國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
other unaccusative existence/appearance verbs, such as OCCUR or APPEAR, would also be overused in the V-ed form by L2 English learners. On the other hand, even though the research assumed that there might be a close relationship between the grammatical form happened and its Chinese counterpart with the Chinese perfective auxiliary發生了 fāshēngle ‘happen-le’, no empirical acceptability judgment task was conducted to investigate the L1 Chinese transfer to L2 English, which is worth discussing deeply in the present thesis. In Chapter Three as to the corpora analysis, we will re-categorize these five error types so as to clearly divide them into a larger scale of error classifications.
2.5 Gaps of the Previous Research
From the discussion of the previous four sections, we found that many scholars generally centered on two points. The first is the different uses among unaccusative existence/appearance verbs through L2 English-based syntactic perspectives in transitive versus intransitive structures, perfectivity (in L2 English and L1 Chinese), and L1 Chinese lexical semantic analysis via corpora. The second is the error analysis of unaccusative verbs through either empirical psycholinguistic experiments or corpora. However, there are some gaps in previous studies of unaccusative verbs.
First, there seemed to be little research integrating the L2 English syntactic-based perspectives with L1 Chinese grammatical patterns (e.g., 車禍發生 chēhuòfāshēng
‘The car accident happened’ versus 發生車禍 fāshēngchēhuò ‘The car accident happened’) on the lexical semantic perspective, which is paramount because the relationship between L2 English and L1 Chinese could be elicited to explain the causes of frequent error types of unaccusative verbs.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Second, for the research methods regarding the error analysis in the related unaccusative verb studies, generally these two common errors (overpassivization and transitivization errors) were usually examined through either empirical psycholinguistic experiments (e.g., the grammatical or acceptability judgment tasks) or the corpus-based approach. However, fewer studies could be found to integrate the two research methods. According to Gilquin and Gries (2009), the two types of research methods, corpora versus psycholinguistic experiments, have their indispensable advantages. For corpora, the data were extracted from natural linguistic contexts, which would be much more objective in language analysis. As for psycholinguistic experiments, they can be utilized to investigate the less frequent linguistic data in corpora. Additionally, the variables among the linguistic data could be controlled systematically, which would also be one of the criteria to verify corpora analysis. Hence, this research pointed out the importance of integration of both corpora and psycholinguistic experiments to investigate overpassivization and transitivization of unaccusative verbs in SLA. Therefore, an integrated approach to combine corpora analysis with psycholinguistic experiments seems necessary, which can provide more objective evidence for analyzing L2 unaccusative verbs as well.
Third, it is of vital importance to set up a criterion for examining unaccusative verbs between corpora and psycholinguistic experiments and use this criterion to compare the impact of L2 English influence versus L1 Chinese transfer in SLA.
Hence, section 2.5 would stress the grammatical forms as the criterion between corpora and psycholinguistic experiments, along with the language transfer issues discussed in the SLA research.
The grammatical form criterion (e.g., the V-ed form of PLAY is played) could be found in many previous studies (e.g., Krashen, 1977; Lightbown & Spada, 2006;
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Gabriele, Martohardjono, & McClure, 2005). Some studies in SLA regarded the grammatical forms as the grammatical morphemes (Lightbown & Spada, 2006) and claimed that the grammatical forms, such as V-ing or V-s, would be developed by L2 learners through different stages. For instance, Krashen’s (1977) study summarized that the acquisition of V-ing in progressive aspects was found to be earlier developed than that of V-ed form in past or perfective aspects in comparing the accuracy among these grammatical forms, while this development sequence of the grammatical forms may not always follow the same order. Therefore, Lightbown and Spada pointed out three main crucial factors for acquiring the grammatical forms. They are saliency (how easy it is to notice the grammatical forms), linguistic complexity (how many small parts of the grammatical forms L2 learners have to process), and semantic transparency (how clear the meaning of the grammatical forms is). In addition, the similarities and frequency of language uses between L1 and L2 should be also considered when the grammatical forms are investigated in SLA.
Moreover, the grammatical forms are used to investigate verb tense or aspects as well as different verb types. In Housen (2002), Bardovi-Harlig (1999), and Gabriele, Martohardjono, and McClure (2005), they all mentioned the Aspect Hypothesis, in which the grammatical forms would be influenced by the semantic properties of the verb meanings. There are four concise principles of the Aspect Hypothesis given in (11):
(11)
a. Learners firstly associate the imperfective grammatical form V-ing with dynamic/activity verbs, such as work or laugh.
b. The perfective and past grammatical forms (V-ed or V-en) are limited to
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
accomplishment/ achievement verbs, such as drop, stop, die, fall, and happen.
c. The 3rd person grammatical form V-s dominates the stative verbs, such as know and like.
d. In languages with distinction between perfective and imperfective in terms of the perfectivity, imperfective past grammatical forms (e.g., was playing basketball) emerged later than perfective past ones (e.g., have played basketball).
