| A00-1002 | the text using this memory . The | translator | is free to make any changes to |
| A00-1002 | routine procedure . The human | translator | translating from the source language |
| A00-1018 | process to have a reviser look at a | translator | 's output . His job will be to |
| A00-1002 | target text created by a human | translator | is then compared with the text |
| A00-1002 | between our system and the TRADOS | Translator | 's Workbench . The method is |
| A00-1001 | base Schema was defined , and a | translator | from FOL to SQL was implemented |
| A00-1018 | prove quite embarrassing to the | translator | . Yet a diversity of situations |
| A00-1015 | The current implementation of | TRANSLATOR | uses a context-free grammar , |
| A00-1004 | text being drafted by a human | translator | ( Langlais et al. , 2000 ) . |
| A00-1003 | results ranging from 68 % of human | translator | performance for German , to 100 |
| A00-1002 | target language . When a human | translator | starts translating a new sentence |
| A00-1015 | with IBM 's VIAVoicE . The job of | TRANSLATOR | -- Or a different module conforming |
| A00-1018 | of argument , let 's consider a | translator | to be a black box with source |
| A00-1015 | can replace JSL and still use | TRANSLATOR | and even EXECUTER 's low-level |
| A00-1015 | utterance is shown in Figure 3 . Once | TRANSLATOR | has processed an utterance , |
| A00-1002 | method is simple -- the human | translator | receives the translation memory |
| A00-1015 | Though JSL is presented here , | TRANSLATOR | has been used to form Prolog |
| A00-1002 | systems . We have chosen the TRADOS | Translator | 's Workbench as a representative |
| A00-1002 | translation memory ( of TRADOS | Translator | 's Workbench ) amounts mainly |
| A00-1015 | passes to the NLP component , | TRANSLATOR | . We are using the IBM implementation |
