1 of 33

Comparing reflexive and personal pronouns in Chinese locative prepositional phrases

Jiaxing Yu and Shannon Bryant

Rutgers University

Linguistic Society of America January 4-7, 2024

2 of 33

Reflexives vs. personal pronouns

In many languages, reflexives (e.g., herself) and personal pronouns (e.g., her) can both be used to refer to someone previously mentioned in the sentence.

Sometimes the choice between pronoun forms is fixed:

(1) a. Michele congratulated herself / *her.

b. Michele’s brother congratulated her / *herself.

2

3 of 33

Reflexives vs. personal pronouns

Strict complementarity is captured by syntactic binding conditions.

Condition A: a reflexive must be bound in its local domain.
Condition B: a personal pronoun cannot be bound in its local domain.

3

Throw-type vs. Spray/Load-type: B=0.21, p=0.155

(Chomsky 1981, a.o.)

4 of 33

Pronoun choice in LPPs

Complementarity often breaks down in locative prepositional phrases (LPPs):

(2) Michele set a glass next to her / herself.

We assume that Conditions A and B are both satisfied for this kind of sentence.

(Reinhart & Reuland 1993, Büring 2005, Bryant 2022, a.o.)

LPPs are ideal for investigating non-syntactic factors that shape pronoun choice.

4

5 of 33

Two semantic factors: �Event type and relation type

Event type: motion vs. perception

(3) a. Chloe poured some glitter on her/herself.

b. Chloe noticed some glitter on her/herself.

Relation type: +contact vs. -contact

(4) a. Chloe poured some glitter on her/herself.

b. Chloe poured some glitter next to her/herself.

5

6 of 33

Event type and relation type in English LPPs

6

> Reflexives are relatively better in motion sentences than in perception sentences, while personal pronouns are the opposite.

> Reflexives are relatively better in contact LPPs than in non-contact LPPs, while personal pronouns are the opposite.

Throw-type vs. Spray/Load-type: B=0.21, p=0.155

(Bryant 2022a-b, N=122)

7 of 33

7

Is this pattern specific to English, or does it occur in other languages, too?

8 of 33

Today’s talk: LPP pronoun choice in Chinese

Part 1: Chinese pronoun forms

Part 2: Acceptability judgment experiment

Part 3: Main takeaways

8

9 of 33

Part 1: �Chinese pronoun forms

9

10 of 33

Three pronoun forms

> Simplex reflexive ziji (自己)

> Complex reflexive ta-ziji (她自己/他自己)

> Personal pronoun ta (她/他)

10

11 of 33

Reflexives vs. personal pronouns

Reflexives and personal pronouns are in complementary distribution in simple transitive sentences:

(5) 李娟祝贺自己/她自己/*她。

(5) Lijuan zhuhe ziji / ta-ziji / *ta

(5) Lijuan congratulated self / herself / her

(5) ‘Lijuan congratulated herself / *her.’

(6) 李娟的哥哥祝贺她/*自己/*她自己。

(5) Lijuan de ge-ge zhuhe ta / *ziji / *ta-ziji

(5) Lijuan DE brother congratulated her / self / herself

(5) ‘Lijuan’s brother congratulated her / *herself.’

11

12 of 33

Pronoun choice in LPPs

All three pronoun forms are allowed in LPPs:

(7) 李娟把亮片洒到她/自己/她自己身上。

(18) Lijuan ba liang-pian sa dao ta / ziji / ta-ziji shen shang.

(18 )Lijuan ba glitter pour to her / self / herself body on

(18)‘Lijuan poured glitter on her / herself.’

12

13 of 33

Main empirical questions

Is pronoun choice in Chinese LPPs sensitive to event type and relation type?
How do simplex ziji and complex ta-ziji compare?