From the four principles, we can discover that the grammatical forms are not distributed randomly among most languages. The combination between a certain grammatical form and a specific verb type may imply some functional or semantic similarities in L2 learners’ mind (11a), particularly for some similar features between L1 and L2 (11d). Owing to the feature possessed by the grammatical forms in English, other research line (e.g., Granger & Rayson, 1998) centered on the application of the grammatical forms via corpora and investigated the the grammatical form distributions of the verbs between native speaker and learner corpora.
In Granger and Rayson (1998), they found that the overall frequency of verbs between native speaker and learner corpora was similar (e.g., for V-ed form, 38% in the native speaker corpus and 35% in learner corpora), while the grammatical forms of verbs would vary. For instance, both the past particples V-ed and the present participles V-ing were underused, whereas the infinitives to-V were easy to be overused by L2 learners. From their finding, the frequency of each grammatical form with verbs may reflect different difficulty of acquiring L2 and this phenomenon would be regarded as the evidence for the varieties of L2 among the learners.
With the data of different frequencies in grammatical forms between the native speaker and learner corpora, it would be more important to find out the association of
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
grammatical forms with L1 language transfer. Though using L2 linguistic data to discuss the influence brought by L1 transfer were noticed by many studies (e.g., Pavlenko & Jarvis, 2002; Odlin, 1989), fewer previous studies would emphasize the relationship between grammatical forms and language transfer.
As for the definition of language transfer, Odlin (1989) provided some classification. The main two divisions of language transfer are positive transfer and negative transfer. Positive transfer refers to the positive effect brought by L1 in acquiring L2. Most of time, the degree of positive transfer would be determined by the similarities between L1 and L2, in which L2 learners may have less difficulty in processing L2. By contrast, negative transfer implies the negative influence and sometimes is also called ‘interference’ because the features of L2 induced greater degree of learning difficulty so that L2 learners would make more errors, which made the data of L2 learners differ from those of native speakers. Furthermore, negative transfer also includes the subcategories of learner errors, such as underproduction (or underuse), referring to the lower frequency of a certain linguistic elements by L2, overproduction (or overuse), indicating that L2 learners simply produce too many linguistic data with the same feature, and misinterpretation, which means that L1 language structures would influence the interpretation of L2.
From the classification of language transfer, one paramount point may emerge.
That is, it appears to be critical for SLA to bridge the gap between the grammatical form of a certain verb type, such as unaccusative existence/appearance verbs, and the language transfer from L1 to L2 learning. Therefore, it would be necessary to combine the corpora analysis with the grammatical forms and then apply the grammatical forms in the psycholinguistic experiments to find out the effect of the L1 language transfer in the present thesis.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
2.6 Summary of the Chapter
In sum, from the overview of the related previous studies in this chapter, the distinction among the different unaccusative existence/appearance verbs should be re-investigated from language users’ perspectives on either L2 English syntactic or L1 Chinese lexical-semantic in order to identify the features of each unaccusative existence/appearance verb within one language or between two languages, such as HAPPEN versus OCCUR or HAPPEN versus發生 fāshēng ‘happen.’ On the other hand, the relationship between the features of individual unaccusative existence/appearance verbs and the causes of L2 learners’ errors in acquiring this verb type should be elaborated more profoundly through an integrated approach with corpora and psycholinguistic experiments. Therefore, in this thesis, the following chapters will center on these two issues. Chapter Three will present Study I—the corpora analysis section.
We will detailedly introduce the way of extracting and collecting data from corpora. Then the results of corpora will also be displayed to compare the differences between unaccusative existence/appearance verbs in Chinese and English, such as HAPPEN versus 發生 fāshēng ‘happen’, as well as discover the frequency and percentages of errors made by L2 learners
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER 3
STUDY I—CORPORA ANALYSIS
From the overview of the related studies in unaccusative verb differentiation based on L2 English syntactic structures, perfectivity, and L2 Chinese lexical semantic in grammatical patterns, along with some SLA research in terms of error analysis of unaccusative verbs and L1 Chinese transfer, we realized the importance of the integrated approach to combine both corpora analysis and psycholinguistic experiments when analyzing unaccusatve verbs. Therefore, in this chapter, Study I will focus on the corpora analysis section. We will first demonstrate the way of extracting data and display the findings of the unaccusative existence/appearance verb HAPPEN and its three other synonyms OCCUR, APPEAR, and EXIST through corpora. The conducting procedures and the findings of psycholinguistic experiments will be later discussed as Study II in Chapter Four.
