Re: Cleaning House

Maciej Stachowiak:
> Are the semantics defined solely by the specification  
> (Prescriptivism) or informed by actual use (Descriptivism)?

We could employ descriptivism for classes with accepted semantics  
based upon extensive observation which values the |class| attribute  
gets in the wild and why.

We could also employ limited descriptivism for the redefinition, i.e.  
the refinement of previous definitions, of existing element types.

Where people used |i| where they should have used |em| (or |var|, | 
cite|, |dfn|) they did not use the language correctly. They should  
not be pardoned in retrospective by weakening the distinction. For  
all the other cases where phrases, but not sections, have been  
enclosed by |i| tags we may try to find out the prevailing reasons,  
although most have already been named probably, and invent new  
element types for them or assign these semantics to |i| (expressis  
verbis). The second option requires less work on all sides, but/ 
because |i| would remain a wide-purpose element type.

> For human languages, linguists generally take the Descriptivist  
> approach. This turns out to be a more productive way to interpret  
> artifacts in human languages such as English.

Computer languages tend to be prescribed in a standard, but markup  
languages are a border case possibly.

Received on Sunday, 6 May 2007 20:24:13 UTC