zero-marked verbs

Tom Givon tgivon at OREGON.UOREGON.EDU
Wed Jan 22 01:46:18 UTC 2003



Dear JW--

Now we are talking. But, I think, about a different topic. In clause-chains in
discourse in an SVO language (like English), the first clause in the chain is most
typically fully marked with all the finite trimmings. Most of the chain-medial
clauses dispense with TAM marking, or use an invariant (less finite) "narrative"
marker that simply indicate "same as in the previous clause". Sort of like zero
anaphora (or anaphoric pronouns). [In OV languages the tendency is to mark the
last clause in the chain more fully. But still chain-medial clauses tend to be
less marked].

In fact, you can show this in English too, but only up to a point. Notice how the
auxiliary TAM marker in English--just like the more-marked subject--do not repeat
chain-medially:

  a. She was sitting there, eating dinner and minding her business, thinking deep
thought....
   b. She had come earlier, looked around and decided to....
   c. She will come tomorrow, check the place out and then make her move...

In English (alas?) one cannot dump the more obligatory -ing (a) and -en (b) in
English. But in other languages clitic TAM morphology can be dumped or
neutralized. In Swahili almost all  major TAM markers can be replaced with the
"narrative" -ka- in chain-medial clauses. But in Akan, two of the main TAM markers
cannot be dumped in chain-medial contexts. The other two can be replaced with an
invariant "narrative" marker, however.  The general principle is amazingly
transparent (universal?), but the gory little details exhibit a lot of
language-specific (and construction-specific) quirks. Ultimately, most of these
quirks have cogent diachronic explanations, altho in many cases the evidence is
not available any more.

Best,  TG

===================

jaw300t wrote:

> Joan & Tom
> I guess I was not totally aware that "isolating" languages are slow to
> grammaticalize.  But Kuche is not an isolating language--it is similar to
> Bantu, with a nearly identical noun class system and verbs that may have as
> many as 4 or 5 prefixes.  That is, if they should be called prefixes when they
> are not obligatory.  They are definitely phonologically bound to the beginning
> of the verbs, but their order is not absolutely fixed (although generally
> predictable).  Neither are they serial verbs--there's a distinct difference
> between the TAM prefixes and serial verbs.  There are a few auxiliary
> constructions as well.  Elicitation for various tenses/aspects/moods
> invariably brings out the overtly marked form, but those forms are seldom used
> in discourse.
>
> But then, I guess that means that a transition to written literature may not
> be the pressure that moves a language towards obligatory TAM markers either.
> Chinese has enjoyed written literature for centuries.  JWilson
> >===== Original Message From Tom Givon <tgivon at OREGON.UOREGON.EDU> =====
> >Dear Joan,
> >
> >Very nice discussion. A student of mine, Mark Post, is working on a
> comparison
> >between Thai and Mandarin Chinese. Superficially, they are supposedly very
> similar,
> >both with with serial-verb clauses of very similar type & functions. But
> while
> >serial construction (and others) in Mandarin grammaticalize copiously, in
> Thai the
> >very same S-V constructions, using the very same lexical verbs (and
> performing the
> >samediscourse functions) somehow don't undrego the last phases of
> >grammaticalization. Mark is testing a hypothesis about lexical replacement in
> >Mandarin, and the steady--indeed profuse--creation of compounds. In Mandarin,
> this
> >apparently leads to repeating waves of obsolescence of simplex lexical verbs
> (or,
> >at least, the obsolescence of the old lexical senses of those verbs). So that
> only
> >the grammaticalized senses survive. Mark  is doing comparative
> frequency-counts on
> >this, as one of his predictions, using Pear Story texts.
> >
> >I am not sure this will explain the broader typological differences, tho.
> Virtually
> >all so-called "isolating" languages (I've always dloathed the term...) are
> >serial-verb languages (the only exception are true Plantation pidgins...).
> And I
> >think there are several factors that conspire against grammaticalization in
> this
> >typology. One factor that needs to be considered is the frequency of
> >operator-operand adjacency.  Serial-verb languages thend to have a much lower
> >probability of this in general, particularly the adjacency of modal/aspectual
> verbs
> >to their complement verbs. I discuss this in the GR chapter (#5) of  my
> >"Functionalism & Grammar" (1995), in terms of the two major typological
> strategies
> >for clause-union.
> >
> >One way or another, the explanations we are likely to discover, are most
> likely to
> >involve multiple factors and their dynamic interaction(s). And what we need
> to
> >discover is not labels, but rather highly specific performance mechanisms
> that
> >motivate diachronic change. If all we can say about "isolating languages" is
> that
> >grammaticalization is retarded there, we have not explained much, yet. Best,
> TG
> >====================
> >
> >Joan Bybee wrote:
> >
> >> Dear Janet,
> >>
> >> I have also worried about what makes a category become obligatory in one
> >> language while a similar category may remain optional in another. Based on
> >> Garcia and van Putte's discussion (1989, 'Forms are silver, nothing is
> >> gold' Folia Linguistica Historica 8, 365-384) of the inferencing strategies
> >> that lead to the infusion of meaning in zeroes, I have proposed that
> >> different cultures utilize different inferencing strategies in interpreting
> >> discourse (in my paper in Bybee, Haiman and Thompson. 1997.  _Essays on
> >> Language Function and Language Type_ Benjamins). The absence of obligatory
> >> categories (say, in isolating languages) leaves all inferences open
> >> throughout a stretch of discourse. Obligatory categories close down certain
> >> options right away.
> >> Of course, it is also necessary for the overtly marked categories to gain a
> >> certain level of frequency and redundancy before the inference can be made
> >> that the zero signals the opposite of the overt member.
> >>
> >> Unfortunately I have found no way to test this hypothesis, as it very
> >> difficult to know what inferences are being made in an ongoing discourse.
> >> However, if my hypothesis is correct, then it would suggest that discourse
> >> strategies may be very stable over time explaining why languages without
> >> inflection, that is, isolating languages, tend to stay isolating over time,
> >> despite ongoing grammaticization. As we argued in Bybee, Perkins and
> >> Pagliuca 1994, some languages simply do not carry grammaticization as far
> >> as others do.
> >>
> >> Whether this is really a hypothesis or just a hunch, I hope you find it
> >> stimulating.
> >>
> >> Joan Bybee
> >>
> >> At 09:15 AM 1/21/03 -0600, jaw300t wrote:
> >> >Colleagues,
> >> >For the past couple of years I've been following a certain unmarked verb
> form
> >> >in Kuche (also known as Rukuba, listed in the Ethnologue as Che), a
> language
> >> >of Nigeria.  I had the feeling it ought to be the simple present tense.
> In
> >> >elicitation sessions, informants never gave this form in response to an
> >> >English simple present tense sentence, although in conversation it would
> often
> >> >be used that way.  In discourse, it is the most common verb form of all,
> >> >seldom used as a simple present tense verb; there are about 20 different
> verb
> >> >markings that can be used, but the unmarked form is used in many texts
> more
> >> >than all the others combined.
> >> >
> >> >Fleischman's "Tense & Narrativity" and Bybee, Perkins, & Pagliuca's "The
> >> >Evolution of Grammar" opened up some possible explanations.  But what
> really
> >> >crystallized it for me was something in an earlier article by Bybee ("The
> >> >Grammaticization of Zero" in a volume edited by Pagliuca).  She says,
> "When a
> >> >grammaticizing OVERT morpheme becomes obligatory, it may happen that other
> >> >meanings within the same functional domain, which previously had no
> >> >grammatical expression, come to be expressed by a meaningful zero."
> >> >
> >> >It seems Kuche is at a point in its history where overt tense/aspect
> morphemes
> >> >are NOT obligatory.  What happens to the unmarked form, then, is that it
> has
> >> >NO SPECIFIC meaning--it means just about anything, depending on context.
> >> >Overt tense/aspect markers are used at the beginning of a discourse, and
> those
> >> >tense/aspect interpretations hold over long stretches of discourse.  This
> is
> >> >much like Longacre's "consecutive" tense, except that this unmarked form
> is
> >> >remarkably flexible.  It can even be used after an imperative with
> imperative
> >> >force--a second imperative verb would be even less marked, because the
> >> >unmarked form IS marked for subject agreement, while the imperative is
> not.
> >> >
> >> >I suppose Kuche is not unique among non-written languages in having a
> >> >flexible, relatively unmarked form.  I don't see them mentioned in Bybee,
> >> >Perkins, & Pagliuca.  Then, again, I suppose it's a matter of degree--in
> some
> >> >languages, the interpretation assigned to a zero-marked form might be a
> little
> >> >flexible, and in other languages, the interpretation of such a form might
> be
> >> >quite specific.
> >> >
> >> >What is it in a language community that causes OVERT TAM marking to become
> >> >obligatory?  Is it the transition to written literature?  Fleischman says
> >> >(quoting somebody else, if I remember correctly), that "The meaning of
> written
> >> >literature is in the text; the meaning of oral literature is in the
> context.)
> >> >
> >> >Janet Wilson
> >>
> >> Joan Bybee jbybee at unm.edu               phone:  505-277-3827
> >> Department of Linguistics               fax:    505-277-6355
> >> Humanities 526
> >> University of New Mexico
> >> Albuquerque, NM 87131-1196
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/funknet/attachments/20030121/1e7a3c6b/attachment.html>


More information about the Funknet mailing list