[developers] MOD value in predicative only adjectives

Emily M. Bender ebender at uw.edu
Fri Apr 25 02:35:49 CEST 2014

Hi Dan,

Thank you for the further elaboration. I have been taking "attributive" to
mean "attaches via the head-modifier rule" (and in the case of intersective
guys, does the LTOP identification thing) as opposed to predicative meaning
"attaches via the head-subject rule/or as complement of the copula" and
does not do any LTOP sharing.  That is, in terms of a predominantly
semantic distinction.  This is the distinction that I'd expect to find
active crosslinguistically.  (In French, for example, there are pre-head
and post-head attaching adjectives, but as far as I know, that distinction
does not align with the ability to appear as the complement of the copula).

Do the "a-" guys have the post-nominal modifier possibility?

"Anything ablaze should be put out immediately."
"Anyone still abed needs to get up now."

(The second one of those two sounds better, but that could just be the
weight thing again...)  Also, isn't it true that the ordinarily pre-nominal
ones can be post-nominal if they're phrasal, so:

"Anything other than that is off-limits."
"Children teenage or younger must be supervised."

(Here the second one is maybe suspect to me...)

At any rate, it seems to me that there are two separate distinctions that
happen to correlate in English (more or less) that I don't think we can
count on correlating cross-linguistically.  I'd rather use POSTHEAD to
control the direction of modification (in the Matrix) and save PRD for
can/can't appear as the complement of the copula (or as the head of a
sentence, depending on the language).  Do you see anywhere we'd run into
trouble by using MOD < > to indicate adjectives that can't appear as

(You might then ask why call them adjectives if they can't appear as
modifiers, either left or right, of nouns. I think the motivation would
likely be morphological/relate to other distributional properties.  That
is, imagine a language that has a class of things that all inflect the same
way, form comparative/superlative constructions the same way, and take the
same kind of degree modifiers. But some of those guys can be adnominal
modifiers and some can't...)


On Thu, Apr 24, 2014 at 4:55 PM, Dan Flickinger <danf at stanford.edu> wrote:

> I'm not quite sure, Emily, what you mean by "potentially attributive".
>  There are adjectives like "mere" that can only appear prenominally, and it
> is this positional ability that I associate with "attributive" adjectives.
>  The other common places for adjectives are either as postnominal
> modifiers, or in the copula construction, and both of these I take to be
> characteristic of "predicative" adjectives.  Maybe I'm overlooking some
> other context in which an adjective like "mere" could occur, but I don't
> see it.  Hence the property [PRD -] exactly means "I only appear as a
> prenominal modifier" (equivalent to "I am an attributive adjective"), and
> [PRD +] means "I do not appear as a prenominal modifier, but I can
> (potentially) appear either postnominally, or with a copula" (equivalent to
> "I am a predicative adjective").  Adjectives like "afraid" are lexically
> constrained to be [PRD +].  Most adjectives are underspecified for PRD, and
> hence can behave as either attributive or predicative.  The ERG lexicon
> includes a large set of lexical entries of the attributive-adjective type,
> but the great majority of these are hyphenated phrases treated as
> multi-words (a kind of grammar hack), or are words like "downstream" that
> also have a predicative entry which might also serve as a VP-modifier, so I
> divide their work into "attribute adjective" and "lexical PP" (where PPs
> can only be post-head or appear with the copula).  Some other `true'
> attributives include "everyday", "foster", "future", "inner", "kindred",
> "lone", "mock", "other", "overall", "teenage", "utter", and maybe
> "veritable".  Some true "predicative-only" adjectives include "a-" guys
> like "abed", "afraid", "ablaze", etc., but also "galore" (probably not with
> copula), "so" meaning "true" as in "that is so", maybe "ready" meaning
> "prepared", where "a ready smile" has a different sense, and probaby "well"
> meaning "healthy".
>  - Dan
> ----- Original Message -----
> From: "Emily M. Bender" <ebender at uw.edu>
> To: "Dan Flickinger" <danf at stanford.edu>
> Cc: "Woodley Packard" <sweaglesw at sweaglesw.org>, "developers" <
> developers at delph-in.net>, "T.J. Trimble" <trimblet at me.com>
> Sent: Thursday, April 24, 2014 4:08:44 PM
> Subject: Re: [developers] MOD value in predicative only adjectives
> Thanks for the clarification Dan.  I'm a bit puzzled about the distinction
> that PRD +/- marks then.  PRD - apparently doesn't mean "can't be
> attributive" (which is what we were taking it to mean) but rather "can't be
> a pre-nominal modifier".  Do you think that all adjectives must be
> potentially attributive, just not necessarily pre-nominally?
> (I don't have any examples of non-attributves off the top of my head.  I'm
> looking here for the counter part to "mere", which is attributive-only.)
> Emily
> On Thu, Apr 24, 2014 at 3:50 PM, Dan Flickinger <danf at stanford.edu> wrote:
> > Hi T.J. (and Woodley) -
> >
> > As Woodley notes, there are two contexts in which predicative adjectives
> > can appear:
> > (1) in copula constructions as in |the ships are afloat|
> > (2) as postnominal modifiers as in |the ships afloat reappeared|
> >
> > While you're right that the MOD value is unneeded for the copula use,
> that
> > non-empty value is necessary for the postnominal use, since the ERG
> > combines "ships" with "afloat" using a head-modifier construction.  Hence
> > the boolean PRD feature is used to distinguish attributive from
> predicative
> > adjectives, and not the MOD value.
> >
> > Woodley's example |the dogs awake arise| might sound awkward, but |anyone
> > awake at that hour must be crazy| is impeccable, suggesting that there
> is a
> > "heaviness" element involved in the acceptability of single-word
> > post-nominal adjectives, and hence maybe something about information
> > structure is at play.  However, even the single-word ones can sound fine:
> >  |the only rooms available are doubles|
> >  |the first person awake was the old man|
> >
> > As for Woodley's |I found the dogs awake|, this has several readings
> which
> > may obscure the issue of postnominal modification -- one with the
> > three-argument "find" as in |I found him (to be) amusing|, and one with
> > transitive "find" and a "depictive" adjective, as in |I found the keys
> > (while I was) blindfolded|.  Perhaps more to the immediate point is an
> > example like |I fed the children awake at dawn an early breakfast|.  The
> > awkwardness of |I fed the children awake an early breakfast| is, on my
> > current view, not an issue of grammaticality, but something (mysterious)
> to
> > do with pragmatics.
> >
> >  Dan
> >
> > ----- Original Message -----
> > From: "Woodley Packard" <sweaglesw at sweaglesw.org>
> > To: "T.J. Trimble" <trimblet at me.com>
> > Cc: developers at delph-in.net
> > Sent: Thursday, April 24, 2014 2:16:32 PM
> > Subject: Re: [developers] MOD value in predicative only adjectives
> >
> > Interesting question, T.J.; I would also like to know the answer to this
> > one.  One side effect seems to be the prediction that the following
> string
> > is grammatical:
> >
> > The dogs awake arise.
> >
> > i.e. the predicative-only adjective "awake" is allowed to modify
> > post-nominally.  To me this is a surprising prediction; but maybe those
> on
> > the list with more flexible minds won't have an issue with it.  Other
> > constructions such as "I found the dogs awake." get their own analysis,
> > with the "awake" predicate a scopal argument of "find," so they can't be
> > the explanation here.
> >
> > -Woodley
> >
> >
> > On Apr 24, 2014, at 1:57 PM, "T.J. Trimble" <trimblet at me.com> wrote:
> >
> > > So, two related questions about the ERG or about these sort of
> > constructions in other grammars/languages:
> > >
> > > 1) Are there any examples of the MOD value of these predicative only
> > adjectives being utilized?
> > >
> > > 2) Is there any compelling reason to use PRD +/– to constrain this
> > instead of MOD < >?
> >
> >
> >
> --
> Emily M. Bender
> Associate Professor
> Department of Linguistics
> Check out CLMS on facebook! http://www.facebook.com/uwclma

Emily M. Bender
Associate Professor
Department of Linguistics
Check out CLMS on facebook! http://www.facebook.com/uwclma
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.delph-in.net/archives/developers/attachments/20140424/521b5d26/attachment.html>

More information about the developers mailing list