ACL4 Machine Translation Exam 1997

Andy Way and Lynne Bowker
Dublin City University

Attempt five questions.

1) Critically evaluate the following comments from Arnold et al. (1994:7)

MT is a waste of time because you will never make a machine that can translate Shakespeare.
There was/is an MT system which translated The spirit is willing, but the flesh is weak into the Russian equivalent of The vodka is good, but the steak is lousy, and hydraulic ram into the French equivalent of water goat. MT is useless.
Generally, the quality of translation you can get from an MT system is very low. This makes them useless in practice.
MT threatens the jobs of translators.
The Japanese have developed a system that you can talk to on the phone. It translates what you say into Japanese, and translates the other speaker's replies into English.
MT systems are machines, and buying an MT system should be very much like buying a car.

2) Assuming a translation bureau has integrated MT into its translation process, one might isolate the following subtasks as being integral components of this process:

Document preparation
Translation
Document revision

How do each of these impact on Document Managers, Authors and Translators? What tools, other than the MT engine itself, might also aid these three groups of people in their work?

3) In a reversible MT system, the processes of Analysis and Generation are viewed as mirror images of each other. One might, however, take an alternate view that they are quite different processes. Show how their demands differ by outlining some of the choices open to the system developer in constructing these components.

4) Empirical methods have recently made a comeback in many areas of NLP, including MT. Why do you think this is the case? Describe the processes involved in Example-based MT, and outline some of the advantages of this method over more traditional rule-based methods.

5) Why might you opt for a transfer-based system over an interlingual system? Substantiate your opinions with insights from translational ambiguities (e.g. river --> rivière, fleuve).

6) There are three main types of lexical ambiguity which pose problems for MT systems: polysemy, homography and homonymy.

Explain the differences between these three, supporting your explanations with examples.
While it is important to resolve all three types of lexical ambiguity in MT, some errors are more serious than others. Rate these three types of lexical ambiguity in order of importance in which they should be resolved, and justify your ranking order.

7) Briefly explain the main differences between the TRADOS Translator's Workbench tool and the Globalink TELEGRAPH tool. In what type of environment would the Translator's Workbench be preferable, and why? In what type of environment would TELEGRAPH be preferable, and why?

8) Discuss the advantages and disadvantages of adopting a sublanguage approach to MT. Briefly describe an MT system which has adopted this approach.

Andy Way, 10th March 2000.