Transblawg

I was at a loose end on Sunday so I decided to investigate the MemoQ translation memory program (CAT tool). I didn't get very far, so despite the fact that I attended a Webinar (bizarre word) today at ProZ, the following remarks may not be well-informed. The program looks very good, I must say.

My situation: I've been a happy user of STAR Transit since 1998, even though I'm a legal translator, and I've often heard other legal translators say TM is no use to them. I like it for terminology (all the terms I've recorded are shown highlighted on screen), for quality control (automatic checking of numbers and, if I want, of term consistency), and sometimes for its usual purpose: if I get two almost identical contracts at the same time, it's much easier to process them this way, because every single deviation of one from the other is shown in colour. I never supply clients with a TM (except for just one project, for which I had to use Trados Workbench but fortunately was able to import it into Transit).

Many buyers of these programs are translation companies or translation managers in companies, who use it to coordinate the work of many translators. At the moment Kilgray, the developers of MemoQ, are obviously trying to build up market share, so there are special offers around and lots of help (they may be just as helpful in future - I'm probably just suspicious, as usual). There was some emphasis on their server version today that wasn't relevant to me. I didn't listen very hard either. I think I missed something about using the Server for working from your laptop when you're away.

Although I intend to stick with Transit, I have long thought it would be a good idea to have a comparison and to know what would happen if STAR stopped selling it - I suppose they won't, but they don't always seem to make it as easy to buy and use as one would hope. In addition, their free version, Satellite, is unpopular with translators - in fact, handling Transit projects is one of MemoQ's selling lines. I never got round to investigating Déjà Vu, but that was very much aimed at freelances. Anyway, you can download a full version of MemoQ and try it out for 45 days, I think.

I was disappointed to read recently that, hardly had I learnt to pronounce MemoQ memmock, stress on first syllable, before attenders of a conference in Budapest persuaded Kilgray to change the pronunciation to memmockcue (stress on first syllable).

Notes:

1. It seems easy to understand (although it's easier for those who already know a TM system). The Getting Started Guide is short (about 38 pages, half of them relating to the Server) but there's more in the help.

2. Projects are stripped of their format (Word, PPT, HTML etc.) and exported back to it at the end.

3. I haven't yet worked out what to do about spaces. I have spaces showing in Word and Transit. Here, it seems the program inserts a space after each full stop, but if you already have one, you end up with two. There's probably a way around this.

4. Bold and italic can be seen as such. Other format features (footnotes, for example) appear as tags - numbers in curly brackets - in the source text, and I gather they have to be copied over. Didn't see this. I'm used to seeing tags in Transit (but Transit has views without tags too).

5. Segmentation was good (so was alignment).
But how do you add segmentation rules? Apparently you use some regex characters - I haven't tried it.
(When I import a text into Transit, I am asked to identify abbreviations. GmbH I would not mention, because if it has a full stop after it, that's always the end of a sentence, so it can be segmented, whereas in running text it has no full stop. But Art., Rdnr., ff. and so on could be added to the rules).

6. We were told that you can even export bilingual documents. As far as I can tell, this means you export a document to MS Word in the typical Trados style, which is not interesting. I may have overlooked another possibility.

7. I imported, slightly processed and exported an XML file, to make sure everything worked. I was impressed to see not only the DE and EN columns, but also another window showing the XML text with tags, just as you would see it in an html program. What I didn't notice until the Webinar was that this text was a real-time preview - it changed into English as I processed it. The window showing the original format of various text types is obviously something you get with other programs now, but the real-time preview probably not (although I must say, just seeing the German XML in its original format is the most important thing for me).

8. I was most interested in the subsegmentation search in the concordance. I didn't create a big TM, so I didn't have a very detailed impression of it, but I saw how phrases can be marked automatically and you can look them up in the concordance if you want. This is something I'd like to do with contracts - scarcely ever are whole segments identical, but phrases often are. It's been pointed out to me that I can do something similar if I reduce the percentage by which the fuzzy index has to match the new text down to 20% or 10%. I will try that out. Transit NXT also apparently can be set to automatically search through the concordance in this way if no fuzzy-match sentences appear.

9. MemoQ has not only subsegment search, but also fragment search. I did not understand that till the Webinar: it means that a complete segment, such as the heading 'Gründe' ('Grounds') - however many words in length - is treated rather like terminology and a translation is offered when it crops up as part of a later sentence.

10. One of the nice things about Transit is that instead of keeping one or more TMs, you keep sets of files. A pair of files ending in DEU and ENG are available to be used in the same way as a memory. This means that I might have a folder labelled Contracts, full of pairs of files, and another folder labelled Websites, and so on, but I can mix and match the folders and use as many or as few as I want. With MemoQ, you can have as many TMs as you like (and as many term bases) as far as I can see, but when you create a project you have to decide which TM it is to feed into, whereas in Transit you don't need to make a decision. I also think a TM can take up more room on the drive, but I'm not sure about that.

