3. STUDY I—CORPORA ANALYSIS…
3.3 Findings of Learner Corpora Analysis
3.3.1 Findings of Grammatical Form and Erroneous Rate Analysis in Learner
國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
would be also analyzed and discussed with particular concern in section 3.4.2.
3.3 Findings of Learner Corpora Analysis
From the analysis in section 3.3, we have discovered the distributions of the grammatical forms of the four unaccusative verbs in the English native speaker corpus.
In section 3.3.1, the comparison between native speaker corpus and the other three learner corpora will be stressed.
3.3.1 Findings of Grammatical Form and Erroneous Rate Analysis in Learner Corpora
This section is comparing the similarities and differences among the native speaker corpus (BNC) and the three learner corpora (the LTTC, the ICLE, and the NCCU). In order to clearly show the features of the four verbs among the four corpora, we utilized the bar chart to present the percentages of HAPPEN, OCCUR, APPEAR, and EXIST. The result is shown in the following four figures.
Figure 3.7 BNC Frequency of the Four Verbs
Figure 3.8 LTTC Frequency of
the Four Verbs
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
Regarding the grammatical form distributions, the four figures (from Figure 3.7 to 3.10) present the percentages of the four frequent grammatical forms (V-ed, V-base, V-ing, and V-s) possessed by the four unaccusative verbs among the four corpora. As mentioned previously, the result shows that either V-ed or the V-base form of the four unaccusative verbs in BNC appears most frequently in the BNC even though only EXIST appears extremely frequent as the base form exist. This corresponds to a similar distribution of the other three learner corpora that the highly frequent grammatical forms almost appear in both V-ed and V-base forms. However, some grammatical forms distributions, such as existing (35.86%) in the ICLE corpus and occurs (32.14%) in the NCCU corpus possess higher percentages than those in BNC.
Particularly for existing (35.86%) in the ICLE, this grammatical form shows a great difference with that of BNC with only 2.05% of occurrences, which is viewed as the striking variance between native speaker and learner corpora.
After the general analysis of the grammatical form in the four corpora, we then respectively analyze the frequency of the four verbs as well as their erroneous rates from each learner corpus and each verb. In the LTTC learner corpus, compared with the other three verbs, HAPPEN displays a similar pattern of the discrepancy between the V-ed form (45.16% for happened) and the V-base form (37.09% for happen), and
Figure 3.9 ICLE Frequency of the Four Verbs
Figure 3.10 NCCU Frequency
of the Four Verbs
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
the erroneous rate of HAPPEN is proportional to the grammatical form. That is, the erroneous rate of the happened form (71.42% with 28 instances≒32.25% of all HAPPEN’s instances) appears the most, and then the happen form (52.17% with 23 instances≒19.35% of all HAPPEN’s instances) appears the second most in the LTTC corpus. On the other hand, the overall erroneous rate of HAPPEN in the LTTC corpus is more than 50% (53.22% from 62 instances), which indicates that the L2 learners may have some difficulty acquiring the uses of the specific grammatical forms for the unaccusative existence/appearance verb HAPPEN.
As for the other three synonyms of HAPPEN in the LTTC corpus, OCCUR has a dominant percentage of V-ed form (75% for occurred) with the most erroneous rate (66.66 % with 3 instances≒50.00% of all OCCUR’s instances) even though the frequency of OCCUR (4 instances) is much lower than that of HAPPEN (62 instances). APPEAR also presents its distributions on the V-ed form (40.90% for appeared) and the V-base form (36.36% for appear), while the two highest erroneous rates are on the base form appear (50.00% from 36.36%≒18.18% of all APPEAR’s instances) and the V-s form appears (66.66% with 3 instances≒9.09% of all APPEAR’s instances).
The last synonym EXIST shows its salient percentage of the grammatical forms on the V-base form exist (80% for exist) with the most erroneous rate (75% from 80%
≒60% of all APPEAR’s instances), which may probably indicate that EXIST tends to be used in the V-base form exist, whereas the overuse of a certain grammatical form would cause more errors. The next figure will display the verb form distributions of the four unaccusative verbs in the ICLE corpus.
