May 29, 2006

Confusing web language with the web world

Says Adam Cohen in a New York Times article on corporate threats to the democratic ethic of the World Wide Web [print: Sunday May 28, 2006, Week In Review, p. 9]:

"The blogging phenomenon is possible because individuals can create Web sites with the World Wide Web prefix, www, that can be seen by anyone with Internet access."

The remark suggests at least two mistakes. One (relatively minor and perhaps debatable) is linguistic. The other is more technical, and concerns a false belief that is probably fairly widespread.

1. Morphologically, the www in such words as www.languagelog.com is more like a combining form (see chapter 19 of The Cambridge Grammar) than a prefix. While there is great freedom in creating URLs, the most usual practice follows the convention of having a name for a server or a department followed by a name for the site or the company followed by a suffix indicating either a type of domain (commercial, non-profit, educational) or a specific country, and only the last element has to be picked from a predetermined list and has a semantics that the owner does not control. Thus we find email.sjsu.edu for the email server at San Jose State University; ftp.debian.org for the ftp (file transfer protocol) of the Debian Linux organization; ling.ucsc.edu for the departmental server of the Department of Linguistics at the University of California, Santa Cruz (where I work when I'm not at Language Log Plaza or away on sabbatical); and so on. The www portion of the millions of URLs that have it is really a server name: www.webster.com is the URL for the main web server (or server array) of the Merriam-Webster Corporation in the commercial arena, matching the typical pattern of server + site + domain. To the linguist's eye, the www is more like an initial combining form in a multi-component word (like the geo in geophysical or the psycho of psycholinguistic) than like a sense-modifying suffix (un- in unhelpful or pre- in prenatal).

2. You don't have to have www as the first component of the URL in order to have a blog or any other kind of web site. What you have to have is an Internet-connected machine running http server software. You can see this immediately from any of the livejournal blogs (like clunis.livejournal.com, to cite a random example), or from the fact that ling.ucsc.edu and lib.harvard.edu are web servers, etc. Cohen appears to believe that in order to have a blog, or perhaps in order to have a web site at all, you have to create a site with www as the first element of the URL. You don't. Doing so is purely a very widespread linguistic convention. One more case of confusing your language with your world.

Posted by Geoffrey K. Pullum at May 29, 2006 10:46 AM