On Chinese Dative Alternation: a Corpus- Based Study on Object - - PowerPoint PPT Presentation

On Chinese Dative Alternation: a Corpus- Based Study on Object Properties

Paper No.: 181 Author(s): Huichen S. Hsiao and Shih-Wei Chan National Taiwan Normal University, Taipei, Taiwan The 21st Chinese Lexical Semantics Workshop (CLSW2020) This project is funded by MOST107-2410-H-003 -025 -MY2

• 1. Introduction
• This paper aims to investigate how the semantic features of direct objects influence

Mandarin dative alternation between double object constructions (DOC) and prepositional dative constructions (PDC).

• Constructions are form-meaning pairings (Goldberg, 1995).
• In Zhang (1999)’s Chinese dative constructions, (a paraphrase conveys)

– DOC : “the agent with purposeful transfers of the patient to the recipient, and this action is successfully completed on site.” – PDC : “an explanation of an action of delivery toward the path.”

• The association between constructions and verbs has been explored via

collostructional analysis (Gries and Stefanowitsh, 2003; Colleman, 2009).

• In this present study, we used collexeme analysis to investigate what the verb

types are in dative alternation (DA) constructions.

– (1) Double Object Construction: [NP1 VERB NP2 NP3] – (2) Prepositional Dative Construction [NP1 VERB NP3 GEI NP2]

• 1. Introduction
• Zhang (1999) also points out the core-extended mechanism for dative alternation.

– The giving objects: spatial entities (e.g. a brick), non-spatial entities (e.g. eye contact) and speech entities (e.g. oral message).

– Typical DOs for DOC: spatial entities/ the extension: other entities

• Liu’s (2006) core/extended meaning of transfer:

– The indirect object’s core semantic role in dative alternation is the recipient. – If the semantic role of an indirect object is a patient or a goal, the dative alternation is less possible. – Liu (2006) doesn’t discuss the core-extended properties of direct objects in dative. alternation

• 1. Introduction
• Thus, we further investigated Liu’s verb classification, as well as

referencing Xu’s (2007, p.257) verb semantics.

• Xu: defines three kinds of English verbs occurring in DOC.

– 1. any verb with a clear giving action, e.g. give. – 2. verbs with an action of giving and manner, e.g. send. – 3. verbs with manner but no giving action, e.g. bake.

• We integrated the concept of “transfer” from Liu (2006) and the

concept of “giving” and “manner” from Xu (2007) to form our semantics features/rules, which allow us to classify dative verbs in this study and see what semantic ranges that the alternate/ non- alternate verbs share.

To sum up

• 1. We conducted collostructional analyses on dative alternation and by calculating

the associations between constructions and verbs.

• 2. We determined whether the properties of direct objects and the properties of

verbs individually influence dative alternation.

• We adopted a corpus-based analysis on COCT 2016 and implemented a sentence

completion task to investigate whether the semantic nature of direct objects affects its occurrence in specific constructions as well as tested cf. Liu’s (2006) classification

• f verb classes in Chinese dative constructions
• Three Research questions:

• 2. Methodology
• We utilized a corpus-based analysis in the COCT 2016 and a

sentence completion task to investigate whether the semantic nature

• f direct objects affects its occurrence in specific constructions.

– COCT 2016: written data collected from 2014 to 2016, comprising 170,119,181 words in total and restricted to only collected data in Taiwan.

1. Collexeme analysis

Verbs Direct objects

2. Analyze the alternate

Xu’s verb semantics Zhang’s DOs

3. Co-varying analysis

The semantic frame in PDC

2.2

Sentence-completion task NSs’ dative preferences

2.1 A corpus-based Analysis

2.1. Corpus Analyses

• While the prepositional dative constructions are marked, it is difficult to extract

double object construction-related data from the corpus.

– We investigate dative alternation observed in the PDC and then manually examine in the COCT 2016 whether DOC occurs.

