How do search engines handle hyphenated words?

Question

I am not sure my title fully explains what I mean. I thought this might be an interesting question.

If I had a set of keywords, broken with a dash or 2, will search engines consider the dashed split keyword as maybe a full keyword?

Say I have a site that sort of breaks words down, like the dictionary sites do. So a keyword for that page, might end up in the page, and / or the URL, as broken by dashes.

Key-word = keyword
Co-op-er-at-ive = cooperative
Pho-to-gra-phy = Photography

www.example.com/key-word/
www.example.com/co-op-er-at-ive/
www.example.com/pho-to-gra-phy/

I know search engines will consider a dash (at least Google) as a space, and understand it as multiple words. But in the English language, a dash can also break a word down (at least I think it can, can't it?), so will search engines also take this into consideration? I did a 'little' research, I Googled some words and placed random dashes, and it returned the words I searched for, but this could be considered a typo from the user on Google's search end, so really I am wondering if I can purposely put a dash in a keyword, and have the search engine spiders still catch that keyword as the real word without dashes?

I've done a little Googling and looking here on Stackoverflow, but everything comes down to dashes for multiple words, not really the specific thing I'm trying to figure out.

Hopefully that makes sense, I am not an expert in SEO, yet, but get the basics and have been playing, and this is just really a random question to satisfy my knowledge of playing :P

Answer 1

Search engines usually index the separate words rather than the phrase as a whole. So in your example it would index key and word but not keyword.

However, search engines are also very well-versed in synonyms so if there are enough non-hyphenated versions of the word out there, they will likely become associated.

For example, if you search for key-word, Google says "Did you mean: keyword" and shows results for keyword. Furthermore, a search for free-standing the top result is for Merriam-Webster's dictionary entry for freestanding.

Answer 2

I've answered my question...

After using google to search for this page specifically, with the following keywords (or other variations)

http://webmasters.stackexchange.com: "keyword" Cooperative "Photography"

You'll find that a little ways down the results, this page was found. The 'strong' words being "Cooperative" and "Co-op" leaving out the rest. To get a solid match, you 'must' type the spaces. So, "Pho-to-gra-phy" != Photography, it will match, as expected, if each section of the word was split by spaces or hyphens in the search query. So, to answer my question, since if google cant do it, then it's pointless to try any more.

If you have words on your site that require hyphens at the syllable's of the word, or anywhere else for any reason (OF 1 WORD SPLIT UP, NOT MULTIPLE WORDS), it will not get crawled as a keyword of the original word. This is all I needed to know.

Thanks everyone for the prompt replies.

Answer 3

Generally speaking a search engine will treat a dash / hyphen (-) as a delimiter, and so split the given words up into multiple tokens (roughly equivalent to words), for example:

nineteenth-century = [nineteenth] [century]
blue-green = [blue] [green]

Which usually means that a search for "word" would potentially match "key-word".

The reason for this becomes a little clearer once you start to look at commonly hyphenated words - most of the time^* a hyphen is used to combine words that would probably make sense separated, for example pigeon-hole vs pigeon hole. Indexing them as two separate tokens allows for a search on "green" to successfully match "blue-green", which is probably what people would expect.

Many search engines have additional mechanisms used to weight searches based on the proximity of words, so that (for example) a search for "key word" will show a match against "this is a key word", higher up in the results than a match against "the key thing is that they are all words".

Of course as Ed says - it's completely down to the implementation.

_{(*) Based on absolutely no evidence / data.}