As for the data from the ICLE corpus, HAPPEN also possesses the most percentages in terms of the grammatical forms on the V-ed form happened (35.20%)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
and the V-base form happen (36.00%) with the greater erroneous rate (27.27% from 35.2%≒9.6% of all HAPPEN’s instances for happened and 33.33% from 36%≒12%
of all HAPPEN’s instances for happen respectively). The second verb OCCUR shows the most frequent grammatical form on the V-ed form (36.11% for occurred) as well as the V-base form (52.77% for occur). However, the two highest erroneous rates are displayed on the highly frequent V-ed form occurred (76.92% from 36.11%≒27.78%
of all OCCUR’s instances) and the low frequent V-s form occurs (50% from 5.55%≒
2.78% of all OCCUR’s instances).
The third verb APPEAR in the ICLE corpus has its higher frequent grammatical forms on the base form (59.61% for appear) and V-ed form (26.92% for appeared), while the two highest erroneous rates are on the V-ed form appeared (28.57% from 26.92%≒7.69% of all APPEAR’s instances) and the low frequent V-s form appears (14.28% from 13.4%≒1.91% of all APPEAR’s instances). The same pattern in terms of the two highest erroneous rates can be found in EXIST, with 66.66% from 13% (≒
8.67% of all EXIST’s instances) for existed and 63.63% from 11.9% (≒7.57% of all EXIST’s instances) for exists, whereas the most frequent grammatical forms are the V-base form (39.13% for exist) and the V-ing form (35.86% for existing), which is the most distinctive finding in the ICLE corpus.
In a nut shell, as for the data from the ICLE corpus, some findings can be summarized. Different from its three synonyms, HAPPEN possesses two most frequent grammatical forms possessing the most erroneous rate (happened and happen). By contrast, EXIST displays separate distributions on highly frequent grammatical forms (exist and existing) and erroneous grammatical forms (existed and exists). On the other hand, OCCUR and APPEAR present a similar pattern on the two highest erroneous rates, including one of the most frequent grammatical forms of the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
three verbs (occurred and occurs for OCCUR, appeared and appears for APPEAR).
Regarding the third learner corpus NCCU, HAPPEN, similar to the previous two corpora, possesses the two most frequent grammatical forms on the base form and the V-ed form (40.84% for happen and 36.61% for happened), with higher erroneous rates (15.51% from 40.8%≒3.14% of all HAPPEN’s instances for happen and 13.72% from 36.61%≒5.02% of all HAPPEN’s instances for happened). On the other hand, OCCUR presents similar distributions in terms of the most highly frequent grammatical forms with the two highest erroneous rates on the V-base form occur and the V-s form occurs (16.66% from 46.42%≒7.14% of all OCCUR’s instances for occur and 22.22% from 32.14%≒7.14% of all OCCUR’s instances for occurs). The rest of the two verbs (APPEAR and EXIST) have a similar tendency on only one grammatical form (the V-base form) with errors (16.00% from 58.13% for appear≒9.3% of all APPEAR’s instances and 9.52% from 61.76%≒58.8% of all EXIST’s instances for exist), which indicates that the other three grammatical forms (V-ed form, the V-ing form, and the V-s form) of APPEAR and EXIST suggest less difficulty in the NCCU corpus.
To summarize the findings of the grammatical forms as well as erroneous rate across the four corpora, we can look at Table 3.3. In this table, we chose two top grammatical forms in each corpus. The grammatical forms in bold-face type refer to the grammatical forms also found in BNC. Additionally, the grammatical forms with underlines refer to the grammatical forms possessing not only higher frequencies but higher erroneous rates as well.
