中文事件計算結構的阻擋效應之實證研究

An Empirical Investigation of the Blocking Effect of Eventuality Counting Expressions in Mandarin

羅佳雯(Chia-wen Lo)

指導教授 ： 麥傑 ； 張寧

國立中正大學/文學院/語言學研究所/碩士(2012年)

摘要

本研究嘗試運用實驗方法去測試事件計算結構(Eventuality counting expression)的阻擋作用(blocking effect) 以及中文裡的殊指性(specificity)。在中文裡，不像其他語言(例如，土耳其語)有明顯的構詞可以清楚區分殊指現象。因此，在中文裡，人們可以很清楚區分殊指(specific)與非殊指(non-specific)以及能夠察覺到事件計算結構與殊指現象的互動在現今研究依舊存有疑問。換言之，即使中文的殊指沒有明顯的構詞能區分，是否接受度(acceptability)可以反映出中文殊指的現象？本研究嘗試測試張(2010)對於中文事件計算結構與殊指互動的主張。藉由測試此主張，我們也可探究文法(特別是指邏輯形式移位LF-movement)與行為的實驗法(例如句子接受度判斷測驗)之間的關係。
本研究的另外一個目標為運用實驗設計方式去決定我們可以遵循何種理論。在判斷句子時，每位語言學家都會有自己的判斷標準。由於判斷句子的標準不同，也會造成他們所提出的主張和解釋也會有所不同。既然理論是建立在句子與資料的收集判斷上，如何得到最有信度的資料與句子判斷也變的相對重要。在這之後，若我們能用理論去預測句子的文法度(grammaticality)，句子的文法度應該也可以符合理論的預測(Sprouse & Almeida 2012)。由於句子判斷的標準不同，湯(1991, 1994)、龔(1993)和蘇(1998)對於事件計算結構及殊指的互動分別提出不同的分析。於是本研究嘗試理解何種分析我們可以遵循。
本研究嘗試了三項句子接受度實驗。實驗一測試事件計算結構位於不同的位置以及運用接續句代表殊指及非殊指進行句子判斷實驗。結果顯示不管是否殊指，事件計算結構若位於右邊的位置，會比另外兩邊的接受度都要高。實驗一不足證明事件計算結構及殊指的互動。
為了排除受試者可能在看到目標句直接作判斷，而沒有把表達殊指或非殊指的接續句考慮進去，實驗二用語境判斷的方式進行問卷實驗。受試者首先會看到有殊指意思的語境或是非殊指意思的語境，然後受試者才會看到目標句。受試者必須評分先看到的語境有幾分可以描述及代表目標句的意思。實驗二結果顯示不論殊指或非殊指，位於右邊的事件計算結構依舊比其他兩個位置的評分高。實驗二顯示了和實驗一相似的結果。由於接受度不等於文法度，實驗二中運用了填充項(filler)來當文法判斷的基準。在配對比較(Tukey’s test)目標句與填充項後，結果顯示位於右邊的事件計算結構與無定賓語(indefinite object)的句子傾向合文法。而位於左邊及中間的事件計算結構與無定賓語的句子傾向不合文法。
由於實驗一及實驗二測試事件計算結構及無定賓語皆顯示相似的結果，實驗三測試是否左邊及中間的事件計算結構本身跟賓語就不相容。實驗三測試事件計算結構與特定賓語(definite object)的組合。實驗三結果和實驗一及實驗二相似，依舊顯示事件計算結構的位置有顯著性差異。實驗三也使用了填充項來當檢視文法度的基準。配對比較目標句及填充項後，結果顯示所有位置和特定賓語的組合皆合文法。實驗結果支持湯(1991, 1994)及蘇(1998)的句子判斷標準。
總言之，由於事件計算結構的三個位置及特定賓語的組合皆合文法，本研究發現事件計算結構不會在表層結構(surface structure)裡阻擋賓語向上移位。然而，事件計算結構的確會阻擋在邏輯形式(logical form)上的量化詞提升(Quantifier raising)以及存在範疇的認證(licensing from existential closure)。除此之外，雖然文法度不能夠被直接檢驗，而接受度也不等於文法度，基準的設置可以運用在實驗中檢驗文法度。特別是在不同的判斷而造成不同的理論分析中，正式實驗可以讓我們更清楚文法如何在理論中被塑造及成型。