• 1. Collexeme analysis

Fisher-Exact Test P- value<.05

Verbs (top 25 frequent) – PDC [_V* _N* gei _N*] Direct objects (top 10 frequent)– PDC verbs Representative Direct Objects

• 2. Manually check the

alternate between DOC & PDC

PDC collocation (the representative

• bjects + VERB)

Whether the DOC also exists (alternate) [VERB * DIRECT OBJECT]

• 3. Semantic features

Verbs based on Xu’s classification Direct objects based on Zhang’s

• 4. Co-varying

collexeme analysis PDC’s semantic frame Between two slots [VERB] [DIRECT OBJECT]

u by manually checking whether the most significant verbs in the direct object-to-PDC verb overlap with the verbs which are also significant in the verb-to-PDC collexeme analysis.

2.2. Empirical Study

• An online questionnaire: SurveyCake

– To explore native speakers’ dative preferences

• Using five alternative verbs designated with [+/-concrete] direct
• bjects. The questionnaire was composed of ten items in total with

each item providing subjects with one verb, one direct object, one indirect object and one subject.

– We designed two respective sentences for each verb, which differ in the property of direct objects, e.g. concrete or non-concrete.

• Participants were asked to produce a sentence using the provided

words and optionally complementing sentences with other words if necessary.

• 3. Results

lThe preliminary result indicates that

– (1) our corpus-based findings via collexeme analyses are mostly in line with Liu’s (2006) claims on verb classes and further support Zhang’s (1999) claims based on the prototype-extension theory. – (2) In Mandarin, verbs prefer PDC when occurring with [-concrete] objects. The concreteness of direct object might increase the probability of dative alternation. For instance, the semantic features of “manner” and “giving” for a dative verb determine the possibility of dative alternation.

• Corpus analysis (3.1 & 3.2): verbs prefer PDC when occurring with

[-concrete] objects (e.g. message)

– we infer and argue that direct object concreteness might increase the probability of dative alternation.

• Online task (3.3): native speakers have strong syntactic preferences for PDC

when completing sentences with non-concrete objects.

3.1 Results of Collexeme Analyses

• 22 verbs in PDC (Table1) and 9 direct objects in

PDC are found significant through collexeme analyses.

• Based on Liu’s (2006, p.870) verb lists, we

identified 17 verbs that can be grouped into two lists.

– 1. Alternate between PDC and DOC. – 2. Only PDC allowed. – The remaining 5 verbs are not listed in Liu’s list à grouped into 2 lists based on Liu’s classification.

– 1. Alternate 2. Only PDC allowed

3.1 Results of Collexeme Analyses

• We chose the alternate verbs of the 22 significant PDC verbs and chose
• ne verb from each verb class as a candidate to analyze their verb

semantics. – based on Liu’s (2006) verb “transfer” property (causing a theme to be successfully transferred) and Xu’s “giving” (the action of giving), “manner” (apparent manner of giving) properties. – we included two verbs that do not entirely correspond to Liu’s grouping in terms of dative alternation, fu ‘pay’ and ci ‘bestow’ in

• rder to serve as testing materials in the sentence-completion task.

3.1 Results of Collexeme Analyses

• It is inferred that the lexical property of

“manner” and “giving” determine the possibility of dative alternation.

• In particular, we checked alternation in

Liu’s verb lists, where song ‘give’, tigong ‘provide’, rang ‘yield’, jieshao ‘introduce’ are able to alternate between DOC and PDC. If the verb is embodied with the property of “giving”, it is often correlated with the lexical property of “transfer”.

• Other verbs are only allowed to occur PDC

because the lexical verbs lack the meaning of “giving” and some of them only entails the property of “manner”, causing them to be mainly compatible with the construction meaning of PDC.

Some cases that are different from Liu’s theory

• The verb fu ‘pay’ belongs to

“transfer of possession” based on Liu’s classes, which means this verb can alternate between DOC and PDC.

– Cannot alternate if the direct object is not a concrete entity

• The verb ci ‘bestow’ belongs to

Liu’s “promise” type and thus is not allowed to be DOC.

– Can alternate in COCT2016 with a concrete direct object

Find the alternated patterns

• Table 3 shows the collocated pattern in PDC in order to address whether the

direct object properties influence dative alternation.

– We manually check in the COCT 2016 whether the same collocated patterns also

• ccur in DOC by using the search pattern [VP * NP] as the DOC.

