Saeed,

There are many ways in which one might classify a text, and you have 
made a common distinction, which I will summarize using other words:

A) classification by form (letter, poem, novel, play, etc.)

B) classification by subject (about shipping, about biology, about 
literary criticism, etc.)

What you call (A) and (B) varies considerably in different communities. 
  So while you were using "genre" to mean (B), the term "genre" is used 
by library cataloguers to refer to (A).

The TEI does not make this particular distinction, or at least does not 
prioritize these two types of classifications over other types of 
classifications.  I think it's legitimate to use <textClass> for both 
(A) and (B) because the definition of <textClass> is quite broad 
("information which describes the nature or topic of a text"), and while 
the examples (the three bullet points) given in section 2.4.3 all relate 
to (B), I think that (A) deals with the "nature" of a text.

You can document which controlled vocabulary you used in your text 
classification by using the scheme= attribute, so a user can figure out 
whether the particular text classification is of type (A) or (B) if this 
distinction is important to that user.

--Kevin

On 5/10/12 8:07 AM, Saeed Sarrafzadeh wrote:
> Hello all
> Thanks to Martin and James for the recommendation. As the<textClass>  element description noted, it is for the classification of text. Although it said that this element: groups information which describes the nature or topic of a text, but don't you think that the nature of a computer file (which may only use the teiHeader to encode the metadata for a file such as an audio recording, an image, or a video) should not be distinguished in some ways with its classification due to its topic? I think that the categorisation of a digital item into (book, audio, video, thesis, etc. or a multilevel categorisation such as: text [:>book, thesis, article, document, ...], audio, video, ...) is somewhat different from genre or topic classification (which may be done using different approaches) of digital objects. So, what I find suitable to distinguish that is @scheme attribute especially in<catRef>. Is it enough to not confuse between the nature of the object and other classification a
 pproaches
?
> Best wishes, Saeed