Eldin Raigmore wrote:
> BTW thanks for your posts on that thread!
> On Mon, 11 Aug 2008 14:04:29 +0200, Benct Philip Jonsson
> <[log in to unmask]> wrote:
>> [snip]
>> How does "caste" differ from "cast" and "type" and
>> "word class" in your terminology? Instinctively
>> I'd say that "caste" may differ from "word class"
>> in that the former is semantically defined while
>> the latter is morphologically defined, but that's
>> just my mind trying to make sense of what to me is
>> an alien terminology.
> I probably mis-used the adverb "linguistically" in my original statement.
> "Caste" and "type" are (used by me above as) nouns, 

Well, they are nouns in standard English    :)

> and "cast" and "type" are (used by me above as) verbs.

'type' is, of course, also commonly used as a verb with various 
meanings. As it has been observed, English often uses the same word in 
different grammatical categories without affixing or any other modification.

'cast' was in its basic meaning verbal ("to throw or fling") and is not 
etymologically related to "caste" - the latter being derived from 
Portuguese 'casta' ("breed, race" <-- Latin adjective 'castus, casta, 
castum' 'pure"), while the former is from Old Norse 'kasta' ("to throw").

'cast' came also to be used nominally to mean "act of throwing or 
flinging" and has since come to develop a whole range of related 
meanings. But one meaning it does not have is "caste" (a social class in 
India; an exclusive social class).

> "Caste" is a linguistic term; 

Is it? I find no mention of it as such by Larry Trask (A Dictionary of 
Grammatical Terms in Linguistics) nor by David Crystal (A Dictionary of 
Linguistics and Phonetics); nor can I find it mentioned by Thom Payne 
(Describing Morphosyntax).

> it's linguistic meaning is derived from its meaning in 
> sociology (cf. "the caste system in India"). 

...which in turn is derived from its meaning denoting the social 
divisions in Indian Society.

> There are some Indian languages, 
> including some Dravidian languages, in which some word-classes (nouns? 
> verbs?) are divided into "castes"; the high-caste ones have a somewhat 
> different morphosyntax than the low-caste ones.

This seems a bit vague. Are we talking about a system in which the 
different registers found, for example, in English and, I guess, most 
languages have become formalized? Or what? Can you give examples to make 
this clearer?

> "Word-class" is a linguistic term, too; I used it almost as a synonym for "part-
> of-speech".  I do not know what the difference is, if any, between "part-of-
> speech" and "word-class".

Let's try Larry Trask first of all:
"The traditional name for *lexical category*."

"(also *part-of-speech*, *word class*) Any one of a dozen or so classes 
into which the lexical items of a language are defined by their 
morphological and syntactic behaviour, such as Noun, Verb, Adjective, 
Determiner and Preposition. Traditional grammar recognized only about 
eight parts of speech, but all current theories of grammar have found it 
necessary to increase this number by the addition of such 
non-traditional categories as Determiner, Degree Modifier and 
Complementizer. There is as yet no complete consensus as to precisely 
which lexical categories should be recognized, though the divergence of 
views is not dramatic."

"See *lexical category*."

Now David Crystal:

"The TRADITIONAL term for a GRAMMATICAL CLASS of words. the main 'parts 
of speech' recognized by most school grammarians derive from the work of 
ancient Greek and Latin grammarians, primarily the NOUN, PRONOUN, VERB, 
PARTICIPLE and others often added. Because of the inexplicitness with 
which these terms were traditionally defined (e.g. the use of unclear 
NOTIONAL criteria), and the restricted nature of their definitions 
(reflecting the characteristics of Latin or Greek), LINGUISTS tend to 
prefer such terms as WORD-class or FORM-class, where the grouping is 
based on FORMAL criteria of more UNIVERSALLY applicable kind."

Crystal does not give a specific entry for "word class", but under the 
heading WORD he does have:
"At a more specific level, *word-classes* can be established, by 
analysing the various GRAMMATICAL, SEMANTIC and PHONOLOGICAL properties 
displayed by the words in the language, and grouping them into classes 
on the basis of formal similarities (e.g. their INFLECTIONS and 

Thus, as far Trask is concerned 'part of speech' and 'word class' are 
indeed synonymous. But Crystal does make a distinction between the 
traditional 'parts of speech' of western grammars derived from 
Graeco-Latin models and 'word classes' derived from a more modern 
linguistic analysis of language.