‧
TABLE 3.3 Two Top Grammatical Forms with High Frequency and Errors
Corpora Two top highly frequent grammatical forms
Two top erroneous grammatical forms HAPPEN
(66)
BNC happened (41.96%) happen (27.11%)
LTTC happened (45.16%) happen (37.09%) happened (32.25%)happen (19.35%)
ICLE happen (36.00%) happened (35.20%) happen (9.6%) happened (12%)
NCCU happen (40.84%) happened (36.61%) happen (3.14%)happened (5.02%)
OCCUR (43)
BNC occur (35.78%) occurred (34.63%)
LTTC occurred (75%) occurs (25%) occurred (50.00%)
ICLE occurred (36.11%) occur (52.77%) occurred (27.78%) occurs (2.78%)
NCCU occur (46.42%) occurs (32.14%) occur (7.14%) occurs (7.14%)
APPEAR (53)
BNC appear (35.83%) appeared (34%)
LTTC appeared (40.90%) appear (36.36%) appear (18.18%) appears (9.09%)
ICLE appear (59.61%) appeared (26.92%) appeared (7.69%) appears (1.91%)
NCCU appear (58.13%) appears (20.93%) appear (9.3%)
EXIST (42)
BNC exist (47.76%) exists (28.16%)
LTTC exist (80%) existed (20%) exist (60%)
ICLE exist (39.13%) existing (35.86%) existed (8.67%) exists (7.57%)
NCCU exist (61.76%) exists (17.64%) exist (58.8%)
In Table 3.3, the two top grammatical forms with the highest frequencies and erroneous rates can be found and compared across the corpora. When comparing the grammatical forms, we can discover that the highly frequent grammatical forms in BNC bear a close resemblance to those in the three learner corpora. Among those highly frequent grammatical forms, more than half of them are highly erroneous.
Therefore, the overuse of the L2 English verb forms can be observed from this section in corpora comparison. The next section 3.3.2 will focus on the types of errors for the four unaccusative existence/appearance verbs.
3.3.2 Findings of Categorizing the Errors
The second part of the learner corpora findings to show the result of the distributions of the error types among the four verbs based on Table 3.2. Each of the four verbs (HAPPEN, OCCUR, APPEAR, and EXIST) is displayed according to the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
frequency and the percentages of the five error types within the schematic errors and the unaccusative errors. The schematic errors include Type 1 (mismatches in subject-verb agreement or tense marker, e.g., *Why the 現 象 xiànxiàng
‘phenomenon’ happened?), Type 2 (mismatches in infinitive usages, e.g., *But you may say what is the reason cause this happen?), and Type 3 (mismatches in present participle usages, e.g., *To avoid this thing happen, we should always keep clearly in a good range.), while the unaccusative errors contain Type 4 (overpassivization, e.g.,
*First problem is always happened. When you eat noddles you will find glass bluring.), and Type 5 (transitivization, e.g., *This situation I have never happened before!).
As for the grouping of the four verbs, since HAPPEN and OCCUR have closely related meaning, the two verbs are discussed as one group in Table 3.4. The other two unaccusative existence/appearance verbs (APPEAR and EXIST) shown in Table 3.5 are sorted as the other group. We first select two most frequent error types in each corpus and then observe the common error types across the three corpora.
