RE: RT list: Degrees of inference?

Hi,
 
This discussion is interesting but I think Ronnie Sims' observation that
somehow "it doesn't hit the spot" is very pertinent here. It seems that too
little attention is paid to a couple of important points of RT. .
 
1) Optimal relevance is not universally defined, but always context
dependent - which means, in effect, dependent on the mutual cognitive
environment of the communication partners. The point from which this
discussion started, however, seem to be cross-cultural judgments: in this
case, scholars from one culture comparing the processing effort, esp. that
incurred by inference, in acts of communication in another culture. This
means, however, comparing two most likely non-commensurate things if the
cognitive environments of both acts of communication are not the same, which
in these cases they hardly will be. So what seems to be happening in these
evaluations is something like the following: one is trying to compare the
processing cost [e.g. steps of inference] that would one would have to spend
in one's own cognitive environment to obtain an utterance considered
satisfactory in one's own cognitive environment with the effort it takes one
to process, in one's own cognitive environment, the utterance actually
offered in quite a different cognitive environment. For example, a Western
person considers the processing cost he/she would need to invest in
processing the utterance of an East-African person to obtain the information
a Western person would expect to be told, and comes to the conclusion that
this information is either only very weakly implied or requires a lot of
inferencing, which could have been avoided by a more direct answer, of the
kind expected in a Western cognitive environment. From an RT point of view,
this does not make sense. Optimal relevance, with both processing cost and
cognitive effects involved, have to evaluated relative to the cognitive
environment in which they take place.
 
2) Also, the 1995 version of optimal relevance includes a very significant
modification in its second part, bringing in the preferences of the
communicator: "b) The ostensive stimulus is the most relevant compatible
with the communicator's abilities and preferences" (Sperber, D., and D.
Wilson. 1995:270). This means, among other things, that the communicative
POR does not oblige the communicator to supply, without unnecessary
processing effort, the information the addressees might like to get but a)
only information that is in line with the abilities of the communicator -
e.g. the communicator can't supply info she does not have or can only come
up with as good a stimulus as her particular skills allow - and b)
information that the communicator is prepared to supply (preferences).
 
This has far-reaching implications for the topic under discussion here.
Suppose the mutual cognitive environment of a group contains the belief that
information is a very valuable, and potentially also risky, commodity: one
never knows what others might do with that information. When a person with
that cognitive environment is asked by a stranger how far it is to a certain
place or where such and such a person lives, then that person would, if
indeed he felt he had to answer, tend to keep the information quite vague:
"Oh, it's over there, quite far" - even if he was in a position to give a
much clearer answer. While the vagueness of the answer probably would not
surprise a person sharing a similar cognitive environment - it's only what
any stranger could expect to be told - an outsider to the culture might
conclude that either these people don't know how to be precise - how to
formulate an optimal stimulus - or that they expected the addressee to work
hard to figure out from that answer where they'd have to go. In the latter
case, in the spirit of the discussion going on here, one might conclude that
this is a culture that generally demands higher processing efforts from the
audience than some other cultures. Similarly with business negotations,
perhaps in an Asian context: if it is customary to reveal only as little of
one's own interests to the other party as possible in order to get the best
deal for oneself and to keep open as many options as possible, then, again,
one will naturally tend to be rather vague, saying only as much as one feels
one has to say to remain polite and to keep things moving. The other party,
if they share the same cognitive environment, will interpret the answers
given along these lines - probably also trying to figure out what the first
party is trying to hide. Such interaction obviously involves a lot of mental
work. This would not be needed in another group with less sophisticated
conventions where one is quite ready to reveal information.
 
The important point is that, in all these cases, people are actually guided
by the communicative POR (adequate cogn. effects with minimal processing
effort) - keeping in mind the preferences of the communicator, and the
mutual cognitive environment in which the communication takes place. To take
the observable difference in processing cost as evidence for cultural
parameters that somehow - arbitrarily? - determine a sort of "default level"
of processing effort or inference for that group that is e.g. higher than
the "default level" of some other group would be missing the point entirely.
It would be much more useful to study the "contents" of the cognitive
environment concerned and find out what cultural assumptions there are in
that cognitive environment that would make the information and the stimulus
used to convey it optimally relevant. That would actually help outsiders to
better understand those communications - and the culture in which they take
place.
 
