RST and implicit communication: Relational Propositions

[Bill Mann]

Dear list:

I would like to give a short status report on an old RST topic: implicit
communication that arises out of discourse structure.

Some very early RST reports, as well as later publications, were devoted to
implicit communication, under the name Relational Propositions (RPs).

The basic idea is this: discourse structure itself is capable of creating part
of the overall communicative effect that a text produces.  The source of this
is conceptually separate from  the clauses and  the words of the text, although
of course these are not independent.

In particular, each relation that is part of the structure of a text can assert
a proposition.

The act status of these assertions is perhaps always as deniable (defeasible)
assertions by the writer.  If a text is altered so that the Relational
Proposition is denied, then the corresponding relation no longer applies.  In
that case the text will become incoherent if no other plausible relation can be
assigned in its place (or, more generally, if no other complete RST analysis is
plausible).

Here are several examples from the RST website,
http://www.sil.org/linguistics/RST .  All of the propositions are stated
informally.

  1) One example would be the analysis of "Lets be clear:" which is unit 10 of
the Common Cause text.  (The analysis is on the RST website among the published
analyses.)  Unit 10 is linked to units 11-13 by a Justify relation. The
assertion would be that the author's expressed desire to be clear gives him a
right to state the nuclear portion, units 11-13.

  2) Another example from the same text would be the analysis of the following
fragment, units 7 and 8: " <7>Rather, I think we will be
stronger and more effective <8>if we stick to those issues of governmental
structure and process, broadly defined, that have formed the core of our
agenda for years. "

The Condition relation links these two, with 7 as the nucleus.  The
proposition would be that

provided that  the conditions in <8> hold, the CCC organization will be stronger
and more effective.

In the second example, the Relational Proposition is (also) expressed
explicitly.

3) A third example would be the Elaboration relation in the same analysis
between units 7-8 as nucleus and unit 9 as satellite:

<7>Rather, I think we will be
stronger and more effective <8>if we stick to those issues of governmental
structure and process, broadly defined, that have formed the core of our
agenda for years. <9>Open government, campaign finance reform, and fighting
he influence of special interests and big money, these are our kinds of issues.

Here, the concept represented by the phrase "Open government, campaign finance
reform, and fighting the influence of special interests and big money" is an
elaboration on the concept represented by "issues of governmental
structure and process."    (Both set::member and abstraction::instance patterns
apply.)  Based on set::member, the Relational Proposition would be something
like "{open government, ...}  are issues of governmental
structure and process.

In these examples as in RST generally, it is not the text forms that produce the
RPs but rather what they communicate in conceptual terms.

We can divide the attributes of Relational Propositions into two roughly
distinct groups based on how well established and how clearly understood they
are.

The better established attributes of RPs would include these:

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

RPs may be either signaled or implicit.

Recognition of RPs is not essentially  form based.

RPs are always assertions, not simply propositions without a speaker attitude.

RPs are distinct from other recognized phenomena of implicit communication,
including in particular presupposition, Gricean implicature, indirect speech
acts, anaphoric presupposition, invited inference, zero anaphora, pronominal
reference, deixis, metaphor, simile, euphemism, irony, memory of shared
experience,  intertextuality, ritual fulfillment, interpretation of intonation.

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

The attributes of RPs that are NOT well established or clearly formulated
include these:

        000000000000000000

It is believed that RPs are derivable from the relation definitions and an RST
analysis as they apply to the text (or nearly so), rather than simply being
stipulated. (A very small amount of work is in progress on how derivation of
RPs can be modeled.)

 ** A crucial and logically prior problem is to identify just what RP is
asserted.

The communicative act performed or established by a relation in a discourse
structure is necessarily always assertion;  that assertion is defeasible
whenever it is implicit.

        000000000000000000

Of these poorly developed aspects of RPs, I think that the one marked with ** is
the most crucial.  Before formalization, there needs to be a more extensive
account of what might be formalized.

I plan to circulate a list of references on RPs in a separate message.

I look forward to our discussion.

Bill Mann