– song ‘give’, tigong ‘provide’, mai ‘sell’, jie ‘lend’ are found to be alternate among these nouns. (Table 4)

For the concrete direct objects, dongxi ‘stuff’, shu ‘book’, qian ‘money’ and liwu ‘gift’, the individual frequency of PDC is higher than that of DOC. In general, the frequency of PDC for each direct object is also higher than DOC.

• We found that the alternate verbs for the concrete direct objects, such as dongxi

‘stuff’, shu ‘book’, qian ‘money’, liwu ‘gift’, are song ‘give’, mai ‘sell’ and jie ‘lend’. (cf. Table 4)

• We further collected statistics of the frequency for each alternate pattern (Table 5).

###The reason why PDC dominates DOC is that our direct objects are primarily searched from PDC, which logically leads to more PDCs than DOCs. Although the number of DOC is less than PDC, we can still pay attention to whether the verb is able to alternate or not.

• Further results found in Tables 4 & 5 are regarding the non-concrete

direct object, xunxi ‘message’ only having association with the verb tigong ‘provide’ .

• We thus further investigated more non-concrete direct objects that

are collocated with tigong ‘provide’ by analyzing its frequency in the COCT 2016.

– The other possible alternate verb with xunxi ‘message’ is song ‘give’, but we don’t find this correspondence in the COCT 2016. – Both tigong ‘provide’ and song ‘give’ belong to the type of transfer of giving, but only tigong ‘provide’ is possibly collocated with the non-concrete direct object.

• Even though tigong ‘provide’ is alternate with DOC and PDC when

the direct object is non-concrete, the percentage of PDC (61%) is higher than DOC (39%).

3.2 Results of Co-varying Collexeme Analysis

• Non-concrete direct objects, such as xunxi ‘message’, hua ‘oral speaking’ and jianxun

‘text’ xunxi ‘message’, are highly collocated with verbs with the [-giving, +manner] property, e.g. chuan ‘send’, chuandi ‘deliver’.

• As for concrete direct objects, they are highly collocated with verbs [+transfer] or [-

transfer].

– only song ‘give’ and jie ‘lend’ can alternate. – The verb properties of song ‘give’ and jie ‘lend’ are [+transfer, +giving, -manner], making dative alternation possible.

• Thus, we infer that giving and

manner determine the possibility

• f dative alternation.

3.3 Results of sentence-completion task

• Among the 25 subjects, we collected a total of 250

sentences, but in this section we only show the PDCs and DOCs and show the distribution based on 5 different verbs and 10 different direct objects, where the direct

• bjects

are divided by the property [+concreteness] or [-concreteness].

• Some verbs can address our issue that whether direct objects

can influence subjects’ syntactic production, especially tigong ‘provide’, rang ‘yield’, and fu(chu) ‘pay’.

– rang ‘yield’ and fu(chu) ‘pay’: subjects produced no DOCs when given non-concrete objects. – Although tigong ‘provide’ has been investigated to prefer DOC (Hsiao & Lestari, 2020), it is found to prefer PDC than DOC (20:1) when collocated with a non-concrete object xunxi ‘message’.

• In terms of song ‘give’, the reason why this verb produced

more DOCs with non-concrete object may be related to verb bias (Hsiao & Lestari, 2020) and the construction meaning (Zhang, 1999).

• The verb ci ‘bestow’, which is supposed to be ‘promise’ type

(Liu, 2006) and not to allow DOCs, is found to alternate in this task.

• Some verbs are not highly associated with dative constructions,

especially with non-concrete direct objects. – rang ‘yield’, fu ‘pay’ and ci ‘bestow’, subjects produced more dative constructions with concrete direct objects than non- concrete direct objects.

• It seems that the concrete direct objects trigger

subjects to produce constructions related to PDCs or DOCs and serve as a core role in direct objects.

• 4. Conclusion
• Incorporating corpus studies and empirical study, we have shown that the direct object

properties have an influence on the possibility of dative alternation and whether the semantic features of verbs entails the core meaning [+giving] determines their alternation between PDC and DOC.

• The corpus study indicates that there’s an interaction between the direct objects and verbs,

which both entail prototypical-extended relationships.

