Re: Linguistic term for ease of changing word-class (was: 'out-' affix in conlangs?)

Plura commentaria in uno voluta!

Från: Alex Fink <000024@...>

 >
 > On Fri, 8 Aug 2008 20:14:05 +0200, M. Czapp
 > <0zu149@...> wrote:
 >
 >>>*HIJACK: Is there a better linguistic term for
 >>>the ease with which you can change whether a
 >>>word is a noun, adjective or a verb? The best
 >>>example for weak typing (easy/implicite
 >>>changes) might be Esperanto, German is of the
 >>>languages I know the one with the most
 >>>problematic 'typecasts.
 >
 > Talk of casting tends to make me leery, for the
 > way it seems to make the background assumption
 > that given any two data types there should be
 > exactly one function between them of such
 > paramouncy that it makes sense to elevate it
 > above all others and crown it the Cast between
 > those two types. For some type-pairs I buy this
 > (smaller to larger floating point types, say);
 > mostly not.

Talk of too strict disambiguity in language, at
least naturalistic (qua natural human language-
like) language as opposed to computer language or
the most ivory-towered loglang, makes me leery.
Ambiguity, fuzziness and under-specification are
as much a feature of natural language as is
redundancy! The reason of course is that any
ambiguity at some level (lexical, morphological or
syntactic) will be resolved by context at another
level, or as the last resort by the recipient's
knowledge of the world. Moreover the canonical
communicative situation is not reading a book but
a conversation (be it face-to-face or by email)
and in a conversation the interlocutors can always
ask/clarify if they don't or seem not to get the
intended meaning.

This said vague casting in an interlingua (be it
an auxlang or a translation interlingua) which may
be primarily used for non-conversational written
communication is probably a bad thing.

 > This seems as much to be true of lexical
 > categories as computerish data types. So in
 > Esperanto the "casts" to adjectives are in fact
 > ambiguous between roughly "pertaining to X" and
 > "having lots of X" or perhaps other things yet:
 > _suna_ 'solar' or 'sunny'; _denta_ 'dental' or
 > 'toothy'. Not to mention the whole _broso_ vs.
 > _kombo, kombilo_ thing ('brush'; 'act of
 > combing', 'comb'). Basically, you simply need to
 > specify more for a derivational operation than
 > e.g. "converts nouns to verbs".

The best thing is to do as natlangs do: allow the
sender to be either vague or precise as they see
fit. The problem with natlangs is of course that
the "types" for which they offer precision are
often randomly 'selected' and not nearly
'exhaustive' (note the scare-quotes!) Also it
should be noted that natlangs not *always* offer
vagueness, such as the possibility not to choose
between "dental" and "toothy", unless it be the
genitive "of teeth". In the days when I was
actually reading and writing Esperanto (some 20
years ago now) i often felt that adjectivization
('casting to adjective') and the _de_ genitive
often overlapped semantically. My hunch is that
this is less the case in natlangs, i.e. natlang
adjectives are usually stronger 'typed', as
exemplified by "dental" and "toothy" as opposed to
"of teeth" in English (apart from the fact that
"of teeth" also can mean 'made of teeth', but that
is because "of" has two meanings or, as some would
put it, there are two different "of"s.

 > Anyway, to get back to your original question, I
 > don't know of any such terminology pertaining to
 > changing word class in particular. One could
 > just talk of the general propensity for
 > derivation -- some langs might be rich in
 > productive derivational morphology, others poor.

Why should there be a single (presumably Graeco-
Latinate) term for 'word class changing' any more
than there is for 'valency changing'? -- although
there is "applicative" for 'focus changing'!

 > At the extreme, I suppose, you might have a
 > language where one of these categories (e.g.
 > verbs, or adjectives, I think I've read of cases
 > of both) is a _closed class_, i.e. you can
 > simply never make any more of them, whether by
 > derivation or borrowing or some other means.

Basque comes to mind as a language where finite
verbs are a closed class. I guess one could
analyse Esperanto so that there is only one noun
_o_ and one adjective _a_ which compound with
different verb roots to form the syntactic
equivalent of nouns and adjectives in most
natlangs, or rather such a usage would be possible
under the minimalistic codified grammar of
Esperanto, but does not agree with actual usage,
which is much more influenced by (European)
natlangs, so that in practice some roots are
nominal or adjectival rather than verbal.

