A spelling test I've enjoyed - using the Spellr.us beta

I've been lucky enough to get a beta invite to test Spellr.us, which is a new web service for testing the spelling on your website. As someone with a long list of regular typos behind him on this blog, I thought it might be rather useful.

Getting started was very easy - I simply followed the email invitation and put in the credentials they'd supplied, and I was up and running. Of course, with it being a beta test, there are bound to be some glitches, and it turned out that the first time I tried to test the system they were in the middle of doing some upgrade work. It didn't seem to affect my session though.

Setting up a new spelling scan is pretty much as simple as typing in the URL. During the beta phase scans will only look at a maximum of 500 pages from a site, and there is as yet no ability to add or edit custom dictionaries.

One thing that I was immediately pleased to see was that Spellr.us recognise / recognize that there is more than one variation on a spell-checker required for the English language. You can specify British English, which ought to prevent me seeing a lot of false positives on currybetdotnet.

They also have 5 European languages covered in French, German, Spanish, Italian and Swedish.

Once you request a scan, the job is added to a queue. I would hope that in the fully functioning product you could either set up a regularly scheduled scan for a site, or be able to ping the Spellr.us service asking for a check on a specific new URL.

Potential errors are divided into two categories - words which are probably a mistake, and words which may be a mistake but which also may be names, jargon or acronyms.

When my first report came back it identified over 10,000 errors - but only 2,325 'uniques'. My gut instinct is that my spelling isn't as bad as all that.

As I dived into the report there were an awful lot of false positives. 'currybet' I can forgive, but for a system designed to test websites, I was surprised to see on the first page of my report 'blog', 'online', 'google', 'blogger', 'bloggers' and 'internet' appear. OK, well maybe the latter should be capitalised, and I know I'm inconsistent on that, but 'online'?

The nicest feature of the application is the ability to go to a 'screenshot' page, which shows your page with all of the spelling mistakes highlighted. This looks like it requires a little more work on the parsing side of things (it was flagging the 'href' in links as a spelling mistake for example, as well as 'blockquote') but it seems like a genuinely useful way to browse through a site looking for mistakes.

At the head of each screenshot there is a summary of all the problems found upon the page in question.

Several areas of the site are quite obviously still under a lot of construction - account management is very rudimentary for example, and you only know you are 'logged in' to the site because there is a big 'log out' button in the toolbar.

One area that is also a little hazy yet is the business model. There is a tab marked 'pricing', and when you select it you get the message:

"Pricing. Wouldn't you like to know...Basically - so would we! We will have our pricing sorted out shortly - and rest assured that we have taken all comments (both philanthropic and stingy) on board."

When you apply to join the beta programme, you are asked whether this is the kind of service you would be willing to pay for. I think I said 'no'. Having used it I think now that I might be willing to pay a little bit each month for it, but it would depend on the full feature-set.

For me, custom dictionaries are an absolute must, and also a system that would alert me to newly discovered errors, rather than me having to remember to go and do scans. The ability to also allow me to schedule reports and ping new content at the system would also make it a more attractive proposition.

Definitely worth keeping an eye on, but they seem a long way off from having a finished product yet, and they need to tune up that default dictionary for the early 21st century.