TABLE 3.4 Frequency of Error Types in HAPPEN and OCCUR
HAPPENError type LTTC ICLE NCCU
Schematic errors
Total (Schematic errors) 28 (84.84%) 18 (62.06%) 7 (41.17%) Type 1 (S-V agre.) 15 (45.45%) 13 (44.82%) 2 (11.76%) Type 2 (Infinitive) 8 (24.24%) 1 (3.44%) 1(5.88%) Type 3 (Pres. Part.) 5 (15.15%) 4 (13.79%) 4 (23.52%)
Unaccusative errors
Total (Unaccusdative errors) 5 (15.15%) 11 (37.93%) 5 (29.41%) Type 4 (Overpassivization) 4(12.12%) 10 (30.30%) 4 (23.52%)
Type 5 (Transitivization) 1(3.03%) 1(3.44%) 1 (5.88%)
Others 5 (29.41%)
Grand total 33 (100%) 29 (100%) 17 (100%)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
OCCUR
Error type LTTC ICLE NCCU
Schematic errors
Total (Schematic errors) 0 (0%) 4 (33.33%) 3 (60%)
Type 1 (S-V agre.) 2 (16.66%) 2 (40.00%)
Type 2 (Infinitive)
Type 3 (Pres. Part.) 2(16.66%) 1 (20.00%)
Unaccusative errors
Total (Unaccusdative errors) 4 (100%) 8 (66.66%) 2 (40%)
Type 4 (Overpassivization) 8 (66.66%) 1 (20.00%)
Type 5 (Transitivization) 2 (100%)
Others 1 (20.00%)
Grand total 2 (100%) 12 (100%) 5 (100%)
From Table 3.4, it shows the distributions of the two large scales in terms of HAPPEN and OCCUR. As for HAPPEN, higher percentages of schematic errors (84.84% in the LTTC; 62.06% in the ICLE; 41.17% in the NCCU) indicate that the highly frequent unaccusative existence/appearance verb might be easily misused by L2 learners in general error types. On the other hand, with respect to the unaccusative errors of HAPPEN, the percentages of the overpassivization errors (12.12% in the LTTC; 30.30% in the ICLE; 23.52%) are comparatively higher than those of the transitivization errors (3.03% in the LTTC; 3.44% in the ICLE; 5.88% in the NCCU) across the three learner corpora, revealing that generally L2 English learners tend to make the overpassivization errors with HAPPEN.
As for the cross-corpora comparison of OCCUR, we found that the percentages of the schematic errors (0% in the LTTC; 33.33% in the ICLE) are generally lower than those of the unaccusative errors (100% in the LTTC; 66.66% in the ICLE), except for the NCCU Learner Corpus (60% for the schematic errors versus 40% for the unaccusative errors), which means that OCCUR might easily be misused by L2 learners in the unaccusative errors. Within the unaccusative errors, we found that the overpassivization errors with higher percentages in the two learner corpora (66.66%
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
in the ICLE; 20% in the NCCU) are generally the most frequent errors of OCCUR, even though the percentage of the transitivization errors (100%) in the LTTC dominates the whole error types, which may be partially due to the low frequency of the errors.
On the other hand, compared with HAPPEN, the other verb OCCUR has lower frequencies among the first three schematic error types (Type 1, 2 and 3), except for the outstanding frequency of Type 1 in the NCCU, which may indicate that L2 English learners of the LTTC and ICLE corpora seem to make less schematic errors, such as Type 1 (*Many family problems will occurs.), Type 2 (*Televisions makes that incident occurred.), and Type 3 (*There is a different opinion occurs.). However, this may be also because of the asymmetric frequencies between the two verbs. That is, the error frequencies of HAPPEN is usually approximately three times more than those of OCCUR, while this also indicates that L2 English learners tend to choose HAPPEN rather than OCCUR to illustrate the verb concept of the event becoming to exist. From the findings, we realized that, for HAPPEN and OCCUR, Type 4 (Overpassivizaton) of the unaccusative errors is the identically frequent error type across the three learner corpora.
Furthermore, to know how L2 learners misused the sentences with HAPPEN and OCCUR and to realize some features possessed in the overpassivization errors, we select some instances from the three learner corpora for further qualitative analysis.
Four examples are shown in (1).
(1) a. *When two reasons above are happened frequently, students will get nearsighted soon. (LTTC)
b. *Few crimes will be happened. (ICLE)
c. *The same condition is occurred ion students, too, even more apparently. (NCCU) d. *In recents, many PC cafes are occurred in the city. (ICLE)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
As shown in sentences (1), there are four overpassivization errors of HAPPEN or OCCUR extracted from the learner corpora. For the subjects of the sentences, L2 learners tend to combine some improper subjects, such as two reasons in (1a), or PC café in (1d), with HAPPEN or OCCUR, which are not usually used by English native speakers. Though we might not be sure whether the selection of subjects will increase the overpassivization errors, yet the misuses of the unaccusative errors of HAPPEN and OCCUR would probably not only be related to the English syntactic structures.