11. I am used to the appearance of Transit. But I'm sure I could get used to MemoQ. One thing I find odd is that in Transit, parts of a segment that don't match the fuzzy match (TM example) are coloured, whereas in MemoQ, it's the matching subsegments and fragments that are coloured. I am used to seeing the new marked, but then I don't know how Transit NXT handles that (I plan to try it out shortly) - obviously it must have some way of marking concordance subsegment matches too, and that needs to fit in without conflicting with its other uses of colour/highlighting.
Here's an example from the Webinar:
Old sentence: Mit Bildschirm-Regler MBI und Beleuchtungs-something or other
New sentence: Mit Bildschirm-Regler MBI, ohne Beleuchtungs-something or other

I would expect the new word ohne to be emphasized, whereas MemoQ emphasizes the compound nouns, I seem to remember.

12. When numbers in source and target segments don't match, a red exclamation mark appears in a small column to the right. In Transit, checking numbers is a separate operation at the end. I quite like the immediate indication. Numbers often don't match, for instance where German tends to use figures over 20 or over 12, and English tends to use words up to ninety-nine.

13. Obviously there are lots of general settings, which I haven't tried out yet. I found that every time I started a project, I had to select the source and target languages from a scroll-down list. There is probably a way to avoid that if you always use the same language pair.

14. (Added later) The terminology module in Transit - Termstar - is more complex than the MemoQ one. As far as I can tell, the latter permits a number of equivalents and has separate tabe for usage, grammar and definition. But I haven't really inspected it yet. I presume I could easily import CSV-delimited files with EN and DE terms and could place all my miscellaneous notes in the Definition field - I sometimes save discussions about the translation of legal terms, which can be tricky.

15. It was a good idea to have a webinar. A big problem with these programs is that they tend to be very complex and if you attend a seminar, you may find the trainer doesn't know much about translation but is more interested in software development. So the more opportunities to hear a different speaker, the better.

I think that's all I have to say for the time being. I need to try out Transit NXT, which involves a lot of learning, I'm sure, and see if different fuzzy index match percentages help me use my old contract translations.

LATER NOTE: ProZ Webinar for MemoQ here (not live, of course).

E-LEARNING (free) for Transit NXT here - I'll probably write up Transit NXT eventually, as I'm about to try it out.

Louise Frances Denyer, of (or connected with) Birmingham University, has a PDF on the Web of an 85-page paper (about half of it consists of appendices with language examples):
Corpus study carried out on three 'legal' verbs to demonstrate their similar and different usage for the purposes of legal translators and lawyer-linguists.

The paper aims to show the valule of corpus studies in translation and legal-language teaching. I've only skimmed it so far, but unfortunately the corpora used, which are English, Spanish and French, don't appear to be specifically legal texts. The three verbs were chosen because the author's students wanted to know more about them - I remember this problem with another collection of verbs mentioned here: abolish, override, set aside, quash etc.

The following is really only based on skimming the article:

Breach often collocates with peace, security, contract, confidentiality, rules, rights, and fiduciary (MM: should be fiduciary duty, not just 'breach of fiduciary' as suggested).

Violate collocates with rules, codes, covenants, terms, prohibitions, and court injunction; also with airspace. It is very common in connection with US constitutional rights, whereas in British English it can relate to minor offences.

Infringe collocates with IP rights, such as patents, copyrights, fundamental rights.

I found this article via the programme of a conference on corpus linguistics in Murcia this week, tweeted by Matthew Bennett, seeing the topic and then Googling the author's name.

I have got a book on corpora that seems very good, but I am only part-way through it: Working with Specialized Language. A practical guide to using corpora, by Lanne Bowker and Jennifer Pearson, Routledge 2002.

I've mentioned Stenography before. Today Jack Schofield shows there is also Palantyping, in answer to this question:

I attended a gathering in Richmond Theatre, at which the then mayor and members of the Greater London Authority were available for public questioning. It was very impressive that their words almost immediately appeared on a screen courtesy of a voice recognition system. How is it done?

1. Some people would appreciate sending their MacBook Air in for repair and getting it back with a German keyboard. I use a German keyboard all the time, because I can write German or English on it.

MacMacken: Amerikanische MacBook Air-Reparatur mit Umlauten

2. Some astronomers (simplex by name) do not believe that specialist translators know about their subjects:

Und: gibts wirklich z.B. ausgebildete Übersetzer, die genügend Ahnung von Astronomie haben um mit den ganzen Fachbegriffen klarzukommen? Das war nämlich auch ein Argument gegen einen Übersetzer...