Personally I believe there is no reason to assume that there are groups that
set "minimal processing cost" as such at a higher level than perhaps some
other groups, or put more "responsibility" on the audience than other groups
do. I would expect that in all and every case it will turn out that such
differences arise not from culturally set default values for these two
parameters but are natural consequences of beliefs and values of the
cognitive environments concerned - whether this concerns just two
individuals or large cultural or ethnic groups. (So this should actually be
a testable hypothesis.)
 
Well, this is my two cents' worth.
Ernst-August Gutt
 
 
  _____

From: owner-relevance@linguistics.ucl.ac.uk
[mailto:owner-relevance@linguistics.ucl.ac.uk] On Behalf Of Ronnie Sim
Sent: 28 November 2013 10:20
To: ian mackenzie
Cc: relevance@linguistics.ucl.ac.uk
Subject: Re: RT list: Degrees of inference?

Now you speak of
"the SIL linguists with knowledge of Africa, and the various Japanese
linguists working in RT. Then again, they tend to analyse specific
particles, connectors, discourse markers, etc., i.e. do real work, rather
than trade in airy generalisations like I do!"

I'd add three points:
1. Procedural meaning came out of RT work, and a number of us are applying
it in analysing those particles &c ... I think with useful results,
particularly in turning up particles in Greek, Hebrew and African languages
that mark a string as metarepresentational.

2. "Trading in airy generalisations" is in fact where the new insights come,
that the detail work can pick up. don't deprecate the theoretical
developments you guys uncover; I don't-- and a reason for speaking to the
issue you raise is the hope for some new thinking.

3. "Ranking languages according to levels of inference, or the use of
implicatures" doesn't hit the spot, though, does it? It would presumably not
be IN the language, but in how it is used for communicating in different
interpersonal contexts, some of which would rely more on shared assumptions,
or if Yoshikawa goes further, some of which pay less attention to shared
assumptions and require audiences to spend more processing effort than RT
suggests. I assume that all communities have situations in which speaking is
deliberately less plain, and, if you are correct, that some communities
spread this across many contexts, and make the assumption that the audience
will make effort to understand. If that roughly represents the kind of
societal contexts you have in mind, I then want to ask two further
questions.

a. How would RT be affected if the "minimal processing effort" side is
weakened? I guess the affect can be tolerated, since all societies do
communicate in situations where more effort is required [by some in the
audience]. It presumably links up with the three strategies in Dan Sperber's
paper Understanding Verbal Understanding, which also accepts audiencescan
jack up the effort side.

b. What kind of evidence would be required to show that community A
typically uses language in ways that lay more responsibility on
hearers/readers? Not from less benevolence or to obfuscate communication,
but for some other reason.

Coming back to eastern Africa, yes, I "see" communities that apparently lay
more responsibility on audiences, and do not utter to minimise processing
costs. And, yes, I see literature that requires more processing. Isn't it
the case that much modern English literature is more open-ended than RT has
allowed for?

I'd really like to hear discussion from those in RT who are in a good
position to speak to these questions.

Ronnie

On 27/11/2013 18:08, ian mackenzie wrote:

Thank you, Ronnie Sim.

It’s not only the Horn of Africa. John Hinds (1987) quotes Yoshikawa(1978)
(I can give the refs. to anyone who wants them) saying “the basic principle
of communication in Japan, the fact that what is verbally expressed and what
is actually intended are two different things, is something that Japanese
people are supposed to be aware of.”

S&W do of course argue that language use (in any natural language) is almost
never literal, and that inference is almost always involved, but I’m still
surprised that no-one seems to have tried to rank languages according to
levels of inference, or the use of implicatures. Especially with all the SIL
linguists with knowledge of Africa, and the various Japanese linguists
working in RT. Then again, they tend to analyse specific particles,
connectors, discourse markers, etc., i.e. do real work, rather than trade in
airy generalisations like I do!

Ian MacKenzie

On Wed, Nov 27, 2013 at 12:48 PM, Ronnie Sim <ronnie_sim@sil.org> wrote:

Ian, you asked

"Can anyone tell me whether any work has been done on RT and degrees of
inference considered to be necessary in different languages/cultures?"

