Error Analyses

As discussed, only eighteen instances of erroneous combination with to were found, further proving that they tend to occur in lower frequency. Among these cases, a large number of them seem to be the evidence of negative lan-guage transfer from the learners' native lanlan-guage, Mandarin, in which the equivalents of the preposition to are varied. It can be translated as dao4 ‘to reach a place; up to (a figure or time)’, dui4 ‘for or as regards’, and rang4 ‘to cause or to make someone do something’. Therefore, learners might use to to express similar meaning, as shown in (7).

(7)(a) ^?Then, we finally understood what Nash equilibrium is to the end.

(S007_008.txt)

(b) *Another reason of their individual performing styles is their unique interpretation to the roles they plays. (S007_002.txt)

In example (7a), the end functions as an indication of time instead of the goal of this process, so it should be “in” the end. The expression with dao4 in Man-darin dao4 zui4 hou4 (*to the end) might be the cause of confusion between the preposition to and in. In (7b), the roles they plays is the main issue to be interpreted (in a unique way), so learners included the preposition to to indi-cate this relation, which is usually expressed with dui4 in Mandarin dui4 jiao3 se4 de quan2 shi4 (*interpretation to the roles). It could thus be a possible rea-son of where this error comes from.

In addition to cross-linguistic influence, there are some cases which might be the results of an overgeneralization of the syntactic structure VERB to VERB, as in example (8).

(8)(a) *When we grow older and older, we become to leave home and make our own social life ... (S005_002.txt)

(b) ^?I regretted to tell her the truth because it has made her preoccupied for a long time, and blocked me from life routine. (S010_010.txt) Learners applied the VERB to VERB construction to whatever verbs they use regardless of the different meanings and syntactic structure implied in a spe-cific verb. For the verb become, it is usually followed by a description of what the subject will be like instead of an action that he or she is going to do. For the verb regret, there is a distinction between the forms of its complement that tend to be confusing for learners.

Two instances of misusing the infinitive are found, as in example (9), which could be explained by employing the image-schema theory.

(9)(a) *... you have better to choose online shopping. (S005_005.txt) (b) *Besides, I also have lots of fun to read the articles that the BBS users

have written everyday. (S004_004.txt)

In (9a), the learner used to as a connection that brings the end-point choose online shopping of this motion; however, the collocation have better does not entail a path for this movement. The same goes for to in (9b) as the motion read the articles is what the agent has been doing already, rather than the goal the agent intends to carry out. These analyses reveal that learners mistakenly conceived these events or processes on a path as something or someone that might change from one point to another or appear at the future end. Although misuses occurred in lower frequency, they provide evidence that learners were unsure whether it is correct to involve this path when using to with certain collocation.

4. Conclusion

This study, though carried out as a pilot study, will be expanded when larger amounts of corpora data are collected. However, based on the analyses above, some problems remain unsolved which will be our suggestions for future stud-ies. For the analysis of sense enumeration for to, we found that it is sometimes impossible to distinguish the semantic meaning from the syntactic meaning.

For example, one possible suggestion for this is to apply image schema theory, through which we might have different explanations of the preposition to. In addition to the senses, we also observed that some uses of to are affected by the meanings of the words around it; thus, the concept of coercion (Pustejovsky 1995) may become useful in explaining this phenomenon. Moreover, though this study is advantaged in its use of corpora data for error analyses, only a few errors were found in the two-word analysis. If we further examine three- or four-word combinations of to, more errors may be found, which will be worth exploring in the future.

References

Anthony, Lawrence (2005): AntConc: Design and development of a freeware corpus analysis toolkit for the technical writing classroom. In: Proceedings of the Interna-tional Professional Communication Conference (IPCC), Limerick, Ireland. Lime-rick: International Professional Communication Conference 2005, 729-737.

Gass, Susan / Selinker, Larry (eds.) (1983): Language transfer in language learning.

Rowley, MA: Newbury House.

Johnson, Mark (1987): The body in the mind. Chicago / London: The University of Chicago Press.

Kellerman, Eric / Perdue, Clive (1992): Special issue in cross-linguistic influence in second languageacquisition. New York: Pergamon.

Kleinmann, Howard H. (1977): Avoidance behavior in adult second language acquisi-tion. In: Language Learning 27: 93-107.

Lakoff, George (1987): Women, fire and dangerous things: What categories reveal about the mind. Chicago: University of Chicago Press.

Lewis, Michael (2002): The lexical approach. Boston: Thomson Heinle.

Merriam-Webster Online (2009): Internet: http://www.merriam-webster.com/dictionary/to

(last visited: 15.06.2009).

Odlin, Terence (1989): Language transfer. Cambridge: Cambridge University Press.

Pustejovsky, James (1995): The generative lexicon. Cambridge, MA: MIT Press.

Schachter, Jacqueline (1974): An error in error analysis. In: Language Learning 24:

205-214.

Scott, Michael (1999): Wordsmith tools Version 3. Oxford: Oxford University Press.

Selinker, Larry (1983): Language transfer. In: Gass / Selinker (eds.), 33-53.

Computational Linguistics and Chinese Language Processing

Vol. 16, No. 3-4, September/December 2011, pp. 21-40 21

© The Association for Computational Linguistics and Chinese Language Processing