Comment by author under Lost in Translation at Astrodictum simplex

Via Gabi, who writes:

Da meint wieder einmal jemand, einen Fachübersetzer nicht nötig zu haben und ohnehin alles viel, ja unendlich viel besser zu wissen als ein Mensch, der sich sein Leben lang mit der Übersetzung genau dieses Fachgebietes befaßt hat. Also will er seine Software selbst übersetzen. Das Ergebnis kennen wir alle: eine Software, bei der sich der Anwender fragt, welcher Vollidiot die geschrieben hat … welcher größenwahnsinnige Trottel sich an die Übersetzung einer Software wagt, obwohl er nicht einmal so simple Fachbegriffe wie »pan« und »crossmatch« übersetzen kann.

The Augsburger Allgemeine reports the case of the Augsburger charged with smuggling 180 Greek tortoises into Germany.

The public prosecutor's office produced the following text in the indictment:

Dem angeschuldigten Arschloch ist ein Pflichtverteidiger zu bestellen.

(Defence counsel shall be assigned to the accused arsehole.)

This was apparently the result of speech recognition software and office incompetence. The error was corrected but the correction not properly saved to disc (I think it should simply have read 'Dem Angeschuldigten' (the accused).

Leitender Oberstaatsanwalt Reinhard Nemetz ist betroffen. Er spricht von einem "nicht akzeptablen" Fehlverhalten seiner Behörde. "Ich muss mich bei dem Bürger ausdrücklich und in aller Form entschuldigen", sagte Nemetz am Montag unserer Zeitung. Der Behördenchef hat eine penible Überprüfung des Vorgangs angekündigt und betont: "Ich will wissen, wer dafür verantwortlich ist."

The case has been stayed for witnesses to be summoned from abroad. The defendant has a previous tortoise-smuggling conviction, but denies the present charge.

I suppose it makes a change from blaming it on the interpreter.

(From Digital Diktieren - the speech recognition blog - which reports that Dragon Naturally Speaking contains such words in the new version, but some programs delete them).

Thank goodness they're still selling it.

Goodbye Kompatibilitätsmodus.

Mouse soap from yeduo in Taiwan.

Via Gizmodo.de

There must be a lot of errors made with search-and-replace on newspapers. But somehow I seem only to read about the PC ones.

For instance, Tyson Gay is not called Tyson Homosexual.

I also recall reading of Nelson Mandela referred to as an Afro-American.

Aus für Windows Vista. Zumindest auf dem Papier. Das in Fürth ansässige Medienunternehmen Computec Media AG nimmt die Computerzeitschrift nach der Juli-Ausgabe, die am 18. Juni erscheint, vom Markt. Objektleiter Thilo Bayer begründet den Schritt auch damit, "dass Windows Vista als Betriebssystem die hohen Erwartungen bis heute nicht erfüllt hat."

Fürth leading the way.

From textintern, via Volker Weber.

Fränkische Dasdatur (and others)



(Via downwärts)

Josip M. Korbar of LTT Technical Translations has prepared a PDF file with lots of clear photos on how to set up several monitors on a laptop, especially with the Matrox DualHead2Go.



Not much room for a cat there, of course.

Download the file Multi-Monitor-Arrangement_How_To.pdf (I hope).

Gizmodo UK surprises with the headline Keep your pussy off the keyboard.

Needless to say, the US original sticks to 'cat' (not even 'kitty').



The Kit-In Box is by The Refined Feline, who also do the Refined Litter Box, with mahogany, cherry or birch stain.

Now back to the exercises.

It sounds very much as if Steven Levy's wife threw his MacBook Air (he'd been lent it for review) into the trash compacter - see Gone, Without a Trace, in Newsweek.

Well, one of Beuys' Fettecken was disposed of even when it was in full view, rather than hidden in a pile of papers like the MacBook. And another German victim was Gustav Metzger at Tate Modern:

The spokesman said: "The bag was removed the morning before we first opened. It was not put in a crusher but in a skip. It was removed but it was seen that it was damaged. That's why the artist put another bag in its place."

He pointed out that Metzger had not actually had to pick up new rubbish to put in his replacement bin liner but had simply found one that had already been filled and discarded. "The bags are taken from where they are found and put in the [gallery] space," said the Tate. "He doesn't manipulate what's in the bag."

I think I could help.

Deutsche Fassung bei Spiegel Online.

Jost Zetzsche produces a biweekly newsletter on computer tools for translators (biweekly here means every two weeks). There is a fuller version for $15 p.a., and also The Translator's Tool Box as a $50 password-protected PDF file (table of contents etc. as PDF). I've only used the free version so far, but the quality suggests the money wouldn't be wasted.

Incidentally, the idiosyncratic TEnT means Translation Environment Tool (what we call CAT). It took me ages to find that out!