 |
|
General Trends Observed
A number of general trends came to light during the analysis of the data from this pilot study.
14 Meta, XLIII, 4, 1998
3.8.1. General trends in dictionary use vs. corpus use
We observed that for each of the five categories of error, the students using the corpus made fewer total errors than the students using the conventional resources. For three of the categories, namely subject field comprehension, term choice, and non-idi- omatic construction, the improvements shown were reasonably significant. For the remaining two categories, grammatical error and incorrect register, the improvements were marginal. With regard to grammatical error, the majority of errors made by both the conventional resource users and the corpus users seemed to result from sloppiness (e.g. misused punctuation, incorrect capitalization), rather than from a lack of infor- mation in either resource. In the case of register, the problems seemed to be largely related to particular students, rather than to either of the available types of resources. Table 8 shows the breakdown of errors made by each student for each category in text i, and table 9 shows a similar breakdown for text ii. Table 10 summarizes the relative improvement made by corpus users over conventional resource users for each of the categories of error.
| 1-14=student; D=dictionary user; C=corpus user; i=text i | Comprehension errors | Production errors | |||
| Students | subject comprehension error | incorrect term | non- idiomatic construction | grammatical error | incorrect register |
| 1-D-i | |||||
| 2-D-i | |||||
| 3-D-i | |||||
| 4-D-i | |||||
| 5-D-i | |||||
| 6-D-i | |||||
| 7-D-i | |||||
| Total errors | |||||
| Average errors per student | 4.71 | 4.71 | 0.42 | ||
| 8-C-i | |||||
| 9-C-i | |||||
| 10-C-i | |||||
| 11-C-i | |||||
| 12-C-i | |||||
| 13-C-i | |||||
| 14-C-i | |||||
| Total errors | |||||
| Average errors per student | 1.86 | 3.29 | 3.14 | 2.71 | 0.14 |
Table 8
Errors made by each student in each category in text i.
NATIVE-LANGUAGE CORPORA AS A TRANSLATION RESOURCE 15
The general trend shows that for the categories of subject comprehension, term choice, and non-idiomatic expression, students using the corpus made significantly fewer errors than students using conventional resources. For the remaining two catego- ries, namely grammatical errors and register, students using the corpus performed only marginally better than students using conventional resources
| 1-14=student; D=dictionary user; C=corpus user; ii=text ii | Comprehension errors | Production errors | |||
| Students | Subject comprehension error | Incorrect term | non- idiomatic construction | Grammatical error | Incorrect register |
| 8-D-ii | |||||
| 9-D-ii | |||||
| 10-D-ii | |||||
| 11-D-ii | |||||
| 12-D-ii | |||||
| 13-D-ii | |||||
| 14-D-ii | |||||
| Total errors | |||||
| Average errors per student | 2.14 | 4.29 | 4.43 | 3.43 | 1.14 |
| 1-C-ii | |||||
| 2-C-ii | |||||
| 3-C-ii | |||||
| 4-C-ii | |||||
| 5-C-ii | |||||
| 6-C-ii | |||||
| 7-C-ii | |||||
| Total errors | |||||
| Average errors per student | 1.23 | 2.29 | 3.43 | 3.29 | 0.86 |
Table 9
Errors made by each student in each category in text ii.
The general trend shows that for the categories of subject comprehension, term choice, and non-idiomatic expression, students using the corpus made significantly fewer errors than students using conventional resources. For the remaining two catego- ries, namely grammatical errors and register, students using the corpus performed only marginally better than students using conventional resources
16 Meta, XLIII, 4, 1998
| Errors made by dictionary users | Errors made by corpus users | Relative improvement | |
| Text i | |||
| Subject comprehension error | 38% | ||
| Incorrect term | 30% | ||
| Non-idiomatic construction | 33% | ||
| Grammatical error | insignificant | ||
| Incorrect register | insignificant | ||
| Text ii | |||
| Subject comprehension error | 40% | ||
| Incorrect term | 47% | ||
| Non-idiomatic construction | 23% | ||
| Grammatical error | insignificant | ||
| Incorrect register | insignificant |
Table 10
A summary of the relative improvement made by corpus users over users of conventional resources in the five different error categories
3.8.2. Individual student trends in dictionary use vs. corpus use
In the case of individual students, the majority of students also showed either a general trend of improvement or no change when using the corpus. Only the work of student (13) showed a general tendency to decline when using the corpus. This is illus- trated in table 11.