Från: Jim Henry <jimhenry1973@...>

 > On Fri, Aug 8, 2008 at 5:23 PM, Alex Fink
 > <000024@...> wrote:
 >
 >>> On Fri, 8 Aug 2008 20:14:05 +0200, M. Czapp
 >>> <0zu149@...> wrote:
 >
 >>>>>*HIJACK: Is there a better linguistic term
 >>>>>for the ease with which you can change
 >>>>>whether a word is a noun, adjective or a
 >>>>>verb? The best example for weak typing
 >>>>>(easy/implicite changes) might be Esperanto,
 >>>>>German is of the languages I know the one
 >>>>>with the most problematic 'typecasts.
 >
 >>> This seems as much to be true of lexical
 >>> categories as computerish data types. So in
 >>> Esperanto the "casts" to adjectives are in
 >>> fact ambiguous between roughly "pertaining to
 >>> X" and "having lots of X" or perhaps other
 >>> things yet: _suna_ 'solar' or 'sunny'; _denta_
 >>> 'dental' or 'toothy'.
 >
 > Or "made of X" or "resembling, savoring of
 > X" (though there's also the more specific
 > "-eca" for that), or "for the benefit of X"
 > or "suitable for X" or "originating from
 > X"... Issues like these were why I came up
 > with the set of adjective-deriving suffixes
 > I did for gzb.

Would you mind to give a list of those suffixes?

 > Though in practice E-o adjectives derived from
 > substantial roots are not often ambiguous in
 > context, I think. Still, for practical use I
 > prefer a language where the derivations are a
 > little too vague and occasionally ambiguous to
 > one where you can't derive words easily and you
 > have to memorize apparently unrelated words for
 > closely related concepts.

Yes. If I were to design an interlingua I'd
definitely provide for both loose and strict
'typecasts' to be used at the writer's/speaker's
discretion.

 > I think the derivations in the other direction,
 > from adjective to noun or verb to noun, are much
 > less problematic (at least w.r.t. Esperanto);
 > you can be pretty confident that the basic
 > meaning of such a nominalization will be
 > "quality X in the abstract, x-ness" or "an
 > instance/act of doing X" -- though in some cases
 > 120 years of unregulated usage have given some
 > such nominalizations additional conventional
 > meanings.

Yes, that's my feeling to. Cf. what I said above
about adjectivization and genitive overlapping
semantically in Esperanto.


 >>> Not to mention the whole _broso_ vs. _kombo,
 >>> kombilo_ thing ('brush'; 'act of combing',
 >>> 'comb'). Basically, you simply need to specify
 >>> more for a derivational operation than e.g.
 >>> "converts nouns to verbs".
 >
 > Indeed it helps reduce ambiguity to do so,
 > though given how well Esperanto (and even Toki
 > Pona) work in practice, I'd hesitate to say you
 > *need* to do so. I would strongly recommend
 > doing so in an auxlang or engelang; in an alien
 > lang or more or less naturalistic artlang, do
 > whatever you like.

You don't *need*, but you probably should
provide the possibility, which both Esperanto
and most natlangs provide only in a random
fashion. Ido does this much more extensively but
still randomly.

 >>> Anyway, to get back to your original question,
 >>> I don't know of any such terminology
 >>> pertaining to changing word class in
 >>> particular. One could just talk of the general
 >>> propensity for derivation -- some langs might
 >>> be rich in productive derivational morphology,
 >>> others poor.
 >
 > And of those that have productive derivational
 > morphology, some tend strongly toward zero-
 > derivation (English and Toki Pona, for
 > instance), some mark the derivations vaguely
 > (most of the derivational affixes being
 > semantically broad; Volapuk is more extreme
 > about this than Esperanto, and my understanding
 > is that Ido's derivation system is supposedly
 > one of the areas where it improves on Esperanto,
 > but it's still vaguer than an ideal engelang)
 > and some do so with rather more semantic
 > precision (Ithkuil; my gzb, I hope).

Ido at least does the right thing in that it
allows the choice between vagueness and precision,
while engelangs tend to offer only precision. This
may not be a very big problem if you are composing
an original text, but what if you're translating
from a natlang where the original expression is
vague? You'd have to choose a precise expression
which may not be 100 per cent justified by the
context in the original, but the greatest
objection is that vagueness is often stylistically
and pragmatically desirable in human comunication
just as much as precision is.

Från: Eldin Raigmore <eldin_raigmore@...>


 > Linguistically I think of those as "castes" or
 > as "types", not as "casts". I refer to the
 > affixes (or other operations) which change part-of-
 > speech as "typecasting" affixes or operations;
 > in programming I refer to a function whose
 > purpose is to change data-type as a
 > "typecasting" function.