13

14 of 33

Comparison between ziji and ta-ziji

Both forms allow local and non-local antecedents. (Pan 1998)

Ta-ziji shows a stronger bias for local antecedents than ziji. (Lyu & Kaiser 2023)

Ta-ziji is more permissive of non-local antecedents than ziji in the absence of a local antecedent. (Dillon et al. 2016)

14

This second question ties into growing body of literature comparing ziji and ta-ziji

Pan (1998) showed from corpus evidence that both reflexive forms are compatible with local and non-local antecedents.

However, experimental work suggests important differences between ziji and ta-ziji.

For instance, Lyu & Kaiser (2023) found that ta-ziji shows a stronger bias for local antecedents than ziji inside complement clauses.

On the other hand, Dillon et al (2016) found that ta-ziji is more permissive of non-local antecedents than ziji in the absence of a potential local antecedent, specifically in so-called sub-commanding configurations.

To the best of our knowledge, no prior work has compared Chinese reflexives to personal pronouns, and no prior work has investigated Chinese pronouns within LPPs.

This project therefore offers an opportunity to add to the empirical outlook on Chinese pronoun forms, as well as the typology of pronoun forms more broadly.

15 of 33

Part 2: �Acceptability Judgment Experiment

15

16 of 33

Stimuli

Stimuli were 16 sentence sets crossing:

event type – motion, perception
relation type – contact, non-contact
pronoun form – simplex reflexive, complex reflexive, personal pronoun

16

	MOTION	PERCEPTION
CONTACT	Lijuan ba liang-pian sa dao ta / ziji / ta-ziji shen shang. Lijuan ba glitter pour to her / self / herself body on ‘Lijuan poured glitter on her(self).’	Lijuan zhuyi-dao liang-pian zai ta / ziji / ta-ziji shen shang. Lijuan notice glitter at her / self / herself body on ‘Lijuan noticed glitter on her(self).’
NON- CONTACT	Lijuan ba liang-pian sa dao ta / ziji / ta-ziji shen pang. Lijuan ba glitter pour to her / self / herself body next ‘Lijuan poured glitter next to her(self).’	Lijuan zhuyi-dao liang-pian zai ta / ziji / ta-ziji shen pang. Lijuan notice glitter at her / self / herself body next ‘Lijuan noticed glitter next to her(self).’

17 of 33

Stimuli

All LPPs include a preposition (dao ‘to’ or zai ‘at’) and a postposition:

17

(8) dao ta / ziji / ta-ziji shen shang

(8) to her / self / herself body on

(8) ‘on her / herself.’

(9) zai ta / ziji / ta-ziji shen pang

(8) at her / self / herself body next

(8) ‘next to her / herself.’

We assume the same underlying structure for English and Chinese LPPs.

(Wu 2015; cf. Svenonius 2007, a.o.)

18 of 33

Stimuli

Most pronouns were followed by a body term (shen ‘body’ or tou ‘head’). The possessive particle de was not used.

18

(8) dao ta / ziji / ta-ziji shen shang

(8) to her / self / herself body on

(8) ‘on her / herself.’

(9) zai ta / ziji / ta-ziji shen pang

(8) at her / self / herself body next

(8) ‘next to her / herself.’

19 of 33

Question format

Survey materials were presented in Chinese characters; translations are included here for clarity.

19

20 of 33

Controls

Controls include the reflexive sentences (10) and anti-reflexive sentences (11).

(10) Context: 朱莉烤了一个红薯。

Context: ‘Zhuli baked a sweet potato.’
( Target: 朱莉把*她/自己/她自己烧伤了。

‘Zhuli burned *her / herself.’

(11) Context: 李萍和她哥哥在外面玩耍。

Context: ‘Liping played with her brother outside.’
( Target: 李萍追他/*自己/*他自己。

) ‘Liping chased him / *himself.’

20

21 of 33

Procedure

74 adult Chinese speakers (71 Mandarin, 2 Cantonese, 1 Wu) were recruited online via Prolific Academic.

Instructions asked participants to compare sentences with the same meaning by choosing whether each is something they would say. Training trials demonstrated that more than one sentence could be endorsed per question.