Student Comments
In addition to doing the translation using either the conventional resources or the corpus, the students were asked to comment on the usefulness of these tools.17Several general patterns emerged from these comments.
3.9.1. Comments about the conventional resources
There were some feelings of dissatisfaction regarding the information provided by conventional resources. The most common frustration was that the type of information the students were looking for was not found in these resources,18specifically, informa- tion about different types of scanners:
(8) "I didn't find the monolingual dictionaries very useful in this situation because they didn't really provide you with the specialized terms relating to scanners. The only information they provided was a very basic and broad definition of the concept. It never mentioned the types: hand-held, flat-bed, etc."
(10) "I did not find the specialized dictionaries very helpful for this exercise. Some of the words I looked up were not listed (e.g. flatbed scanner) or the information was insufficient (e.g. scanner)."
(11) "I didn't find the specialised dictionaries helpful really at all. I looked up 'scanner' thinking it would give the names of the different types but it didn't."
(13) "I did not find the technical dictionaries to be of great assistance as the terms were either very broad (dictionary of computing terms) or very technical (like the engineering dictionary). I could not find any entry referring to the different types of scanner."
NATIVE-LANGUAGE CORPORA AS A TRANSLATION RESOURCE 17
Two students (4, 5) indicated that they actually preferred working with the con- ventional resources because they were more used to this method of working:
(4) "Maybe it's force of habit, but I felt I used the dictionaries more."
(5) "I really didn't find the corpus tools much use; the dictionaries were much handier, quicker and easier to use. Then again, I may have been a little afraid of the corpus tools!"
Whenever a new method of working is introduced, there is always a learning curve
| 1st number=dictionary use; 2nd number=corpus use; *=more errors made using corpus | |||||
| student | Subject comprehension | Incorrect term | non-idiomatic construction | Grammatical error | Incorrect register |
| 3-1 | 4-2 | 3-3 | 5-4 | 1-3* | |
| 3-2 | 4-2 | 4-3 | 1-3* | 0-0 | |
| 2-1 | 6-3 | 7-5 | 3-4* | 0-0 | |
| 5-0 | 6-3 | 4-4 | 1-3* | 0-0 | |
| 4-2 | 5-2 | 3-3 | 3-3 | 0-0 | |
| 2-1 | 2-1 | 5-2 | 5-3 | 2-3* | |
| 2-2 | 6-3 | 7-4 | 3-3 | 0-0 | |
| 2-2 | 2-4* | 4-1 | 6-4 | 1-0 | |
| 3-2 | 6-2 | 7-7 | 4-1 | 2-1 | |
| 2-2 | 7-4 | 2-0 | 2-2 | 2-0 | |
| 3-2 | 2-2 | 4-5* | 3-2 | 0-0 | |
| 2-1 | 5-2 | 8-4 | 5-3 | 2-0 | |
| 13* | 2-3* | 4-4 | 4-5* | 3-4* | 1-0 |
| 1-1 | 4-4 | 2-0 | 1-3 | 0-0 |
Table 11
Difference in errors made by students when using dictionaries (first number)
and corpus (second number)
involved. Hopefully, improved training on the use of the corpus analysis tool and more practice applying it in a translation environment would result in a reduction of this type of resistance to using the corpus. Another related point is that pre-conceived attitudes and expectations can have a significant impact on the acceptance of a new method of working. Student (14) made the following comment: "Although I was initially dubious of its possible use, the concordancer was in fact quite useful." Fortunately, this student was open-minded, but care must be taken to counter pre-conceived notions by prop- erly educating translators about both the benefits and limitations of computer aids.19
3.9.2. Comments about the corpus
A number of the students (8, 10, 12, 14), indicated that they found it time-con- suming to work with the corpus; however, student (10) did add that, "if one was more used to the system, it hopefully wouldn't be so time-consuming."
18 Meta, XLIII, 4, 1998
Those students who did find the corpus useful indicated that it was helpful in the following ways:
student (8) seemed to find the corpus useful for subject-field understanding, stat- ing "I think that the concordances were useful for determining the exact sense of a particular word regarding the subject of scanners, considering I don't know much about them."
students (2, 6, 12, 14) found the concordancer useful for helping them to decide on an appropriate term (e.g. hand-held scanner as an equivalent for scanner à main).
students (2, 9, 14) found the frequency information useful for helping them to choose between two possible terms (e.g. scanning mechanism vs. scanner mecha- nism).
students (12, 13, 14) found the corpus useful as a means of verifying or confirm- ing that a term did in fact exist or that it was being used correctly.