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.

 >>>This seems as much to be true of lexical
 >>>categories as computerish data types. So in
 >>>Esperanto the "casts" to adjectives are in fact
 >>>ambiguous between roughly "pertaining to X" and
 >>>"having lots of X" or perhaps other things yet:
 >>>_suna_ 'solar' or 'sunny'; _denta_ 'dental' or
 >>>'toothy'. Not to mention the whole _broso_ vs.
 >>>_kombo, kombilo_ thing ('brush'; 'act of
 >>>combing', 'comb'). Basically, you simply need
 >>>to specify more for a derivational operation
 >>>than e.g. "converts nouns to verbs".
 >
 > Yes. In strongly-typed computer languages there
 > is frequently an operation with no other
 > function than to change the data-type. In
 > natlangs, though, there are usually several
 > nouns pertaining to each verb, so several ways
 > of nominalizing it (action nominalization, agent
 > nominalization, patient nominalization, location
 > nominalization, time nominalization). There are
 > frequently also more than one ways to
 > adjectivize it (passive participle, active
 > participle, maybe realis vs irrealis participle,
 > maybe past vs future vs present participle,
 > etc.) And there may be several verbs associated
 > with a given noun, too; to use an N on, to
 > change into an N, to treat as if it were an N,
 > to give an N to, to take an N from, ....
 > Similarly, probably, for most productive
 > operations that can change one part- of-speech
 > into another. (The resulting word-class is
 > always an open one, and usually a large one; the
 > starting word-class is also usually a large,
 > open class. Other than that the only limit seems
 > to be there is frequently only one way to change
 > an adjective into an adverb (if the language has
 > both as large open word-classes and they are
 > different parts-of-speech) and may be no way to
 > change an adverb into some one or another of the
 > other word-classes.) All such operations are
 > likelier to be "derivations" than "inflections",
 > because one of the differences between
 > "derivation" and "inflection" is that
 > "inflection" usually leaves the word in the same
 > class while "derivation" frequently does not.

Has someone made a list of such types, whether
actually distinguished in natlangs or semantically
distinguishable or logically possible?


 > I know there are languages with no class of
 > adverbs distinct from their class of adjectives;

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). What about Dutch. English is
actually quite strange in having made the Old
English adverbial ending _-lice_ (actually the
dative of the adjectival ending _-lic_! :-) not
only very productive but practically required.

 > but aren't many "semantic cases" (that is, cases
 > other than "syntactic cases", that show
 > something other than the "grammatical relations"
 > of Subject, Object, or Indirect Object) also
 > "adverbial cases"? Isn't a noun in a case other
 > than Nominative, Accusative, Dative, or
 > Genitive, essentially an adverb? So, the
 > "changing of a noun into an adverb" is likely to
 > be fairly "easy" -- highly productive -- in most
 > languages with a robust case system, right? And
 > Genitive, in those languages that have one, is
 > essentially a way of changing a noun into an
 > adjective, isn't it?


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

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.

Från: Eldin Raigmore <eldin_raigmore@...>

Datum: Sun, 10 Aug
 > 2008 13:55:51 -0400 Till:
 > CONLANG@listserv.brown.edu
 >
 > BTW I once read a scholarly book by a true
 > professional linguist on the ease of forming new
 > words of one class from words of another class.
 > I forget the title of the book and the name of
 > the book's author (though I still have it --
 > somewhere -- ...)
 >
 > However, I remember it said (among other things)
 > that in most languages (though I recall only the
 > English examples) it was particularly easy to
 > make a verb from another word-class, especially
 > from a noun. (example: "to pipe someone aboard".
 > "Pipe" was a noun; it can however be used as a
 > verb, whether transitive or intransitive, with
 > no morphology having been done on it at all -- a
 > kind of "zero-derivation".)
 >
 > I can quote two other writers on the subject,
 > however; "Verbing weirds language." (Calvin &
 > Hobbes).

Actually only Calvin. This was one of the
instances when Hobbes was just a bemused
spectator.

 >
 > Note that the noun "verb" becomes (via "zero-
 > derivation") the verb "verb" which is
 > morphologically altered to the action-nominal
 > "verbing". Note that the adjective "weird" is
 > zero-derived into the verb "weird".
 >
 > Also note that originally -- back in the mists
 > of English etymology -- "weird" was a noun. But
 > in more modern English it has become an
 > adjective.