As for the corpora, there are two main resources—two native speaker corpora (English and Chinese) and three L2 English learner corpora. Via the comparison of both native speaker and L2 English learner corpora, the similarities and differences of the uses of unaccusative existence/appearance verbs will be shown. On the other hand, we can discover the frequency and erroneous rate of each error type for the four unaccusative existence/appearance verbs in L2 English learner corpora, since error types might have something to do with the word frequency chosen by L2 learners.
The next section will first focus on the methods and analysis of Chinese and English native speaker corpora.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
3.1 Methods and Findings of Analyzing Native Speaker Corpora
Regarding the analysis of the native speaker corpora, three main focuses will be emphasized. First, the synonyms between the English unaccusative HAPPEN (with OCCUR, APPEAR, and EXIST) and its Chinese counterpart 發 生 fāshēng
‘happen’ (with 出現 chūxiàn ‘appear’ and 存在 cúnzài ‘exist’) will be compared.
The reason for the synonym comparison between HAPPEN and its Chinese counterpart 發生 fāshēng ‘happen’ is to identify the different tendency with their synonyms so as to elicit the differences in verb meaning between L2 English and L1 Chinese. The result can serve as the references for the data analysis to understand the different uses of unaccusative existence/appearance verbs within the same verb concept (e.g., HAPPEN in English and發生 fāshēng ‘happen’ in Chinese belong to the Happen concept). The investigation of the synonyms for HAPPEN and its Chinese counterpart 發生 fāshēng ‘happen’ is also conducted through both English (BNC) and Chinese (GW 2.0).
The second focus of the corpora analysis is to observe the Chinese grammatical patterns of 發生 fāshēng ‘happen’ with its synonyms (出現 chūxiàn ‘appear’ and 存 在 cúnzài ‘exist’). The purpose to find out the frequently used Chinese grammatical patterns of 發生 fāshēng ‘happen’ with its synonyms (出現 chūxiàn
‘appear’ and存在 cúnzài ‘exist’) through the Chinese native corpus is that we can identify the most representative grammatical patterns in the collocations of Chinese unaccusative existence/appearance verbs used by the native Chinese speakers. With these Chinese grammatical patterns, we could design an empirical acceptability task of L1 Chinese transfer in psycholinguistic experiments, which will be applied and explained in Chapters Four.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Moreover, the third focus is to find out the English grammatical form distributions of HAPPEN, OCCUR, APPEAR, and EXIST, which can be used to compare the similarities and differences between the English native speaker corpus and L2 English learner corpora in the learner corpora analysis section. For the basis of this analysis, we follow the analyzing approach in a pilot study of HAPPEN (Wang &
Chung, 2009; 2010) and some previous studies advocating the relationship between language acquisition and grammatical forms (Bardovi-Harlig, 1999; Housen, 2002).
The result of analyzing grammatical form distributions can help realize how the distributions of the grammatical forms (e.g., happened, happen, happening, and happens) are used by English native speakers.
3.1.1 Native Speaker Corpora
Concerning the native speaker corpora we utilize in this thesis, we adopt an English native speaker corpus British National Corpus (BNC) with approximately 110 million words for L1 English as well as a Chinese native speaker corpus Chinese GigaWord 2 Corpus (GW 2.0) with nearly 455 million words for L1 Mandarin Chinese.
As for the BNC, it was established in 1991 and was completed in 1994. It was collected from a wide range of sources, including written and spoken samples. The written part, with 90% in the BNC, was collected from national newspapers, specialist periodicals, journals, academic books, popular fiction, etc. On the other hand, the spoken part, with 10% in this corpus, contains orthographic transcriptions of informal conversations, government meetings, radio shows, and so forth. For the features of this corpus, the encoding system in accordance with the automatic parts-of-speech
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
taggers, along with other structural properties in texts, such as headings or paragraphs, are provided for the selections of searching.7
On the other hand, regarding the Chinese native speaker corpus, the GW 2.0 was advanced from Chinese GigaWord 1 Corpus (GW 1.0), created by scholars at Academic Sinica, Taipei, Taiwan and collected from 1991 to 2004. Additionally, the GW 2.0 contains an archive of newswire texts from Central News Agency of Taiwan (CNA), Xinhua News Agency of Beijing (XIN), and Zaobao Newspaper of Singapore (ZBN). All of the data in GW 2.0 were tagged with the parts of speech in Chinese, and the accuracy of unknown words was improved, compared to the previous version Chinese GigaWord 1 Corpus (GW 1.0) (Hong & Huang, 2006). We adopted the second version as the Chinese native speaker corpus because the more various sources of Chinese, including China, Taiwan, and Singapore, can be taken into consideration, which can make the grammatical patterns searched in Chinese more objective and representative.8
3.1.2 Analyzing Synonyms of HAPPEN and its Chinese Counterpart 發生 fāshēng ‘happen’
The first analysis via native speaker corpora is to display the synonym comparison for HAPPEN in English and 發生 fāshēng ‘happen’ in Chinese. We would first demonstrate how we extracted the data from both native speaker corpora and show the general findings of the comparions between HAPPEN and 發生 fāshēng ‘happen’, as well as the relationship with their synonyms (OCCUR, APPEAR, and EXIST with HAPPEN; 出現 chūxiàn ‘appear’ and 存在 cúnzài ‘exist’ with 發