CONCLUDING REMARKS
On the whole, the pilot study appears to have been successful in supporting our hypothesis that a specialized monolingual native-language corpus can be a useful resource for translators translating into their native language. Translators using a corpus have access to a greater amount of data, which they can interrogate more easily. The results of the study showed a general trend towards improved quality translation for the categories of subject-field understanding, correct term choice, and idiomatic expres- sion. Although there was no significant improvement in the categories of grammatical error and incorrect register, we are happy to observe that the corpus did not contribute to a decline in performance either. We hope to eventually expand this study to include experiments involving other text types, other subject fields, and other language pairs to see if we can draw more definitive conclusions about the usefulness of a monolingual corpus as a translation resource.
Acknowledgements
I would like to thank the students who participated in this project, and also Dr. J. Williams (Dublin City University) and Dr. I. Meyer (University of Ottawa) who provided helpful feedback on an earlier version of this paper.
Notes
1. There are, of course, exceptions to this practice. For example, cases where it is desirable, for reasons of both national cultural identity and economic survival, to have translation out of a socalled minor language (e.g. Finnish) into a socalled major language (e.g. English), but where the volume of work exceeds the num- ber of available majorlanguage native speakers (McAlester 1992: 292).
2. Discussions with colleagues at other universities in both Europe and North America have revealed that my experiences in this regard are not isolated incidents, nor are they limited to native English speakers.
3. At present, DCU students have only one course in English language skills which they follow during the first year of the translation program.
4. In addition to providing them with such resources, we should also endeavour to provide them with the skills necessary to compile similar resources of their own in the future.
5. The use of corpora in the discipline of computerassisted language learning (CALL) is wellestablished; however, the most common application is for the learning of foreign languages (e.g. Kettemann 1996), rather than for the finetuning of nativelanguage skills. Conventional CALL applications may be useful for helping translators improve their foreign language skills, but such investigations are beyond the scope of this study.
6. This statement describes the situation today; however, it may change in the future as more translators make use of tools such as translation memories to create bilingual, parallel corpora.
NATIVE-LANGUAGE CORPORA AS A TRANSLATION RESOURCE 19
7. Tools are also being developed which can be used to conduct conceptual as well as linguistic searches
(e.g. Text Analyzer (Kavanagh 1995)).
8. There are, of course, exceptions, such as the systematically organized Glossary of Computing Terms published by the British Computer Society (see bibliography for full reference).
9. The full references for the resources are listed in the bibliography.
10. One of our research partners, The Language Analysis and Knowledge Engineering (LAKE) Laboratory of the University of Ottawa, subscribes to Computer Select, and through them we obtained permission to extract and use corpora for research purposes.
11. The discs are updated monthly, and the disc issued in December of each year contains all the articles published during that calendar year.
12. See tables 2 and 3 for a breakdown of the text types found in Computer Select.
13. The articles classified in the genre technical were not technical manuals, but rather they seemed to be articles dealing with highly specialized subjects (e.g. “Polyester media development for inkjet printers” in HewlettPackard Journal or “Fuzzy cognitive maps model social systems” in AI Expert). The majority of these articles appeared in learned journals.
14. The creators of Computer Select have indexed each of the articles on the disc according to certain key words (i.e. topics). Users can instruct the computer to 1) locate all the articles that have been indexed accord- ing to a given key word (e.g. scanning technology), and 2) download these articles into a file on a hard disc.
15. These lexicographic and nonlexicographic resources were in paperbased form (see section 3.4.1).
16. The * is a wildcard character that can be used to represent any letter or combination of letters (e.g. a search on “sensitiv*” would find occurrences of both “sensitive” and “sensitivity”). The number in paren- theses after the term indicates the number of occurrences of this term in the vicinity of “sensitiv*.”
17. Unfortunately, not all of the students did provide comments on the tools. No comments were received from three students (1, 3, 7); however, the remaining eleven students did provide some comments, in varying degrees of detail.