The semantic lexical choices of the subjects or, perhaps, the effect brought by L1 Chinese might also be taken into account, which will be discussed in the psycholinguistic experiments in Chapter Four.
After realizing the frequencies and distributions of the five error types within schematic and unaccusative error scales for the first verb group (HAPPEN and OCCUR), we then display the result of the second verb group (APPEAR and EXIST) in Table 3.5.
TABLE 3.5 Frequency of Error Types in APPEAR and EXIST
APPEARError type LTTC ICLE NCCU
Schematic errors
Total (Schematic errors) 3 (42.85%) 1 (11.11%) 0 (0%) Type 1 (S-V agre.) 1 (14.28%)
Type 2 (Infinitive)
Type 3 (Pres. Part.) 2 (28.57%) 1 (11.11%) Unaccusative errors
Total (Unaccusative errors) 4 (57.14%) 8 (88.88%) 4 (100%) Type 4 (Overpassivization) 2 (28.57%) 5 (55.55%)
Type 5 (Transitivization) 1 (14.28%) 2 (22.22%) 4 (100%) Others 1 (14.28%) 1 (11.11%)
Grand total 7 (100%) 9 (100%) 4 (100%)
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
EXIST
Error type LTTC ICLE NCCU
Schematic errors
Total (Schematic errors) 1 (33.33%) 5 (22.72%) 2 (100%)
Type 1 (S-V agre.) 3 (13.63%) 1 (50%)
Type 2 (Infinitive) 1 (50%)
Type 3 (Pres. Part.) 1 (33.33%) 2 (9.09%) Unaccusative errors
Total (Unaccusative errors) 2 (66.66%) 15 (68.18%) 0 (0%) Type 4 (Overpassivization) 12 (54.54%)
Type 5 (Transitivization) 2 (66.66%) 3 (13.63%)
Others 2 (9.09%)
Grand total 3 (100%) 22 (100%) 2 (100%)
In terms of the other two verbs (APPEAR and EXIST), Table 3.5 shows the result of error distributions. As for APPEAR, the percentages of the schematic errors (42.85% in the LTTC; 11.11% in the ICLE; 0% in the NCCU) are comparatively higher than those of the unaccusative errors (57.14% in the LTTC; 88.88% in the ICLE; 100% in the NCCU), indicating that L2 learners tend to make unaccusative errors in the sentences with APPEAR. A further analysis of the unaccusative errors of APPEAR, the data possess an overlapping area on the two unaccusative errors—Type 4 (overpassivization) and Type 5 (transitivization)— across the three learner corpora when it comes to the two highly frequent errors (28.57% for Type 3 (schematic errors) and Type 4 (unaccusative errors) in the LTTC; 55.55% for Type 4 (unaccusative errors) and 22.22% for Type 5 (unaccusative errors) in the ICLE; 100% for Type 5 (unaccusative errors) in the NCCU), which implies that the two unaccusative error types may be the major or typical errors of APPEAR for L2 English learners.
On the other hand, in terms of the common frequent error types, EXIST has similar distributions in the transitivization errors (66.66% in the LTTC; 13.63% in the
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
ICLE), even though the overpassivization errors in the ICLE, accounting for 54.54%
as the most highly frequent error. Thus, for APPEAR and EXIST, the common frequent errors across corpora are the transitivization errors (Type 5) within the unaccusative errors even though Type 4 (overpassivization) and Type 1 (subject-verb agreement) still have some impact on acquiring the two unaccusative verbs by L2 learners.
Furthermore, to display some particular features of the transitivization errors of APPEAR and EXIST, four erroneous instances are selected in (2) for discussion to show some specific L2 English differences from those of the native speakers.