What really bemuses me is that Calvin didn't
analyse "weird" as the past participle of a verb
"weir", illiterate as he presumably is...


Från: ROGER MILLS <rfmilly@...>

 > Eldin Raigmore wrote:
 >>
 >> BTW I once read a scholarly book by a true
 >> professional linguist on the ease of forming
 >> new words of one class from words of another
 >> class. I forget the title of the book and the
 >> name of the book's author (though I still have
 >> it -- somewhere -- ...)
 >
 > Even after almost 20 years in this house, I
 > still have boxes of books in storage. Aargh.

I know the feeling. Nine years after my son
started to tear books out of the shelves and tear
them apart so that we had to move them out of
harm's way everything is still in a mess. Not to
mention all the library loans that I never made
proper notes off/from!

 >> However, I remember it said (among other
 >> things) that in most languages...
 >
 > I'd question that "most"!!

Me too. Even AFMOC only two out of an odd dozen
practice zero derivation.

 >> ...(though I recall only the English examples)
 >> it was particularly easy to make a verb from
 >> another word-class, especially from a noun.
 >> (example: "to pipe someone aboard". "Pipe" was
 >> a noun; it can however be used as a verb,
 >> whether transitive or intransitive, with no
 >> morphology having been done on it at all -- a
 >> kind of "zero-derivation".)
 >
 > 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.

 > and I think it can be a pretty random process.

How so. Purist notwithstanding pretty much any
word in English can be zero-derived into another
word-class, especially verbed.

 > There are of course cases like (vb.)
 > projéct, (n.) próject, and genuine cases of
 > zero-derivation like love (n or vb.). Your
 > pipe example, of course, refers to the little
 > whistle-thingy used in the Navy (I think
 > Brits can use it for bagpiping too) so is a
 > highly specific case.
 >
 > But yes, English can do this easily. Note also
 > "welcome" as interjection, noun, verb or
 > adjective.
 >
 > William Safire, in his NYT column, likes to
 > point out horrors like "surveillance : to
 > surveille" or "liaison : to liase", not quite in
 > the same category but close. Also, purists
 > dislike "fínance" replacing the old distinction
 > finánce : fínance (not sure which is the
 > verb/noun, though it was drilled into me once
 > upon a time :-(( ).

I loathe such purist dislike. Backformation and
zero-derivation are signs that English has made a
word its own, detached from its Latin, Greek,
Romance or whatever origin. Unsophisticated
speakers do it all the time, while the sophists
stand there as the guardians of words' allegiance
to foreign, often dead languages. I feel much the
same WRT borrowing of English verbs into Swedish
FWIW. There was a time when purists demanded that
Latin words borrowed into German, Icelandic or
Swedish (then still a case language) should be
inflected with Latin endings (e.g. "Meinen Jesum
lasse ich nicht"). What Mr Safire and others are
doing is the same thing applied to derivation.

 > Indonesian can have zero-derivation in
 > colloquial speech-- surat can mean 'to write'
 > or 'a letter', cinta 'love' can be noun or
 > verb; but correctly, when used as verbs there
 > ought to be a verbal prefix... I really suspect
 > not many languages can do this as readily as
 > English does. (German/Dutch and Romance lgs.
 > come to mind).

Swedish certainly can't. In order to use a word as
a verb you have to put a verbal ending on it, and
only the closed class of strong verbs have any
zero endings within their paradigm.

 > Sort of OT, but relevant to the question about
 > Basque verbs-- IIRC the verbs that have their
 > own synthetic conjugation (i.e. without the
 > usual person+tense aux.) are a small and closed
 > class, I think mostly intransitive. There's
 > another productive (I think) class formed from
 > NOUN + 'to do/make' (egin?); one that has stuck
 > in my mind is 'to sneeze' (sneeze + egin? +
 > aux). (My Basque grammar is one of the books in
 > storage.......)

The Semitic component of Yiddish vocabulary works
similarly IIRC, using a Semitic verbal noun + a
Germanic verb like 'be, have, do' rather than
tacking Germanic endings to Semitic verbs.

/BP 8^)>
--
Benct Philip Jonsson -- melroch atte melroch dotte se
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  "C'est en vain que nos Josués littéraires crient
  à la langue de s'arrêter; les langues ni le soleil
  ne s'arrêtent plus. Le jour où elles se *fixent*,
  c'est qu'elles meurent."           (Victor Hugo)

Replies