Each participant saw 16 target questions (4 per condition), 4 control questions, and 20 filler questions, randomly presented.

21

22 of 33

Results: control sentences

There was clear complementarity between reflexives and personal pronouns in control sentences, with no difference between ziji and ta-ziji.

22

Reflexive controls:

Ziji vs. Ta-ziji: β=-0.74, p=0.18
Ziji vs. Ta: β=-5.79, p<0.001

Anti-reflexive controls:

Ziji vs. Ta-ziji: β=-0.35, p=0.56
Ziji vs. Ta: β=5.81, p<0.001

Turning to the results, we will start with the control sentences.

Here and throughout, experimental condition is shown on the X axis while the Y axis shows endorsement rate, or the percentage of ‘yes’ responses.

As predicted, we see very clear complementarity between reflexives and personal pronouns in control sentences, with no difference between ziji and ta-ziji in either control condition.

These results make us confident that the binary choice design worked in drawing out acceptability contrasts between pronoun forms and that participants were comfortable endorsing more than one form per question.

Against this backdrop, let us turn to results for LPP sentences.

[SB: stats above are for interaction Pronoun * Condition. There is also no main effect of Pronoun when looking at reflexives: B = -0.358, p = 0.601. Revise to make more transparent.]

23 of 33

Results: ziji

Ziji had high endorsement rates across conditions, showing no sensitivity to event type or relation type.

Lowest endorsement rate: 87%

Highest endorsement rate: 90%

23

Motion vs. Perception: β=0.25, p=0.30

Contact vs. Non-contact: β=-0.17, p=0.45

24 of 33

Results: ta

Ta had higher endorsement rates in perception sentences than in motion sentences, and in non-contact sentences than in contact sentences.

Lowest endorsement rate: 64%

Highest endorsement rate: 85%

24

Motion vs. Perception: β=-1.29, p<0.001

Contact vs. Non-contact: β=0.44, p<0.01

25 of 33

Results: ta-ziji

Ta-ziji had higher endorsement rates in contact sentences than in non-contact sentences. Difference across event types was not significant.

Lowest endorsement rate: 64%

Highest endorsement rate: 73%

25

Motion vs. Perception: β=0.23, p=0.20

Contact vs. Non-contact: β=-0.51, p<0.01

26 of 33

Part 3: �Takeaways

26

27 of 33

Non-complementarity in Chinese LPPs

All three pronoun forms are generally accepted in Chinese LPPs, regardless of event and relation type.

Conditions A and B are both satisfied, such that reflexives and personal pronouns are both syntactically licensed.

27

28 of 33

Form-specific effects in Chinese LPPs

Event and relation type shape relative preferences, but effects are form-specific.

Ta is sensitive to both factors in the same direction as English her.
Ta-ziji is sensitive to relation type in the same direction as English herself, but not so sensitive to event type.
Ziji is insensitive to both factors.

28

29 of 33

New insights into ziji and ta-ziji

Ta-ziji is more marked than ziji in LPPs, even in the absence of a more local antecedent, and only ta-ziji shows sensitivity to semantic factors.

This suggests that ziji is the default local anaphor in Chinese and is subject only to syntactic constraints, while use of ta-ziji requires additional motivation.

29

30 of 33

Cross-linguistic outlook

The effects of event type and relation type in Chinese are not as strong as in English and affect reflexives and personal pronouns asymmetrically.

30

31 of 33

Cross-linguistic outlook

The effects run in the same direction as in English, suggesting something universal in how event type and relation type can shape LPP pronoun acceptability.

31

32 of 33

Thank you!

This project was supported by the

Rutgers University Center for Cognitive Science.

32

33 of 33

Summary of findings

Effects of event type and relation type on Chinese LPP pronouns show same directionality as in English, though effects are weaker and form-specific
New evidence for typology of pronoun forms within and across languages
Possible universal in how event type and relation type can affect pronouns

33