Agentivity and intentionality

Wolfgang Schulze W.Schulze at LRZ.UNI-MUENCHEN.DE
Thu Mar 8 11:20:51 UTC 2001

Dear Funknetters,

may well be that I'm somewhat behind the discussion - I had the chance
to read the last few postings only. Still, the matter is strongly
related to my main research domain - thus you perhaps allow if I add
some thoughts that result from what I have been working on.

[What follows are some rudimentary thoughts on the issue - more details
can be found in the exposition of the underlying frame work ('Grammar of
Scenes and Scenarios') as elaborated in W. Schulze 1998. Person, Klasse,
Kongruenz, chapters I,7 and IV. Munich/Newcastle: Lincom Europa].

I think, the basic problem is whether you deal with 'intentionality'
from a linguistic point of view or from a purely cognitive perspective.
By this I mean that it is important to indicate whether the notion of
'intentionality' is [in a given case] substantiatable with the help of a
specific 'behavior' of the linguistic paradigm, or whether it stems from
the *interpretation* of linguistic data in a psychological or
what-so-ever perspective. For instance: If you discuss the German

'Mein Gott, ich habe den Schlüssel vergessen'
[Oh my God, I've forgotten the keys]

you can easily assume that the whole 'scene' (in the terms of the
'Grammar of Scenes and Scenarios') or cognitive construct of the given
State of Affairs is dominated by an 'unintentional reading'. But no
morphology, morphosyntax, or morphosemantics helps the listener to infer
this feature, it's simply the lexical entry and the 'scenic template'
that lies behind it. It's a well known language play in German to
construe counterfactual structures that are based on such unintentional
scenes, e.g.

'Komm, wir finden einen Schatz!' [instead of: 'Komm, wir suchen einen
[Let's go and find a treasure!] -  [Let's go and look for a treasure!]

In such instances, there are two competing 'scenes', one which is based
on a deontic speech act verb which is structurally coupled with
intentionality ['komm!'], and one that is evoking an unintentional scene.

In such instances, counterfactual linguistic material can result in the
emergence of a 'linguistic' coding of (un)intentionality. But quite
often [at least in German], the degree of (un)intentionality has to
inferred from the co[n]text; in other words - the allocation of features
of (un)intentionality is then strongly related to the state of situative
or episodic (as well as encyclopedic) knowledge of the hearer [not the
speaker! - For them, the degree of (un)intentionality is an
idiosyncratically 'known' feature].

As far as linguistics is concerned, we should (imho) avoid the
interpretation of textual material with respect to possibly emerging
grades of (un)intentionality, as long as we don't have either a
linguistic (or textual) clue to do so, or enough episodic, situativ, and
encyclopedic knowledge of the scene communicated by the linguistic
expression. Anything else would lie beyond the scope of linguistics.

The above mentioned example has referred to what I'd like to call
'relational (un)intentionality': it is the 'relational domain' (in a
cognitive sense) established between 'ich' [I] and 'Schlüssel' [keys],
i.e. 'vergessen' [to forget] that is encyclopedically marked for the
feature of unintentionality. It is not determined by the nature of the
referential domain [the restriction in the scene concerns the degree of
'humanness' only]. Now compare:

Mein Gott, das Baby hat das Auto gefahren!
[My God, the baby has droven the car!]

Though 'fahren' normally evokes a scene marked for intentional features
[in German], here it is used with a referential domain the proponent of
which is normally not associated with such a feature [at least when the
encyclopedic knowledge related to 'fahren' is activated']. In this case,
the encyclopedically controlled feature 'baby' [-intentional>fahren] is
not effected by the 'positive' feature of 'fahren' [at least in a
standard reading of the above given example - it may well change if a
different (constrastive) word accent applies). The example illustrates
the second component of 'scenic intentionality', namely 'referential
intentionality'. It is this domain that is often discussed when dealing
with 'agentivity' and 'intentionality' [though unfortunately, the two
domains - coupled with each other just as the relational domain does not
exist without one or more referential domains and vice versa - are often
mixed up in the linguistic discussion]. But again we have to state that
the fature of (un)intentionality' here is only describable from a
non-linguistic, knowledge based perspective.

