iShare does publish the full data regarding demographics, but for a price! They also only provide for free a very simplistic overview, and their panel in particular has significant bias towards geeks, I feel.

None of the online companies try to correct for national demographics, and not even national newspaper political polls make adjustments; they just present the raw numbers, but of course “correcting” the data itself is not an exact science.

The “real” surveys are the ones from Cabinet Office Japan, as they use random sampling from resident rolls, but they themselves bias towards retirees…

Next is goo Research, who do do some correction as far as I am aware; they are the best of the internet surveyors.

People like Macromill who concentrate on narrow demographics are reasonably accurate too, I think.

iShare/CLUB BBQ is pretty much junk statistically speaking, but it’s entertaining junk, I hope!

Anyway, thanks for the questions.

“I understand and agree but it seems that most international companies are switching to English as a working language”

Yes, absolutely. But that’s not the whole issue. Take academic research again, an area I know well. The research sphere has long switched to English as a world-wide working language. And yes, you can go abroad and not know a word of the local language – I did – and still be able to do your job.

But a researcher without the local language is fairly handicapped and a loss less effective than they could be. There’s big things and small: You can’t order supplies, or pick them out from catalogs – or pick them from the supply room; you can’t interact with experimental subjects; you can’t read or write most forms related to your funding, progress reports or anything involving applying for permission (for hazardous substances, animal handling or whatever). You can’t be safety manager, or act as host for local visitors. You can’t rent your own place or ask the super to help if something breaks by yourself. You may have major problems buying groceries, and a dental appointment that would take you half an hour at home becomes half a day with somebody from your lab helping you find a place, then act as translator.

In general, local people will have to help you with most of these things to a lesser or greater degree. You’re far less effective than you could be, and you’re dragging the productivity of people around you down as well – they’re no doubt happy to help, but that time is still time they could have spent on their own work.

Foreign posts in private companies are no doubt different, but similar issues arise there too. English may be the company language, but that doesn’t mean everybody is equally fluent, and after-work social gatherings will have people happily using their own local language, with the foreign guy mostly shut out. Customers want to talk to people in their own language, and few people are “back office” enough never to actually have to interact with outsiders. Even if you’re a pure technical guy, you may well need to talk to your customers’ technical personnel, for instance.

You don’t need perfect language skills for most of this; a little really goes a long way. But having some skill is vastly better than having none at all.

I understand and agree but it seems that most international companies are switching to English as a working language. For example, my father works in France for an international French company. A few years ago, they switched everything to English for corporate matters. The company documentation is now in English (it used to be in French). They can use French for daily activity, but they have to use English for dealing with headquarters (based in France).

Anyway, as usual, I have some concerns about the statistical validity of the surveys you present to us.

For a sample size of 1000, the margin of error is 3.2%. For a sample size of 100 (question Q4SQ2), the margin of error is 10%.

Furthermore, Club BBQ needs to publish for EACH survey the composition of the sample:
- is it representative of the Japanese population? If not, how does the sample composition differ from the overall population? (social status, income, urban/suburban, political leanings, etc…
- how did they correct the possible bias (persons having answered the survey vs club members, given answers vs real answers…)
- which sampling method did they use (quota sampling, stratified sampling, convenience sampling, etc…). Was there any control to correct the bias in sample selection?

I don’t think you can make any conclusion from this survey (and all the other surveys) because a lot of information is missing. I suspect these are just “random” internet surveys with no proper statistical analysis.

I appreciate the work you’ve been doing translating these “surveys”, but is there a way you also translate real surveys, which would given us a better view of “what Japan thinks”.

Anyway, it should be left to the employer to decide how much language ability is required for an employee or intercompany transfer. If someone has a basic level in a language, they will speed up quickly once they are immersed in it daily.

I know my own Japanese ability increases steadily while I am in the country. When I land at Narita I can order a coach ticket. By the time I leave one or two weeks later, my ear has tuned in, my rustiness has been polished off, and I am speaking much more.

That said, the company I work for apparently takes no notice of language ability when sending people on short or longterm overseas jobs.

But for general employees in Japan I agree that Japanese ability should be a requirement (and, say, German for German office transfers, French for the Paris offices and so on). One problem with it is that the JLPT is a pretty blunt tool for assessing ability; writing or speaking ability is not tested for a start, making it rather incomplete. And as far as I know there is no other good alternative.

Really? But pre-employment tests and interviews are presumably all in Japanese so anybody that gets a job does have a working level of Japanese without imposing any extra requirements.