Re: Nominal and Adjectival Predicates
- On Sun, 16 Jun 2013 11:07:41 +1200, James Kane <kanejam@...> wrote:
>Is there some inherent reason that this is a weird way to do it?Well, I don't know how we can hope to know what's inherent, aside from by looking at what large samples of languages do (and trying to see what features of the semantics they're sensitive to).
>Looking around other natlangs, it seems most of them leave the
>predicate-y part in the nominative, and so does Esperanto, which I
>always found counter-intuitive as the rest of the language is very
>strict in marking the accusative.
>The explanation was always that they were equal, and neither was doing
>anything to the other; but my natural instinct (my L1 is English) is
>that, in something like 'he is John', John comes after the verb and
>thus is a direct object.
>I just want to know if I'm being a noob.
A first blush attempt at that leads us to the theory of thematic relations, where the prototypical transitive object is a patient, something that undergoes (i.e. has no volition or control in) an action, and thereby changes its state. These are not features of the, er, right-hand arguments of predicates "is identical to" and "is an instance of", and that speaks against this argument being in the accusative. Note that there are various other kinds of two-argument predicates which also don't have a patient, and may not get the accusative either: e.g. verbs in non-basic voices, like passives, whose demoted object is not a simple accusative in any case I can think of offhand; verbs of motion, which also typically don't get the accusative, though there are phenomena like the Latin "accusative of place to which"; and verbs which take a _theme_ instead, an argument that doesn't suffer a state change -- but most (if not all?) languages disregard the difference between patients and themes in basic case assignment: http://linguistlist.org/issues/4/4-366.html .
Indeed, to judge by the frequency with which non-verbal strategies (or defective verbs or ...) are used for them, the relationships of being identical with, or being an instance of, aren't actions at all. If accusative case is assigned by verbs to their objects but a given construction doesn't even hàve a verb, then it's not likely to use an accusative.
As for Esperanto, the smart money is that it avoids the accusative in copular clauses because Greek and Latin did, not because Zamenhof put a lot of thought into thematic relations or anything.
English intuitions may lead a person astray here, 'cause in English there are good reasons to say the marked member of the opposition nominative vs. accusative is the nominative, and this is also not the case for most languages with that opposition.
On Sat, 15 Jun 2013 22:17:20 -0400, Rich Harrison <rick@...> wrote:
>Probably going off on a tangent here, but does it help to eliminate the copula? If you don't have a vague be/am/is thing in your language, then you only have subject-verb-object statements and of course it makes sense to put the "predicate" in the accusative:
>let's say "seteni" is a transitive verb meaning "to be a member of the set of"
>John setenas virojn. John is a man.
>Rich rolhavis ondiston. Rich held the office of (played the role of) wave-maker.
I think the copula is a red herring here. A zero-copula language, e.g., would be subject to the same questions of case assignment as one with a copular verb. And if your language was encountered in the wild, your one example doesn't give any particular reason that "setenas" shouldn't be called a copula with weird case and number selection, rather than given this odd gloss 'be a member of the set of'.
If you really wanted to go serious about set-theory based semantics, there's much more to do than change "John is a man" type sentences. (Zach Weaver's Davin, presented at LCC5, is a taste of how this might turn out: conlang.org/cms/wp-content/uploads/zweaver_lcc5_slides.pdf )
(E-prime was a fine constrained writing exercise and therefore is a fine prompt for conlang ideas, and it's certainly true that the IE copula has lots of semantic functions, but I quite disagree with its identification of that fact as a problematic locus of "vagueness". It's one hundred percent possible to be just as (overtly or covertly) subjective or misleading or conflict-inducing or whatever without touching the English copula as it is with it; in inexpert hands the principal effect of missing it is just stiltedness.)
- With this post, I finally do away with my backlog and am back on track :P .
On 16 June 2013 15:43, Jörg Rhiemeier <joerg_rhiemeier@...> wrote:
> Yes. A copula complement simply is not a direct object, and
> this is the reason why IE languages (and many other languages)
> do not put it in the accusative case. Instead, the complement
> appears in the "base form", which in IE languages is the
I just realised that it is exactly what Japanese does! Predicate
constructions in Japanese always confused me, but now I understand why!
As a quick primer, in Japanese both the subject and the object of a verb
are marked by a particle (_ga_ for the subject, and _o_ for the object.
Both can be overwritten by the topic marker _wa_, though). However, nominal
predicates (which in Japanese are mostly nouns and some adjectives) do not
take either. Instead, they appear naked, and are only followed by the
copula (which can be omitted, at least in women's familiar speech
patterns). For instance, the simple sentence "Takuto is a man" becomes in
Japanese: _Takuto wa otoko da_, with _da_ being the familiar copula (it
becomes _desu_ in polite speech, or can be omitted).
So basically that's what's happening here: the nominal predicate takes the
base form, which in Japanese happens to be neither the subject nor the
object form, but the stem alone.
> What regards Arabic, one can argue that the accusative is
> actually the least marked case, and the nominative a marked
> one. This pattern (often called "nominative-absolutive")
> seems to be common in languages of the Afrasian family.
I wonder whether something similar might be happening in Moten, which would
explain why "to be" is treated as a normal transitive verb in that language
(and thus takes predicates in the accusative case). It's true that the
nominative is actually semantically marked in Moten (although it is
morphologically the base form): when the subject of a transitive verb is in
the nominative case, it indicates volition, i.e. that the subject does
whatever is indicated by the verb willingly and on purpose. To mark
non-volition, one has to put the subject in the instrumental. This is true
even of "to be", which with a nominative subject indicates that the subject
is willingly "being" something or someone.
Things get a bit more complicated quite quickly though:
- this pattern (which I call the "split nominative") only exists for
transitive verbs. Intransitive verbs take a nominative subject whether
there is volition or not involved.
- this pattern only works well for animate concepts. For inanimates,
volition is not an option, and according to the rule above that should mean
that an inanimate subject of a transitive verb should always be in the
instrumental. And indeed, in high registers of the language the only known
native speaker of Moten does just that. But in more familiar registers, he
tends to slip and use the nominative instead, even though there cannot be
any volition involved. It's a syllable shorter after all, and as long as
context makes clear that the subject is inanimate (it's a semantic feature
in Moten, not a syntactic one), there's no confusion possible.
I'm still not quite sure what to make of this pattern, but that's how Moten
works. It looks a bit like this "nominative-absolutive" pattern, except
with a split for volition, and with the nominative case being the least
> > Yes, but, as I observed above, you cannot promote John to
> > the subject of an equivalent passive: *John is been by him!
> Indeed not!
Moten has no passive voice, so this issue is moot here.
On 17 June 2013 19:49, H. S. Teoh <hsteoh@...> wrote:
> Yeah, this is one of the neat things about Russian: in verbs of being
> *not* in the present tense, the predicate is in the instrumental case:
> Я бы-л врач-ом.
> 1SG.NOM be-PAST.SG.MASC doctor-INSTR
> I was a doctor.
> The instrumental also occurs with verbs of becoming:
> Он ста-л врач-ом.
> He.NOM become-PAST.SG.MASC doctor-INSTR
> He became a doctor.
> It is ungrammatical to use the nominative or accusative in these cases.
> It seems to me that the instrumental case here is being used in a
> stative sense, or a transition into a state, as opposed to a mere simple
> direct object.
Japanese is similar here. The verb "to become" in Japanese (_naru_) is
actually intransitive, and what you become takes the particle _ni_, which
indicates various things like location (at), destination (to) but also the
person to whom something is given (to). In Japanese you become *to*
something, emphasising the process rather than the final state.