> Parts-of-speech don't exist in every computer-programming language; COBOL 
> and related languages speak of "nouns" and "verbs", but not every computer-
> programming-language does so.

Do they exist in any in the way that they exist in natural languages? I 
managed to avoid programming in COBOL.

I remember one of my students many years back complaining about COBOL; 
he maintained that you had to write an essay to make anything happen! 
The lad got quite excited at my Prolog classes and, indeed, became quite 
proficient in the language. Good lad       :-)

But we've discussed before on this list how far so-called programming 
'languages' can really be considered language in the sense that natural 
languages are and, indeed (most?) conlangs attempt to be. One has to 
remember that so much terminology in computing is _metaphorically_ 
derived from human behavior; thus we talk about a computer's "memory" 
and we speak of programs acting "intelligently." We even talk about 
computers playing chess! They do nothing of the sort; it's merely that 
clever programmers have devised a set of instructions that make the 
machine simulate reasonably intelligent (i.e. non-random) responses.

Ben Schneiderman (Designing the User Interface) reminds us to overcome 
"the obstacle of animism" and warns that "the metaphor and terminology 
of human form can still mislead ..."

As those of us who have ever attempted compiler writing will know, a 
programming 'language' is essentially a *code* which makes life easier 
for the programmer but which is ultimately 'translated' (i.e. decoded) 
into binary digits.

> "Type" and "cast" are computer-programming-language terms.  "Variable" 
> names are "typed" (here "type" is a verb) or given a "data-type" (here "type" is 
> a noun).  

[etc. snip]

This is all very true - but I am very skeptical how useful these notions 
are when applied to human languages. Is it useful to say that English is 
weakly typed but Russian is more strongly typed? And as for casting, 
well ........

> My objection to Alex's use of the word "cast" was his apparent feeling that in 
> deciding about datatypes for computer-programming-languages, you had to 
> decide one was higher than the other in order to make a "typecasting" 
> function.

I agree, that's one objection. My other objection is whether the 
application of programming metaphors to the processes of natural 
language is either valid or helpful. My gut feeling is that thinking of 
parts of speech (or word classes) as datatypes, and of processes such 
denominalization and deverbalization as 'casting' are misleading.

> But I also disagree that "cast", in computer-programming-language 
> terminology, has anything to do with any hierarchy.
> "Caste" has to do with hierarchy; "cast" does not.

I agree 100%. The two words are of different origin and are not connected.

> "Type" is a poor match for "word-class", in analogizing between natlangs and 
> proglangs.  

I agree entirely.

> "Type" is a much better match for "gender" or "noun-class".

In view of the very wide range of meanings that 'type' has, it is IMO 
better avoided even in this context.

>> [snip]
>> In German and the Scandinavian languages the
>> adverbal derivation coincides with the neuter
>> nominative singular of the adjective (which
>> happens to be a zero morpheme in German but not in
>> Scandinavian). 

One finds this to a limited extend in ancient Greek and Latin (the 
grammar books normally say its the accusative of the neuter; but as that 
is _always_ identical to the nominative, it comes to the same thing). Is 
this an inherited IE phenomenon?

>> You are definitely on to something here -- cf.
>> what I said about semantic overlap between
>> adjectivization and genitive in Esperanto.

This has long been noted.

>> At least in inflecting languages the difference
>> would seem to be that an adjective is a derivation
>> (possibly zero-derived from a root) and as such
>> may be inflected (for case, number, gender...)
>> while a genitive is an inflection and as such not
>> further modifiable.

IIRC genitives that inflect like adjectives are found in some ancient IE 
derived languages of Asia Minor. In Classical Latin we have _cuius_ 
/kujjus/ which may behave as an invariant genitive like _eius_ , 
_huius_, _unius_ etc, or may behave as an adjective with feminine _cuia_ 
and neuter _cuium_.

So I don't think the above is universally true.


>>> This seems to be a peculiar ability of English,
>> Agreed. A function of the fact that English nouns,
>> adjectives and verbs have so little morphology.
> Maybe so.

This feature of English is also found in Chinese.

Frustra fit per plura quod potest
fieri per pauciora.
[William of Ockham]