Do you ever compare "reading age"/similar scores of source and target
Thread poster: Neil Coffey

Neil Coffey  Identity Verified
United Kingdom
Local time: 20:33
French to English
+ ...
Nov 3, 2012

A random thought that occurred to me: obviously, in doing any translation, one generally attempts to match the register of the chosen translation with that of the source. And that would generally include trying to avoid using, for example, a "rare" or formal/literary word in the target language to translate an "everyday" or "neutral" word in the source.

But, has anybody actually tried to assess this match a little more formally. For example, one could compare a "reading age" estimation of the source with a "reading age" (or similar schemes attempting to objectively assess the "level of reading difficulty/level of vocabulary") of the target.

That's not that I fully endorse the validity of "reading age" as a concept or system of measurement-- I actually think it's poorly founded pseudoscience to some extent-- but the point is that it's a proposed objective method for which tools are available.

Has anybody had any experience of using such tools to compare source and target texts?


Direct link Reply with quote
 

Phil Hand  Identity Verified
China
Local time: 04:33
Chinese to English
I try to look at sentence length Nov 3, 2012

I haven't been able to find widely used measure of language complexity in my source, so I haven't managed that kind of comparison. I do try to keep a rough eye on sentence length, though the two languages differ a lot.

I don't think reading age would work for my texts - everything would be "adult" or whatever the hardest rating is. But a more general measure of average word frequency would be interesting to look at.


Direct link Reply with quote
 

matt robinson  Identity Verified
Spain
Local time: 21:33
Member (2010)
Spanish to English
Not me, but... Nov 3, 2012

It would make a good PhD. thesis topic. I think it's probably the case that such descriptive attributes are inherent in any decent translation, but objective analysis would be interesting, if perhaps not useful. Having spent time 'downgrading' texts for readers with lower levels, it would be useful to be able to automatically spot anything which may not fit with the target level in terms of reader competence, but that is a matter unrelated to the original question.

Direct link Reply with quote
 

Jessica Noyes  Identity Verified
United States
Local time: 15:33
Spanish to English
+ ...
Different standards by country Nov 3, 2012

One of the services I offer is EN>EN simplifying English texts, so when I translate I tend to automatically break up the Spanish and French texts I work with into shorter sentences, use more active constructions etc. I know nothing about Spanish or French readability scales, but it appears to me that they would be quite different from English, as the Romance languages seem, as a baseline, to tolerate much more complex sentence structures. However, to comment directly on your question, Neil, I think it would be worthwhile to compare the two when fine-tuning the register. Do you or anyone know where to find reputable scales for French or Spanish?

Direct link Reply with quote
 

Neil Coffey  Identity Verified
United Kingdom
Local time: 20:33
French to English
+ ...
TOPIC STARTER
"Highest" rating Nov 3, 2012

Phil Hand wrote:
I don't think reading age would work for my texts - everything would be "adult" or whatever the hardest rating is.


Hi Phil -- you would think that, but some guidance actually advocates that writing-- at least for a general audience-- should have a much lower reading age than you might imagine.

For example, on the Gunning Fog Index, a reading age of 8 is apparently recommended for near-universal writing (bearing in mind, as the article suggests, that a score of 12 would presumably equate to senior high school student level).


[Edited at 2012-11-03 15:41 GMT]


Direct link Reply with quote
 

Helena Chavarria  Identity Verified
Spain
Local time: 21:33
Member (2011)
Spanish to English
+ ...
Native or non-native readers? Nov 3, 2012

I often ask if the text is addressed to native or non-native speakers of English (my language). For example, a couple of weeks ago I translated an opening speech that was going to be read by a non-native to a mixed audience. I kept the sentences relatively short and paid great attention to punctuation.

I'm afraid I still haven't reached the stage when I can translate full time (two years ago I thought the time was drawing near but translations have slowed down a bit for me) and I still have to work as an EFL teacher. However, the fact I teach helps me to be aware of the vocabulary I can/can't use.

When I translate for native readers, I go to town using what I consider 'real English' and when this is not the case I try to stick to 'text book' English, which is the language taught to people who study English as a foreign language.

Although, depending on the text, the level ranges between B2 and C2.

[Edited at 2012-11-03 15:50 GMT]


Direct link Reply with quote
 

Neil Coffey  Identity Verified
United Kingdom
Local time: 20:33
French to English
+ ...
TOPIC STARTER
Not sure of schemes for other languages Nov 3, 2012

I think it would be worthwhile to compare the two when fine-tuning the register. Do you or anyone know where to find reputable scales for French or Spanish?


I confess I don't. My gut feeling is that, say, the Gunning Fog index I mentioned could probably work more or less in Romance languages given that: (a) in English, compounds are apparently intended to be discounted before calculating the score [I wonder how successfully automatic tools actually do this]; (b) the words/sentence component, which would on average vary between English and Romance languages, has a relatively small weight; (c) the index's principal preoccupation, rightly or wrongly, is with words of 3+ syllables, and many such words in English will be of Romance origin anyway.

But... I think the scheme would more or less work in Romance languages precisely for the reasons that make them very approximate: they don't actually test any very specific language features.


[Edited at 2012-11-03 15:52 GMT]


Direct link Reply with quote
 

Phil Hand  Identity Verified
China
Local time: 04:33
Chinese to English
Surely word frequency would be a decent universal measure? Nov 3, 2012

Assuming that "words" are definable in a language, if you just develop a (log?) measure of how frequent all words are in a very large corpus, assign a suitably low value to any word not found in the corpus, then calculate the average for a given text, that should tell you something. You might have to weight that number based on the average score in the corpus.

Neil - for general audiences, certainly, but I do a lot of technical and legal, which would be off the charts. I was translating some blog posts a couple of months ago, and it would be a good measure for those.


Direct link Reply with quote
 


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

Do you ever compare "reading age"/similar scores of source and target

Advanced search







memoQ translator pro
Kilgray's memoQ is the world's fastest developing integrated localization & translation environment rendering you more productive and efficient.

With our advanced file filters, unlimited language and advanced file support, memoQ translator pro has been designed for translators and reviewers who work on their own, with other translators or in team-based translation projects.

More info »
Anycount & Translation Office 3000
Translation Office 3000

Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.

More info »



Forums
  • All of ProZ.com
  • Term search
  • Jobs
  • Forums
  • Multiple search