18. Interestingly, none of the students commented on the usefulness of any of the resources other than the dictionaries (i.e. the journal article, the user manual or the desktop publishing monograph). My observation from supervising the experiment is that they, like many students, relied heavily on dictionaries and made little effort to use the other resources provided.
19. The paradoxical combination of unrealistic expectations for machine translation and the fear of being replaced by a machinetranslation system means that many translators are particularly susceptible to being reluctant to incorporate computerassisted translation tools into their work practices.
REFERENCES
ATKINS, S., CLEAR, J. and N. OLSTER (1992): "Corpus Design Criteria", Literary and Linguistic
Computing, 7 (1), pp. 1-16.
BAKER, M. (1992): In Other Words: A Coursebook on Translation, London, Routledge.
BAKER, M. (1996): "Corpus-based Translation Studies: the Challenges that Lie Ahead", H. Somers (Ed.), Terminology, LSP and Translation: Studies in Language Engineering in Honour of Juan C. Sager, Amsterdam, John Benjamins, pp. 175-186.
BOWKER, L. (1995): A Multidimensional Approach to Classification in Terminology: Working Within a
Computational Framework, PhD Thesis, University of Manchester, UK.
BOWKER, L. (1996): "Towards a Corpus-Based Approach to Terminography," Terminology, 3 (1), pp. 27-52. CORMIER, M. (1991): "Traduction de textes destinés à des spécialistes: Approche pédagogique," Meta, 36
(2/3), pp. 440-447.
DANIELSSON, P. and D. RIDINGS (1996): "Corpus and Terminology: Software for the Translation Program at Göteborgs Universitet or Getting Students to do the Work", Proceedings of Teaching and Language Corpora 1996 (TALC'96), Lancaster, University of Lancaster (UCREL), pp. 57-67.
FOLKART, B. (1984): "A Thing-Bound Approach to the Practice and Teaching of Technical Translation",
Meta, 29 (3), pp. 229-246.
KAVANAGH, J. (1995): The Text Analyzer: A Tool for Knowledge Acquisition from Texts, MSc Thesis, University of Ottawa, Canada.
KETTEMANN, B. (1996): "Concordancing in English Language Teaching", Proceedings of Teaching and
Language Corpora 1996 (TALC'96), Lancaster, University of Lancaster (UCREL), pp. 4-16.
LAVIOSA, S. (1998): "The English Comparable Corpus: A Resource And A Methodology", L. Bowker, M.
Cronin, D. Kenny and J. Pearson (Eds), Unity in Diversity? Current Trends in Translation Studies,
Manchester, St. Jerome Publishing.
MCALESTER, G. (1992): "Teaching Translation into a Foreign Language — Status, Scope and Aims", C.
Dollerup and A. Loddegaard (Eds), Teaching Translation and Interpreting: Training, Talent and
Experience, Amsterdam, John Benjamins, pp. 291-297.
20 Meta, XLIII, 4, 1998
MCENERY, T. and A. WILSON (1993): "Corpora and Translation: Uses and Future Prospects", Technical Report from the Unit for Computer Research on the English Language (UCREL), University of Lancaster.
MILLER, D.R. (1993): Towards Knowledge-Base Systems for Translators, MA Thesis, University of Ottawa, Canada.
NEWMARK, P. (1988): A Textbook of Translation, New York, Prentice-Hall.
PETERS, C. and E. PICCHI (1998): "Bilingual Reference Corpora for Translators and Translation Studies", L. Bowker, M. Cronin, D. Kenny and J. Pearson (Eds), Unity in Diversity? Current Trends in Translation Studies, Manchester, St. Jerome Publishing.
ROUMEN, R. and T. VAN DER STER (1993): "CONTEXT: A New Concept in Computer-Aided
Translation", TKE'93: Terminology and Knowledge Engineering, Frankfurt, Indeks Verlag, pp. 215-220.
SINCLAIR, J. (1991): Corpus, Concordance, Collocation, Oxford, Oxford University Press.
SYKES, J.B. (1989): "Chapter 3: The Intellectual Tools Employed", C. Picken (Ed.), The Translator's
Handbook (2nd ed.), London, Aslib, pp. 35-41.
VAN DOORSLAER, L. (1995b): "Quantitative and Qualitative Aspects of Corpus Selection in Translation
Studies", Target, 7 (2), pp. 245-260.