The short answer from me is a nagative ... but I hope others on the RT list
will have something more positive.

What I can do is agree on the basis of anecdotal evidence.
I'd say in Africa generally -- so, at least in the societies I am most
familiar with, in eastern Africa and the Horn, speaking directly is not the
norm. People tend to talk indirectly about issues, and rely on the audience
doing a lot of inferencing to keep track of what is intended. The more
serious the issue is, the more care there is in not speaking to it directly.
So the western "value" of being up front with what is on our mind is not
appreciated, not valued, and may be read as insensitive. Put like that, I
can also see indirectness, and consequent reliance on inferencing, at work
in western societies in at least some aspects of life.

Secondly, especially in the Horn of Africa, it seems to me that explanation
[whenever called for] is achieved via the presentation of analogies --
'parables' if you like. A asks a question about something that is not clear,
and B responds by offering an analogy. A has to inference in what way the
analogy fits. Again, the west does this, and some tropes depend on analogy,
but in the Horn, in some societies at least, it is an art form--a genre,
almost. Sometimes the analogy might be fairly 'narrow' and be interpretable
in terms of one [or two] quite strong implicatures. At other times, it seems
to depend on a range of weak implicatures, no one of which is consciously
'intended'. Again certain [traditional?] negotiations in the west may show
similar signs. I think of negotiations that used to involve two families
arranging a marriage between one of 'our sons' and one of 'their daughters'.
And analogy and indirectness is the means of carrying it out. some time ago
now, I heard this done in terms of a young man wishing to harbour his boat,
so it can be quite raucous and baudy.

Is indirectness something that surfaces in all societies, in some
communication, and then, along the lines of your question, do some societies
prefer indirectness, and a consequent greater dependence on inferencing?

Anthropologist Mary Douglas set up two parameters for understanding societal
behaviour -- Group and Grid. The Group parameter involves the bondedness
among members of the group; the grid parameter involves stratification of
tasks/roles among group members. Low Group/Low Grid would be individualist;
High group/High grid is hierarchical societies; High group/Low grid is
egalitarian; Low group/High grid is fatalist. The epitomising terms are
extremes; most societies presumably show trends not extremes.

Communication in different Group/Grid societies might reasonably be
predicted to show differences brought about by such cultural factors.

There is a correlation between the Group/Grid concept and the concept of
high/low Context. Plausibly, communication in high shared context situations
relies more heavily on inferencing.

I would like to hear from others who have explored more systematically these
cultural aspects in which communication takes place. Which is my main reason
for saying anything at all, no matter how little! I assume that the work of
Helen Spencer-Oatey, Stells Ting-Toomey and those who are associated with
the same interests could give us much more.

Ronnie [Sim]

On 26/11/2013 17:07, ian mackenzie wrote:

In the 1980s, John Hinds distinguished between speaker/writer and
listener/reader responsible languages. In the former (e.g. English and
French), the speaker/writer is primarily responsible for effective
communication; in the latter (e.g. Japanese and perhaps Korean) it is the
listener/reader. It is often said that in Japanese what is expressed and
what is intended tend to be two different things; there is no obligation to
give full explanations and clarifications, to be linear and direct, or to
use explicit coherence markers and transitional statements. Consequently
listeners/readers need to rely heavily on inference.

In the 1970s, Edward Hall distinguished between high- and low-context
cultures, giving Japan as an example of a high-context culture in which
people tend to have similar experiences and expectations, allowing many
things to be left unsaid, and inferences to be drawn from implicit shared
cultural knowledge.

The authors of RT come from speaker/writer responsible cultures (the French
even have the deluded saying ce qui n’est pas clair n’est pas français!).
Can anyone tell me whether any work has been done on RT and degrees of
inference considered to be necessary in different languages/cultures?

Many thanks,

Ian MacKenzie

Faculté de traduction et d’interprétation, Université de Genève

ian.mackenzie@unige.ch

Recently published: English as a Lingua Franca: Theorizing and Teaching
English, Routledge, 2013.
Received on Thu Nov 28 14:41:31 2013