(2)
a. *the tradition that can appear the culture there and can appeal many teenagers. (LTTC)
b. *It is not only appears the financial problems. (ICLE) c. *Although recycling of waste exists a few problems. (ICLE) d. *The cabbages exist a natural and special flave. (LTTC)
From sentences in (2), it appears that some errors cannot be completely placed in the transitivization errors. For example, in (2b), it reveals a combination of both overpassivization (*It is not only appears…) and transitivization errors (*appears the financial problems) in this case from the ICLE learner corpus. Another problem observed from these instances is that the inconsistency between the grammatical form choices of the verbs and nouns. For instance, in (2a), the grammatical form appear has no relation with either the subject the trandition or the erroneous object the culture. Lastly, the noun problem seems to be frequently used with both APPEAR (2b) and EXIST (2c) by L2 learners, and the particular L2 English patterns, such as
*appear the culture in (2a) and *exist a natural and special flave (2d), may not be
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
used frequently by English native speakers. Thus, all of the four transitivization erroneous instances of APPEAR and EXIST reveal the difficulty for L2 English learners to completely acquire the two unaccusative existence/appearance verbs.
However, the possible reasons for these transitivization errors might not be easily elicited via the only corpora comparison, which inspired us to incorporate the psycholinguistic experiments to examine to find out the cause of the unaccusative errors.
3.4 Summary of the Chapter
In a nutshell, in this chapter, several findings from corpora analysis can be included. First, for synonym analysis, from the findings of native speaker corpora, HAPPEN is closer to the English EXIST concept, while發生 fāshēng ‘happen’ is more related to the Chinese出現 chūxiàn ‘appear’ concept. Second, from the Chinese grammatical patterns found in the Chinese native speaker corpus, 發生了 fāshēng-le
‘happen-perfective auxiliary’, 出現了 chūxiàn-le ‘appear- perfective auxiliary’, and 存在著 cúnzà-zhe ‘exist-imperfective auxiliary’ are found to be frequently used by Chinese native speakers. Additionally, both V+N and N+V patterns for 發生 fāshēng
‘happen’, 出 現 chūxiàn ‘appear’, and 存 在 cúnzà ‘exist’, such as 發 生 意 外 fāshēngyìwài ‘The accident happened’ or 意外發生 yìwàifāshēng ‘The accident happened’, are frequently used in Chinese. Third, in English native speaker corpus, we found that V-ed or V-base for the four unaccusative verbs are frequent, which is similar to the findings from learner corpora. The highly frequent grammatical forms, influenced by English, are usually overused by L2 learners, since the erroneous rates of those grammatical forms are comparatively higher. Fourth, for the error types of
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
the four verbs, HAPPEN and OCCUR are frequently misused in the unaccusative error—Type 4 (overpassivization), while APPEAR and EXIST are similarly misused in the unaccusative error—Type 5 (transitivization).
Despite the findings of L2 learners’ frequent errors of HAPPEN and its synonyms through corpora comparison, there is still a doubt how these errors are caused. In order to test the possible reasons for L2 errors of the four unaccusative existence/appearance verbs, we will conduct psycholinguistic experiments based on the variables of L2 learners (e.g., their age differences) and L1 Chinese transfer.
In Chapter Four, we will demonstrate the methods of how we designed and did the psycholinguistic experiments as well as some detailed discussion related to the experiments. In addition, the results of statistical evidence for the psycholinguistic experiments will be shown as well.
‧ 國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
CHAPTER 4
STUDY II— PSYCHOLINGUISTIC EXPERIMENTS
Chapters Three examines the results from corpora analysis, displaying the overuse of the highly frequent grammatical forms in which L2 learners usually make more errors of HAPPEN and its three synonyms. However, due to the limits of the corpora analysis that the variables within L2 learners (e.g., age differences) and the L1 Chinese transfer or effect (e.g., a particular word order意外發生 yiwaifāshēng
‘The accident happened’) cannot be tested, we then conducted psycholinguistic experiments to re-examine the common errors with specific frequent grammatical forms in the corpora section.
In Study II of this thesis, we intend to investigate whether the ages of the L2 learners and their L1 possible Chinese transfer could actually influence the ratings of the grammatical forms within some problematic syntactic structures of unaccusative
In Study II of this thesis, we intend to investigate whether the ages of the L2 learners and their L1 possible Chinese transfer could actually influence the ratings of the grammatical forms within some problematic syntactic structures of unaccusative