– The prototypical semantics of an alternate dative verb is [+giving]. – For direct objects, it is [+concreteness] that serves as the most prototypical property for a dative verb to alternate between PDC and DOC.

– correspond to Liu’s core-extension verb types and Xu’s classification on direct objects.

• The experiment indicates that the direct object properties influence participants to produce

dative constructions. When the given items contain a [-concreteness] object, the possibility of producing a PDC is higher than that of DOC.

• We further investigated the semantic frame to which PDC belongs to by conducting a co-

varying collexeme analysis. The verb slot of PDC belongs to “manner of motion” verbs, whose lexical property is [+manner]. These verbs are significantly collocated with “message” direct

• bjects, which are [-concreteness]

References

• 1. Arnold, J. E., Losongco, A., Wasow, T., & Ginstrom, R.: Heaviness vs. newness: The effects of structural complexity and discourse status on constituent ordering. Language,

76(1), (2000) 28–55

• 2. Colleman, T.: The semantic range of the Dutch double object construction: A collostructional perspective. Constructions and Frames, 1(2), (2009) 190–220
• 3. Goldberg, A. E.: Constructions: A construction grammar approach to argument structure. University of Chicago Press (1995)
• 4. Gries, S. T., & Stefanowitsch, A.: Co-varying collexemes in the into-causative. Language, culture, and mind, (2004) 225–¬¬¬236.
• 5. Gries, S. T., Hampe, B., & Schönefeld, D.: Converging evidence II: More on the association of verbs and constructions. Empirical and experimental methods in

cognitive/functional research, (2010) 59–72

• 6. Gries, S. T., Hampe, B., & Schönefeld, D.: Converging evidence: Bringing together experimental and corpus data on the association of verbs and constructions. Cognitive

Linguistics, 16(4), (2005) 635–676. Retrieved 15 Jan. 2020, from doi:10.1515/cogl.2005.16.4.635

• 7. Hsiao, Huichen S. & Lestari Mahastuti: A Collostructional Analysis of Ditransitive Constructions in Mandarin. Chinese Lexical Semantics (20th Workshop, CLSW 2019,

Revised Selected Papers), (2020) 37–51. Semantics Lecture Notes in Artificial Intelligence 11831, Springer, 2020. SCOPUS (MOST 107-2410-H-003-025-MY2)

• 8. Hu, J., Zhang, J.Y.: Collostructional Analysis: An Integration of Cognitive Linguistics and Corpus Linguistics. Journal of Foreign Languages. 35(4), (2012)
• 9. Huang, C. R., Chang, L. P., Ahrens, K., & Chen, C. R.: The Interaction of Lexical Semantics and Constructional Meanings. Chinese Language and Linguistics V: Interactions

in Language, (1999) 413–438

• 10. Li, Charles Na & Sandra. A. Thompson.: Mandarin Chinese: A functional reference grammar. Berkeley: University of California Press. (1989)
• 11. Liu, F. H.: Dative constructions in Chinese. Language and Linguistics, 7(4), (2006) 863–904.
• 12. Lu, J.M.: Xiandai Hanyu Yufa Yanjiu Jiaocheng<Lit. Tutorial of Mandarin Chinese Grammar>. Peking University Press (2003) [

2003]

• Shen J.X.: ‘Zai’ ziju han ‘gei’ ziju<Lit. existence-sentence and dative constructions>. China Academic Journal Electronic Publishing House, (2) (1999) 94–102 [1999

294–102]

• 13. Stefanowitsch, A. & Gries, S. T. Collostructions: Investigating the interaction of words and constructions. International Journal of Corpus Linguistics, 8. (2003) 209–243
• 14. Stefanowitsch, A., & Gries, S. T.: Covarying collexemes. Corpus linguistics and linguistic theory, 1(1), (2005) 1-43
• 15. Xu, S.H.: Ditransitive Construction Revisited from the Perspective of Proximity. Foreign Language Teaching and Research, 39(4), (2007) 253–260
• 16. Zhang, Bojiang: Xiàndài Hànyǔ de Shuāngjíwù Jiégòushì <Lit. Dative constructions in Mandarin Chinese>. China Academic Journal Electronic Publishing House, (3),

199916–25 [(3)1999]

THANK YOU!