Welcome to WebmasterWorld Guest from 188.8.131.52
Forum Moderators: phranque
Please read this post with a strong French accent. (Try to immitate Cluzo it will sound funnier)
Here in Québec (a Canadian province) most people speak French. Due to the international nature of the Web most corporate sites are at least bilingual, if not multi-lingual. The problem is that they use the same domain name, in this classic structure : root/language/files. This lead to problems like having a splash home page without a single word but a logo and image buttons to pick a language. Even worst, home page without a single word but a browser language sniffer script, and no links for spiders to follow.
Those technical problems can be easely fixed, but some others problems requires a lot more work to resolve them.
As renke mention in this old tread [webmasterworld.com] :
There is some concern that multilingual sites may not fare so well in major SE:s. I am recommending all my customers to split the language versions into separate sites and register them as either a local language dot-com or (where possible) as a purely local domain. Keep in mind, that the leading SE in each country is usually a local engine, indexing sites in the local language only.
Also, Marcia started this tread [webmasterworld.com] in wich Brett added precious informations about whow
some search engines can recognize keyword relationships and relevancies within "families" or broad groupings.
Combine these two factors (inclusion and theme), add inboud links issue and you get a pretty challenged site to deal with.
More and more search engine add some language translation services but I guess it is a long shot untill they integrate keywords relationships from different languages. If half of my site is talking about "potato" and the other half is dealing with "patate" I guess I defenitely have a problem with the main theme and the relevance of inbound links also from different languages.
I am sold to the idea of having a different domain name for each language version of a site. I think it is the only way to go. Since it requires a lot of work, it is hard to sell such a solution to my clients. The last time I tried I foud myself out of ammo and lost the battle.
Does anyone have any other convincing facts, numbers or alternatives to share?
You probably already do that, but that will at least get you into the proper db classification at the se's.
The welcome mat page is tricky. I think you should identify which group is your largest user. Then make the home page in that language, with a prominent link to the alternative language.
Getting clients to see the value is difficult if not impossible. They want to brand out only one domain name - I can understand that. An alternative might be go use a 3rd level domain such as, "french.domain.com" and the root "domain.com" as the primary site language.
Lastly, if you have the tech, you can identify about 70% of your users simply by the ISP they use. Even if you just hit the big isp groups like (aol.com, msn.com, home.com,..etc) with a default language, you'll guess right about 90% of the time.
As for lang attributes, I got it all wrong! (AAARG!) I was using the meta language only. There is much to learn at W3C...
Your suggestion for using a 3rd domain level is good, but in some cases this can be saved to use elsewhere. Like for a Canadian portal I work for, we use "sports.domain.com" for the English version and "sports.domain.ca" for the French one. This way we can benefit from a broader placement in main directories like Yahoo! and a good keyword in the URL. I believe the inconvenient is that the total of inbound links are split between sub domains.
We applied this structure 3 months ago and it is too early to have some clear figures yet.