Articles and the Givenness Hierarchy

[Warning: Long, and theory-laden.]

In my languages, I've traditionally avoided things like definite
and indefinite articles because: (a) I don't like them, and (b) I
can't seem to avoid making them work like English (or Spanish).
The result is that definiteness/indefiniteness isn't marked with
articles in any of my languages.  Nevertheless, something like
definiteness (i.e., how salient the information presented should
be to the hearer) is marked in every language, though differently
for each.  By trying to avoid the problem, I know doubt unwittingly
reproduced English-like constructions.

To that end, we recently read a paper in my pragmatics class
which presented a theory about what the authors call givenness.
The reference is:

Gundel, J., N. Hedberg, R. Zacharski (1993). Cognitive Status and
	the Form of Referring Expressions in Discourse.  Language,
	Vol. 69, 2:274-307.

(It's available from jstor.org.)

Anyway, I thought the *idea* presented was neat, and might be
useful to conlangers, so I thought I'd give you the gist of it.

[Note: I'm presenting this as an "if-this-were-true-it-might-be-
cool" theory.  The paper and theory have *many* problems.  In
fact, my pragmatics professor is writing a response to the article,
since apparently no one else has critically reviewed it since it's
original publication.]

Okay, the givenness hierarchy.  Essentially, the givenness
hierarchy is a hierarchy of how "in focus" a given NP is.  (How
is "in focus" defined?  You'd be hard-pressed to find an answer
to that in the article.  Think of it as "most relevant", but with a
red flag attached.)  This hierarchy has six members, arranged
in a particular order.  For any given member n, it is assumed
that a hierarchical position that is < n will be entailed by n.
Additionally, for any hierarchical position that is > n, it will be
assumed that n will conversationally implicate *not* n+1, n+2,
etc.  That's a vague description, but just keep it in mind as
we go along.

The givenness hierarchy is as follows (going from least to
greatest):

1. Type Identifiable
2. Referential
3. Uniquely Identifiable
4. Familiar
5. Activated
6. In Focus

Now I'll define what they mean, using examples from English.
For non-native speakers of English, some of these distinctions
will be very subtle.  For the purposes of the theory, you just
have to trust that they work (that goes for anyone, actually.
It's *very* easy to prove how these don't always work, but
just pretend, for the sake of conlanging).

1. Type Identifiable: An NP is type identifiable if it's brand new
information.  Pretty much, you just have to know what it is.

English: a(n) NP

Example: I saw a bird.

Explanation: If you hear this sentence, you don't know what
kind of car it is, and presumably have no memory of it.  All
you need to know is what a car is.  The hearer is not assumed
to have any knowledge of the entity itself.

***

2. Referential: This is a reference to indefinite NP, but the
hearer is supposed to understand that it's going to be the new
topic of conversation, and that for the speaker, the entity is
specified.

English: this NP

Example: I met this great guy yesterday.

Explanation: According to Gundel et al., this construction is
supposed to be less indefinite than using "a", but the hearer
is supposed to recognize that the speaker has a specific one
in mind, the way they don't when they say, "I want to own
a house some day."

***

3. Uniquely Identifiable: The hearer can identify the referent
just by hearing the NP.

English: the NP

Example: I saw the bird.

Explanation: Compare this one to the example in (1).  Imagine
a friend walked up to you and opened up the conversation
with, "I saw the bird."  If you didn't have a bird in mind, you'd
be pretty bewildered, and would say, "What bird?"  That would
be signaling to the speaker that their use of "the" was inappropriate,
because you weren't able to determine the specific referent
based on the NP alone.

***

4. Familiar: An NP is, to a certain extent, in the hearer's long
term memory.

English: that NP

Example: That dog kept me awake last night.

Explanation: The example above is *not* when the dog is
present and is being pointed out.  This is, perhaps, said at work
where there are no dogs present.  In order to use a sentence
like this one, the hearer has to have had conversations about
a specific dog over a long enough period of time that they can
recognize which dog is being talked about when they hear
"that dog".