7 See http://www.natcorp.ox.ac.uk/.
8 See http://www.ldc.upenn.edu/Catalog/CatalogEntry.jsp?catalogId=LDC2005T14 for more detail.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
生 fāshēng ‘happen’) respectively. Figure 3.1 shows the methods of synonym analysis.
Figure 3.1 The Search Results for HAPPEN in BNC and
發 生 fāshēng ‘happen’ in GW 2.0 Using the Thesaurus FunctionIn Figure 3.1, the thesaurus in BNC and GW 2.0 is utilized to show the relationship of the synonyms with HAPPEN and 發生 fāshēng ‘happen’ respectively.
These two verbs in English and Chinese respectively within the Happen concept have different relationship with their synonyms in terms of the similarity and frequency scores, and therefore the priority order of the synonyms as to HAPPEN in English and
Query word
Similarity Frequency
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
發生 fāshēng ‘happen’ in Chinese would vary as well. In this thesis, we set the English lexicon as the basis to compare the different synonym relationship in Chinese because knowing the different uses between English native speaker and L2 English learners is our main focus of the thesis.
On the other hand, in terms of how to generate the similarity score of each synonym with HAPPEN (e.g., the similarity score for OCCUR with HAPPEN is 0.342.), Figure 3.2 provides the relevant information.
In Figure 3.2, the calculation of the similarity score for OCCUR with HAPPEN is displayed. As can be seen in this figure, all of the common parts of speech in terms of sentence construction, such as subjects, are taken into account to calculate the frequency and similarity score for both HAPPEN and OCCUR. The continuum with
Figure 3.2 Sentence Construction Comparison of HAPPEN and
OCCUR in BNC
happen occur happen occur
Frequency Similarity
Query word
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
one green extreme for HAPPEN and one red extreme for OCCUR represents the tendency of the similar senetence constructions with the two verbs. For instance, the subject death is in the red column, which means that death could be collocated with HAPPEN and OCCUR, yet it is slightly more frequent to be collocated with HAPPEN.
With the finding for the thesaurus in both the BNC and the GW 2.0, the synonyms with similar senetnce constructions can be identified. The most salient three synonyms of HAPPEN are OCCUR, EXIST, and then APPEAR, while the Chinese counterpart (發生 fāshēng ‘happen’) of HAPPEN has different distributions for the synonyms, which indicates that, in terms of lexical meanings, there may be some differences of the Happen concept between English and Chinese, and these differences will be reflected on the synonyms of the two languages.
Specifically, for the results of synonym analysis from the thesaurus of the BNC, the main unaccusative existence/appearance verb HAPPEN possesses 31,245 instances from the native speaker corpus BNC, and the most closely related synonym of HAPPEN is OCCUR with the similarity score of 0.342 and the frequency of 15,477 instances. The second and third related synonyms of HAPPEN are EXIST and APPEAR, even though the frequency of APPEAR (29,956 instances) is over two times more than that of EXIST (11,203 instances), the similarity score of EXIST (0.26) is slightly higher than that of APPEAR (0.244), indicating that, in the BNC, the sentence construction of HAPPEN within per million words is more similar to that of EXIST than APPEAR.
In order to find some similar sentence constructions among these verbs above, the shared subjects selected from corpora are displayed and compared to realize the tendency for the meaning of HAPPEN and發生 fāshēng ‘happen.’ For HAPPEN and OCCUR, the subject thing are frequently combined with each of the two verbs.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
However, HAPPEN and EXIST frequently share the word situation as their subject in a sentence, whereas HAPPEN and APPEAR tend to be combined with the subject attack.
Other than the synonym analysis of the English unaccusative verb HAPPEN
Other than the synonym analysis of the English unaccusative verb HAPPEN