Inventory of demonstratives

They are quite possibly related to/derived from the Dena'ina spatial adverbs gu 'here' and ghu 'there'.

Definition

I follow here the definition by Himmelmann (1996:210f.). Himmelmann's goal there is to define demonstratives in contrast to definite articles and pronouns.

• "the element must be in paradigmatic relation to elements which -- when used exophorically -- locate the entity referred to on a distance scale: as proximal, distal, etc.
• the element should not be amenable to the following two uses which are characteristic for definite articles:
• (...) demonstratives are generally not usable for first mention of entities that are considered unique in any given speech community
• associative-anaphoric use (...)"

I originally intended...

to prove a hypothesis such as:

• Demonstratives in Dena'ina function similar to definite articles
• Demonstratives in Dena'ina make noun phrases [+definite] or [+specific], or possibly both
• Demonstratives in Dena'ina are used when the referent is accessible to both speaker and hearer
• the variation in the singular forms is due to factor X or Y or Z

In short, I anticipated some form of result.

Outline

Source: http://fishbowl.pastiche.org

Questions

Data (Language)

Dena'ina/Tanaina is spoken by ~60 people in Southcentral Alaska.

Data (Corpus)

Methodology

Problems

But the biggest problem...

...is that little matter of "discourse relevance". What constitutes a globally important, locally important, unimportant referent?

The suggestion by Givón (1983) to measure topic persistence has been modified by Thompson (1989) to determine the relevance of a referent: The more persistent a topic is, the more relevant it is. But does topic persistence really equal relevance? (Plus, that amount of counting is not feasible in large corpora)

Thompson (1989) also suggests using the title of the narrative as a cue; this only works with some titles though.

For this study, I eyeballed relevance; and this is probably the reason why my results are so muddled. There needs to be a better way to measure relevance, but what could that be? Ask speakers? Ideas?

Structure of NPs

We'll assume that NPs have to have a head and that they can be modified by other elements.

Elements that can be in the head are:

• simple noun stems
• possessed nouns
• compound nouns (the line between possessed nouns and compound nouns is not always clear; see also Rice (1989))
• relativized verb forms
• free pronouns
• demonstratives
• numerals
• and possibly other elements; this is just what I found in my dataset

Structure of NPs

Modifiers can precede or follow the head. The general preference seems to be:

There are of course exceptions to this.

And just like the verbal template, reality usually isn't quite as bad.

Some NPs

Some NPs

Frequency of demonstratives

Two surprizing things:

1. only 1,015 NPs in 1,558 story units
2. over 1/3 of NPs contain a demonstrative

Syntactic properties

• Demonstratives agree with referent in animacy and number:

  (7)	łik'a ghin	'that dog'	(nonhuman)
  (8)	deghk'isen ghun	'that woman'	(human, singular)
  (9)	deghk'isna ghuna	'those women'	(human, plural)

• There are no exceptions in the corpus.

Syntactic properties

• Demonstratives can have pronominal and adnominal function; same form for both functions.

  pronominal	adnominal
  22	339

• Adnominal demonstratives can precede or follow the head noun; again, no difference in form.

  prenominal	postnominal
  3	336

Adnominal vs. pronominal

Prenominal vs. postnominal

There are very few cases of prenominal demonstratives. Their contexts are strikingly similar, though.

Prenominal vs. postnominal

• all prenominal demonstratives are proximal
• they all are clearly deictic
• they all are outside the main story line: 2 have the story as a referent, and the third one is in narrated conversation
• and while there are not a lot of examples from the corpus, they also occur (somewhat more frequently) in elicited sentences

Discourse functions of postnominal demonstratives

• Introduction of a globally important referent:

bare NP	14
NP+Dem	21
Total	35

The exceptions are:

• non-singular (2 times)
• non-specific (5 times)
• non-human (6 times)
• also, if there are several highly relevant referents in one story, then frequently one of them is introduced by an NP containing a demonstrative and the others aren't (9 narratives)

Introducing referents

• In this discourse context, there is a dialect difference:

  Inland		Upper Inlet
  Distal	Proximal	Distal	Proximal
  3	8	8	1

Continuting reference: distal vs. proximal

• If a referent is introduced using a demonstrative, then a demonstrative continues to be used thoughout the text.
• Typically, the same demonstrative is used for a particular referent.
• In the Inland dialect, NPs referring to the main referent(s) usually contain a proximal demonstrative, others contain a distal one.
• In the Upper Inlet dialect, all NPs are more likely to contain a distal demonstrative

	Inland	Upper Inlet	Total
proximal	100	9	109
distal	109	145	254
total	209	154	363

(14) Story units 001-031 of Raven Story (Inland dialect), Tenenbaum (2006:90-92)

(...) [19 lines containing no nominal reference to Raven omitted]

(...) [9 lines containing no nominal reference to Raven omitted]

Going back to our definition...

• "demonstratives are generally not usable for first mention of entities that are considered unique in any given speech community" (Himmelmann 1996:210)

Raven is considered unique. Yet almost all mention of him includes the demonstrative gun. Why?

• Does this mean that in this dialect, the proximal demonstrative is developing into an article?
• Or is Raven not unique?
• Or is the function of the demonstrative here mainly to mark Raven as [+human]?

(15) Story units 002-035 of Glacier People Story (Upper Inlet dialect) by Shem Pete

[15 lines containing no nominal reference to the man omitted]

[16 lines containing no nominal reference to the man omitted]

Why use a demonstrative?

• If we compare the last few examples to the ones with a pre-nominal demonstrative above, we see that there is much less of a deictic function.
• If a highly relevant referent is encoded by an NP, then there is the tendency for this NP to contain a demonstrative.
• However, this only works for globally relevant, [+definite, +specific] referents.
• Since we don't have good measures for relevance, there is no point in doing stats here; I certainly started cheating myself here when coding for relevance in an impressionistic fashion.
• So, this is a claim that must still be proven; and it all hinges on:

Closing question

How can we measure relevance?

Summary

• Globally relevant referents tend to be encoded by NPs containing a demonstrative
• In the Inland dialect, a proximal demonstrative is used
• In the Upper Inlet dialect, a distal demonstrative is used
• At this point, the function of demonstratives in other NPs is not clear; quite possibly, it has to do with local relevance