***

5. Activated: An NP is in the hearer's short term memory.

English: that, this, this NP

Example: I saw that.

Explanation: The above example might be a response to, say,
"I saw Kung Fu Hustle the other day".  If something is activated,
it's in the discourse, or was only a short time before.

***

6. In Focus: This is as prominent and relevant as an NP can
be--it's "on stage".

English: pronouns

Example: It's on the table.

Explanation: In order to use a pronoun, the theory is that the
hearer has to know exactly what's being discussed.  It's the
most prominent, most relevant, most salient entity in the
discourse.

***

Okay, that's the hierarchy.  To rephrase what I stated above
using examples, the theory is that something higher up in
the hierarchy entails something lower down in the hierarchy.
So if you say, "That's the dog", it entails that there is "a dog".
Additionally, the theory states that something lower down in
the hierarchy will conversationally implicate *not* something
higher up in the hierarchy.  That is, a speaker will produce
the member highest in the hierarchy that they can get away
with, based on hearer knowledge.  So it would be weird to
say:

A: "Hey, have you seen my dog?"
B: "Yes, I've seen a dog."

B's statement is true, and can be considered a response to A's
question, but it's bizarre, because A would assume that if he
had seen his dog, he would use something higher up in the
hierarchy--like a pronoun.

***

So that's basically how the thing works.  What's interesting is
that if you start looking at other languages, they express the
various parts of this hierarchy differently.  So, for example,
while English has two separate forms for Referentials and
Type Identifiables, Gundel et al. claim that in Spanish, you
use "un NP" for both Referentials and Type Identifiables.

[Note: I make no claims about the validity of their cross-linguistic
data, since I can verify that their English data is problematic.]

So, of immediate interest to conlangers, is that this is a way
to think about articles, definiteness, topic, etc., that will allow
you to break out of your native language habits, if you have
trouble with that (e.g., like I do, with respect to this).  Whether
or not this hierarchy works, if you take it as true, it gives you
some possibilities to play around with.  For example, what if
you had a language that treated *everything* but what was
in focus one way--say, with null marking?  Or what if you had
a single form that could be used both for type identifiables and
activated NP's?  Presumably, the two wouldn't usually be
confused in specific contexts, so it's certainly possible.

Additionally, with this basic framework, you can also modify
it.  This shouldn't be taken as *the* hierarchy, in my opinion
(especially since, of all of the languages they sampled, *only*
English had a separate lexical entry for each of the six categories.
Highly suspect...), but given the general format, it's easy to
see how it can be modified to fit a particular language, or particular
conculture.  For example, what if there were a separate category
for NP's whose identity the speaker assumes the hearer knows,
but which, should the hearer fail to recognize the NP, should
not be asked about.  In other words, it'd be something that
the speaker wants to talk about, and he assumes the hearer
knows, but if they don't, he doesn't want to go back and have
to explain the referent.  Or perhaps there might be a category
for the first thing ever mentioned in a discourse.  Anything
is possible.

Anyway, even though we really tore this paper and theory
apart in my pragmatics class, I really wished that I'd heard
about it *before* I started creating languages, so that I wouldn't
have been so lazy with definiteness/hearer knowledge, etc.

Oh, and as you can see by the English examples, languages
aren't designed to fill these categories: Lexical items are drawn
from other places to fill the void.  So, for me, for example, there's
no need for me to go back create new lexical items, or delete
old ones: I just need to have a story for how the different
categories are expressed.

All right, that's it.  Sorry for the long, theory-laden post.  I just
thought it was helpful, from a conlanging point of view, so I
thought I'd share.

-David
*******************************************************************
"A male love inevivi i'ala'i oku i ue pokulu'ume o heki a."
"No eternal reward will forgive us now for wasting the dawn."

-Jim Morrison

http://dedalvs.free.fr/

Replies