F(2,66) = 9.740, p < 0.001

• P-PLACEMENT*RELATIVIZER: F(2,66) = 4.217, p < 0.02


-2

-1,5

-1

-0,5

0

0,5

1

1,5

2M

ean

Ju

dg

me

nts

(z-

sco

res)

P+WH

P+That

P+0

prepositional verbs temp/loc adjuncts manner/deg adjuncts

Fig. 1: Magnitude estimation result for P + relativizer

P+WH >> P+that > P+Ø

Fig. 2: Magnitude estimation result for P + relativizercompared with fillers

P+that & P+Ø = ungrammatical fillers violation of “hard constraint” (Sorace & Keller 2005)

-2

-1,5

-1

-0,5

0

0,5

1

1,5

2M

ean

Ju

dg

men

ts (

z-sc

ore

s)

P+WH

P+That

P+0

Filler (grammatical)

Filler (*Agree)

Filler(*ZeroSubj)

Filler(*WordOrder)


-2

-1,5

-1

-0,5

0

0,5

1

1,5

2M

ean

Ju

dg

me

nts

(z-

sco

res)

WH+P

That+P

0+P


Fig. 3: Magnitude estimation result for relativizer + P

WH + P= that + P = Ø + PVPrep > PPTemp/Loc > PPMan/Deg

-2

-1,5

-1

-0,5

0

0,5

1

1,5

2M

ean

Ju

dg

me

nts

(z-

sco

res)

X+P

Filler_Good

Filler(*Agree)

Filler(*ZeroSubj)

Filler(*WordOrder)


Fig. 3: Magnitude estimation result for relativizer + P

VPrep > PPTemp/Loc > PPMan/Deg >> ungrammatical filler violation of “soft constraint” (Sorace & Keller 2005)


Corroborating evidence:

corpus: man/deg PPs: no Pstranded (not even with that/) semantic constraint on Pstranded

experiment:man/deg PPs worst environment for Pstranded yet: better than ungrammatical fillers

(soft constraint violation: processing effect)


What type of hard constraint is P + that?

Sag 1997: case assignment restriction

*P + that = *P + who

new Magnitude Estimation experiment

• informal, restrictive RCs, just VPrep tested for:

P-PLACEMENT (Pstrand, Ppied-piped, Pdoubled )RELATIVIZER (who, whom, that-, Ø-)COMPLEXITY (simple, long-distance Ø- and that-C)

• tokens counterbalanced: 36 material groups a 36 tokens + 48 filler = 84 tokens

• tokens randomized (Web-Exp-software)

• N = so far: 13 BE native speakers

• in progress no in-depth statistical analysis

7. Magnitude Estimation: RC II

-2

-1,5

-1

-0,5

0

0,5

1

1,5

2

SimplePiped

that who whom zero

Fig. 4: Magnitude estimation result for all relativizers

-2

-1,5

-1

-0,5

0

0,5

1

1,5

2

SimplePiped

that who whom zero

Fig. 4: Magnitude estimation result for all relativizers

P + that P + who

7. Magnitude Estimation: RC II

if experiment shows

*P + that *P + who

3 separate constructions?:

(thati) ... Pi

wh-i ... Pi

P wh-i ... ti

In addition to PP-types and Formality effects, variable corpus data (450 finite WH-token)exhibited two more effects (Hoffmann fc.):

1. NP-contained PPs favour pied piping 0.964

2. restrictive RC favour pied piping: (weight: 0.592) nonrestrictive RC clearly inhibit pied piping

(i.e. favour stranding; weight: 0.248)

(Model: Fit:X-square: p = 0,5610 / R2 = 0,92 / multiple adjusted R2 = 0,90 / Cross-validation estimate of accuracy = 0.916)

8. Corpus Study III: Variable RC data

Note: both processing effects

1. NP-contained PPs favour pied piping:

NP itself contained in VP: Pstrand complexity[cf. also Cowart 1997]

2. nonrestrictive RC favour stranding: filler-gap identification process in non-restrictive relative clauses less complex than in restrictive relative clauses (Hawkins 2004: 240ff.)

less complexity Pstrand

8. Corpus Study III: Variable RC data

Preposition stranding: non-canonical English structure

• some properties of Pstranding attributable to processing complexity:

Q > RC non-restrictive > RC restrictiveVPrep > thematic PPAdjunct > non- thematic PPAdjunct

• others call for specific constructions:

formality effect with RCsthat-/-RCs

8. Conclusion

9. References

Aarts, B. 2000. "Corpus linguistics, Chomsky and Fuzzy Tree Fragments". In Christian Mair and Marianne Hundt, eds. 2000. Corpus Linguistics and Linguistic Theory. Amsterdam and Atlanta, GA: Rodopi, 5-13.

Bard, E.G. et al. 1996. “Magnitude Estimation of Linguistic acceptability”. Language 72:32-68.

Bergh, G. & A. Seppänen. 2000. “Preposition stranding with wh-relatives: A historical survey”. English Language and Linguistics 4:295-316.

Cowart, W. 1997. Experimental Syntax: Applying Objective Methods to Sentence Judgements. Thousand Oaks: Sage.

Gries, S.Th. 2002. “Preposition stranding in English: Predicting speakers' behaviour”. In V. Samiian, ed. Proceedings of the Western Conference on Linguistics. Vol. 12. California State University, Fresno, CA, 230-241

Hawkins, J. A. 2004. Efficiency and Complexity in Grammars. Oxford: Oxford University Press.

9. References

Hoffmann, T. 2005. "Variable vs. categorical Effects: Preposition pied piping and stranding in British English relative clauses". Journal of English Linguistics 33,3: 257-297.

Hoffmann, T. fc. “’I need data which I can rely on’. Corroborating Empirical Evidence on preposition placement in English relative clauses”. W. Sternefeld et al., eds. Linguistic Evidence 2006. Berlin: Mouton de Gruyter

Huddleston, R. et al. 2002. “Relative constructions and unbound dependencies”. In: G.K. Pullum & R. Huddleston, eds. The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press, 1031-1096.

Jackendoff, R. 2002. Foundations of Language: Brain, Meaning, Grammar, Evolution. Oxford: Oxford University Press.

Nelson, G. et al. 2002. Exploring Natural Language: Working with the British Component of the International Corpus of English. Amsterdam, Philadelphia: Benjamins.

Pesetsky, D. 1998. “Some principles of sentence production”. In: Pilar Barbosa et al., eds. Is the Best Good Enough? Optimality and Competition in Syntax. Cambridge, MA: MIT Press, 337-83.

9. References

Pickering, M. & G. Barry. 1991. “Sentence processing without empty categories”. Language and Cognitive Processes 6:229-259.

Quirk, R. et al. 1985. A Comprehensive Grammar of the English Language. London: Longman.

Robinson, J. et al. 2001. “GOLDVARB 2001: A Multivariate Analysis Application for Windows”. <http://www.york.ac.uk/depts/lang/webstuff/goldvarb/manualOct2001>

Sag, I.A. 1997. “English relative constructions”. Journal of Linguistics 33:431-484.

Sampson, G. 2001. Empirical Linguistics. London, New York: Continuum.

Trotta, J. 2000. Wh-clauses in English: Aspects of Theory and Description. Amsterdam and Philadelphia, GA: Rodopi.

Van der Auwera, J. 1985. “Relative that — a centennial dispute”. Journal of Linguistics 21:149-179.

Documents

Preposition Stranding in British English: ?Up with how much constraints do you have to put?