關鍵字

接受度判斷 ； 事件計算結構 ； 阻擋效應

並列摘要

This study aims to provide empirical investigations of the blocking effect of eventuality counting expression (ECE) and specificity in Mandarin. This study intends to test Zhang’s (2010) claim about the interaction between the ECE and specificity. It is known that Mandarin does not have overt case markers (cf. Turkish) to represent specificity. It is questionable whether people can notice specificity in Mandarin and be sensitive to the ECE structure related to the interaction of specificity. In other words, can acceptability reflect the grammaticality on the condition that (non-)specificity is distinguished covertly? Can (non-)specificity be reflected in the acceptability judgment in Mandarin? By testing this empirically, we might find out how grammar (LF-movement specifically) relates to behavioral tasks such as acceptability judgments. This thesis found that ECE does not block the object scrambling at PF since all positions of ECE are possible with definite objects. ECE does block indefinites to be licensed by existential closure and Quantifier Raising at LF.
Another goal of this study is to use formal experiment as methods to decide which theory we should follow. Since every linguist has his or her own judgments about the data, the arguments they propose result in different explanations about the data. Since the theory must be built on data, how to get the most reliable data would become important. Then if we use the theory to predict the data, the data should match the prediction of data (see also Sprouse and Almeida 2012). Tang (1991, 1994), Kung (1993) and Soh (1998) give different analyses about the interaction between ECE and specificity due to different judgments about the data. This study tries to figure out which theory would be the direction we should follow.
Three judgment experiments are reported in this thesis. Experiment 1 tested the sentences containing different positions of ECE with a continuation sentence which either yields specific or non-specific reading. The results show that the right position of ECE was rated higher than the other two positions of ECE regardless of specificity. Experiment 1 failed to confirm the interaction between ECE and the notion of specificity.
In order to rule out the possibility that participants had made their judgments without considering the continuation sentence which is either specific or non-specific, an interpretation judgment task was conducted in Experiment 2. Participants saw a context which yields either specific or non-specific reading first. Then they had to rate how much the context can describe the following target sentence containing ECE. The results show that people rated the right position of ECE higher than the other two positions regardless of specificity, which is similar to the results of Experiment 1. In Experiment 2, fillers were controlled as baseline to see (un)acceptability. Comparing the target sentences and fillers, the results show that the sentence of right position of ECE with indefinite object tends to be grammatical. The sentences of left and middle positions of ECE with indefinite object tend to be (un)acceptability.
Since Experiment 1 and 2 show similar results on indefinites, Experiment 3 tested the possibility that the left and middle positions of ECE are not compatible with any following objects. Experiment 3 tested the sentences of ECE with definite object. The results show that there is a main effect on position, which is similar to Experiments 1 and 2. Experiment 3 also used fillers as baseline to examine grammaticality. The results of post-hoc Tukey’s test show that sentences with all positions of ECE with definite object are acceptable. The results support Tang (1991, 1994) and Soh’s (1998) judgments and oppose Kung’s (1993) judgments.
Though grammaticality cannot be examined directly and acceptability is not equal to grammaticality, a baseline can be developed in experiments to examine grammaticality. Especially when judgment variations exist in different theoretical researches such as ECE discussed here, formal experimentation can help us see more clearly about how the grammar shapes in theory.

並列關鍵字

acceptability judgment ； eventuality counting expressions ； blocking effect

參考文獻

Yang, B. C.-Y. (2005). Subject specificity, predicate distributivity, and scope interpretation. Taiwan Journal of Linguistics 3:133-174.

Asadollahfam, H., and A. R. Lotfi. (2010). Quantification scope ambiguity resolution: Evidence from Persian and English. English Language Teaching, 3 (2):180-193.

Baayen, R. H. (2008) Analyzing Linguistic Data. A Practical Introduction to Statistics Using R. Cambridge University Press.

Baayen, R. H. (2011). languageR: Data sets and functions with "Analyzing Linguistic Data: A practical introduction to statistics". R package version 1.4. http://CRAN.R-project.org/package=languageR

Beck, S. (1996). Quantified structures as barriers for LF movement. Natural Language Semantics 4:1-56.