A linguistic reading of (un)intentionality has (imho) to make sure that
there either is a specified linguistic category (or linguistic behavior)
to encode points on the intentionality scale or that there are other
categorial elements (or behavioral patterns) that have the metaphorical
potential (often in co-paradigmatization with other elements) to do the
job. If the 'direct' way is given (i.e., if there as a symbolic relation
form/behavior X <=> pointer on the intentionality scale) things are
'easy'. We can probably start with three basic types: marker in the
referential domain; marker in the relational domain; 3) scenic (or:
clausal) markers (such as intonation patterns, sentence particles etc.).
But typology has shown that such overt pointers are rare. most often,
things are 'covert' (or: metaphorized from other paradigmatic /
behavioral patterns). Now: In the second case, things become difficult:
Are there linguistic (or cognitive based) structures that can be
declared to represent the most 'natural' structures comporting features
of the intentionality scale? Most people will agree when I say: yes:
referential domain > agentivity; relational domain > transitivity scale.
These two domains differ in a very important respect: whereas the
relational domain is normally fixed with respect to its intentionality
parameters, the referential domain is often variable. By this I mean
that 'verbs' are normally defined for their intentional scope through
encyclopedic knowledge whereas referential items can (logically) 'move
along' the intentionality scale (Fluid structures), as long as they have
the option of intentionality at all [which includes 'stones', 'storm',
etc. in a non-poetic/metaphorical reading]. This assumptions helps us to
infer that if a 'verb' is marked for intentional features (exceeding
their inherent intentionality grading), this is more likely done with
the help of specialized morphemes (e.g. in some Salishan languages).
Elements of the referential domain, on the other side, are more likely
to operate through 'hidden' behavioral patterns which allows the Fluid
operation more than a fixed morphology).

 From this it follows that the feature cluster related to the starting
point of 'force' (in a broader sense of force dynamics) [metaphorized to
'power'] both controls the intentionality grading and the way who this
grading is expressed (if it is expressed). If, for instance, this
cluster (or, in my terms, the vector {sem;synt;prag}) has a high value
for {synt} [grounding procedures etc.], then it is rather unlikely that
it is primarily metaphorized with respect to intentionality features (as
it is the case for German). But if the vector has high values for {sem}
or {prag} things may be different. For instance, in Udi (East Caucasian)
the ergative morpheme (covering parts of the agentive relation) has a
rather high value for {sem} - no wonder that it can be exploited for
playing the intentionality scale.

But again: if we ascribe a feature of (un)intentionality to any kind of
morpheme of morphosyntactic behavior, we have to show that this function
is semantically (and cognitively) correlated to the function of the
(less metaphorical) source domain. Anything else would be fiction [I
think Scott'S work has marvelously shown how arguments should run].

A final note: If we accept that force>power dynamics is a major feature
of the cognitive architecture of linguistic (clausal) expressions, we
should be aware of the fact that intentionality may also be crucial for
the 'antipode' of 'power dynamics', namely the Objective domain. If we
claim that the relation between A and O in transitive structures is
basically asymmetric (power verbs less in power), than we can assume
that the O domain normally is not involved in features of
intentionality. Yet, some languages, e.g. some Papuan languages, but
also - I think - Athapaskan and Salishan, show that the O domain can
intervene in this standard asymmetry: in such instances, the referential
O domain establishes some kind of counter force (or: counter power) that
may results from an intentional counter'action' (remember the famous
'finally managed to' principle in some Salishan languages). It such
languages, the O domain may be associated with features of
intentionality that operate against the dimension of intentionality
associated with A. This aspect becomes even more crucial, if we include
the so-called IO-domain (primary or indirect objects). esp. in Papuan

Let me stop here (as I said: I'm not quite sure whether I really met the
current thread - if I didn't, please excuse).

Best wishes [and a big thank you for your patience],


Prof. Dr. Wolfgang Schulze
Institut für Allgemeine und Indogermanische Sprachwissenschaft
Universität München - Geschwister-Scholl-Platz 1 - D-80539 München
Tel.: ++49-(0)89-2180 2486 (secretary) // ++49-(0)89-2180 5343 (office)
Fax:  ++49-(0)89-2180 5345
Email:  W.Schulze at

More information about the Funknet mailing list