Forum: CAT Tools Technical Help
Topic: Efficiency of using CAT tools in comparison to using none
Poster: Samuel Murray
Post title: If you really want to compare
[quote]Sheila Wilson wrote:
Maybe some texts actually take a bit longer, if they're for a new project or a new client. Other texts may be very similar to previous translations and can be at least three times as fast, in my own experience. [/quote]
Yes, I think that if one really wants to answer the OP's question, one has to compare the speeds with texts that contain no repetitions or fuzzy repetitions. Obviously any tool that help find and type repetitions will result in a slight speed increase, but the real question (I believe) is whether translating a non-repetitive text is faster, slower, or the same speed in and outside a CAT tool.
As Sheila says, some translations can take longer, but I think that that depends mostly on the size of the job (a tiny job is likely to take longer in a CAT tool), whether the file needs extra setting up (e.g. tagging) and the complexity of the CAT tool.
I use mostly Wordfast Classic (WFC), with pre-configured project templates, so when doing a translation all I need to do is select the appropriate template (called INI files in WFC), create a new TM and a new glossary (these three tasks take 30 seconds, which includes the time spent navigating to the appropriate folder, typing in the file names, etc.), and off I go. So, from the start, I'm already 30 seconds slower with CAT than without it. (-: But from that point onwards, the fact that there is segmentation speeds me up. After the translation is done, the file needs to be "cleaned" of WFC-related code, and if the file is long, it can take a minute or longer for the cleanup process to complete (other CAT tools also have end-of-job processes, e.g. generating the target file).
Remember also that using a CAT tool with default settings only will mean longer startup times. If you haven't set up any project/TM/resource templates yet, then configuring a project for each new file might take several minutes each time. Once you've created templates for the types of jobs you regularly get, selecting a template for each project will reduce the starting time dramatically. For example, if you have a client for whom you need glossary settings that differ from the default settings, or whose files need to be tagged in a specific way, or whose files call for customised segmentation or tokenization, etc.
What I like about WFC (I don't want to turn this thread into a CAT tool comparison, though) is the ability to edit the source text on the fly. I don't mean edit the source segments, as many CAT tools now can do, but edit the source document itself, e.g. move around sentences within a paragraph, fix layout issues there and then, move around paragraphs, convert lists to paragraphs and vice versa if appropriate, etc. Most modern CAT tools use an intermediary format, which means that once the file is segmented, you must work with what you have (oh, you can sometimes merge and split segments, but that's it). So, with most modern CAT tools, you have to spend time fixing any potential issues before opening it in the CAT tool, or fixing those issues after the CAT tool generated the output file, and you have to bite the bullet when e.g. you stumble upon a section of tag soup that you didn't see initially. I realise that these comments relate more to quality than speed, but fixing up a file before sending it to the client is part of the job, so having to "fix" files in separate steps is something that can certainly impact your translation speed.
[Edited at 2018-08-14 08:32 GMT]
Topic: Efficiency of using CAT tools in comparison to using none
Poster: Samuel Murray
Post title: If you really want to compare
[quote]Sheila Wilson wrote:
Maybe some texts actually take a bit longer, if they're for a new project or a new client. Other texts may be very similar to previous translations and can be at least three times as fast, in my own experience. [/quote]
Yes, I think that if one really wants to answer the OP's question, one has to compare the speeds with texts that contain no repetitions or fuzzy repetitions. Obviously any tool that help find and type repetitions will result in a slight speed increase, but the real question (I believe) is whether translating a non-repetitive text is faster, slower, or the same speed in and outside a CAT tool.
As Sheila says, some translations can take longer, but I think that that depends mostly on the size of the job (a tiny job is likely to take longer in a CAT tool), whether the file needs extra setting up (e.g. tagging) and the complexity of the CAT tool.
I use mostly Wordfast Classic (WFC), with pre-configured project templates, so when doing a translation all I need to do is select the appropriate template (called INI files in WFC), create a new TM and a new glossary (these three tasks take 30 seconds, which includes the time spent navigating to the appropriate folder, typing in the file names, etc.), and off I go. So, from the start, I'm already 30 seconds slower with CAT than without it. (-: But from that point onwards, the fact that there is segmentation speeds me up. After the translation is done, the file needs to be "cleaned" of WFC-related code, and if the file is long, it can take a minute or longer for the cleanup process to complete (other CAT tools also have end-of-job processes, e.g. generating the target file).
Remember also that using a CAT tool with default settings only will mean longer startup times. If you haven't set up any project/TM/resource templates yet, then configuring a project for each new file might take several minutes each time. Once you've created templates for the types of jobs you regularly get, selecting a template for each project will reduce the starting time dramatically. For example, if you have a client for whom you need glossary settings that differ from the default settings, or whose files need to be tagged in a specific way, or whose files call for customised segmentation or tokenization, etc.
What I like about WFC (I don't want to turn this thread into a CAT tool comparison, though) is the ability to edit the source text on the fly. I don't mean edit the source segments, as many CAT tools now can do, but edit the source document itself, e.g. move around sentences within a paragraph, fix layout issues there and then, move around paragraphs, convert lists to paragraphs and vice versa if appropriate, etc. Most modern CAT tools use an intermediary format, which means that once the file is segmented, you must work with what you have (oh, you can sometimes merge and split segments, but that's it). So, with most modern CAT tools, you have to spend time fixing any potential issues before opening it in the CAT tool, or fixing those issues after the CAT tool generated the output file, and you have to bite the bullet when e.g. you stumble upon a section of tag soup that you didn't see initially. I realise that these comments relate more to quality than speed, but fixing up a file before sending it to the client is part of the job, so having to "fix" files in separate steps is something that can certainly impact your translation speed.
[Edited at 2018-08-14 08:32 GMT]