Preliminary Discussion Forum for the 3rd MIC Sorbonne workshop (Paris, Nov. 15-16, 2012)

New Standards for Language Studies
Nouveaux Standards pour les Sciences du Langage

You are not logged in.

Announcement

Announcement:
Researchers (linguists, computer scientists, neurologists, psychologists, sociologists, logicians and philosophers) who are interested in the topics specified in this forum may join us in this preliminary discussion forum by sending their name and surname, affiliation and e-mail address to: You will receive a free subscription account which will allow you to post content.

#1 28/03/12 04:00

Wlodarczyk A
Coordinator

Situation Frames - Semantics and Ontology

The basic assumptions of Associative Semantics are: (a) language is the main (but not the unique) means of communication between humans, whose brain memory limitations shape the representation structure of meaning, and (b) naming objects and situations (semantics) is interrelated with a subset of their being (ontology).

        In this theory, relations (structures), as expressed by utterances, are interrelated with situations (frames). Subsequently, however, I will understand the term “situation” to mean “correlated situation”, i. e. the situation as it can be represented using a specification language (e.g. XML or Description Logic formulae). Although the relationship between natural language and this specification will be omitted here (see the MIC-theoretical perspective), I should emphasise the fact that recently a certain degree of adequacy has been attained between this approach and psycho-neurological results regarding language.

        The notion of information is very complex. For this reason, humans transmit very small chunks of it during their communication acts. The role of the listener is to reconstruct (complete, elaborate, decompress) linguistic messages in order to understand them. I claim that the content of an utterance is always open-ended. For example, the utterance “Brutus killed Caesar” contains not very much more than the following contents:
   
  Meta-information: subject(Brutus), predicate(killed, object (Caesar))
           
  Information: sit-kill, active-role(Brutus), passive-role (Caesar)
   
        Observe that the passive version of the above utterance changes only on the meta-informative level and that the role of syntax is not just “formal” but pragmatic. It is clear then that the linguistic theory of meaning should give an account of this multi-dimensional (roles/situation, relevance/knowledge, salience/information, use/context etc.) character of communication.
             
  Take the classic case in which the verb “to kill” is used with 3 “participants”, as in the utterance “Brutus killed Caesar with a knife.”
             
  Traditionally, in the representation of such a triadic relation, there is one active role but two passive ones:
  
           
       Sit-kill, active-role (Brutus), passive-role (Caesar), passive-role (knife)
             
  In this case, it is however unclear whether the passive-role (knife) participates in this situation with Brutus as an active “partner” or someone/something else.
           
   Following empirical evidence, semantic situations as expressed by linguistic utterances never exceed 3 homogeneous terms: either (1) arguments (participants) or (2) anchors (locations), and it can be built of at most 4 mixed terms. I claim therefore that in order to model this problem, binary relations are sufficient and that the active and passive roles should be our proper starting point when building the ontology of the basic pair of abstract roles. Obviously, there is the third role that I named “median” but this is in fact a term common to the composition of the two associated (binary) situations. In the utterance “Brutus killed Caesar with a knife The “knife” plays a passive role in the relation “Brutus used knife” (this is suggested by the preposition “with”) and a pseudo-active role in the relation “the knife caused death to Caesar”.
             
  {sit-kill, active-role(Brutus), passive-role(Caesar),
           
    sit-use, active-role(Brutus), passive-role(knife),
           
    sit-cause-die, pseudo-active-role(knife), passive-role(Caesar)}
           
     The principle of associativity of (at most, binary) relations is more realistic with respect to natural language than the free cardinality of the arguments of Classical Logic.

     However, situations have their limits that I propose to call 'situation frames'. For the resolution of semantic (information) content the following ontological assumptions are needed (cf. the Polish scholars, logician R. Suszko 1998 and philosopher B. Wolniewicz 2003). Situation frames are determined by the two following functions:

     (a) a participation function (in order to enumerate the entities of the given situation)
     (b) a configuration function (in order to determine how the roles apply to the entities)

     I claim that additional (external and internal) properties of situation frames should also be introduced, namely Analysis and Control. In this way, situation frames can be partially characterised (independently of participants or anchors). The basic frame consists of a 3D space. It corresponds to the State. Additional dimensions are needed in order to 'transform' states into "slices of situations". When the Time dimension is present in the frame, we are dealing with Actions. Dynamic situations (actions) without Progression are called Events, with Progression they are Ordinary Processes and with Progression and Granularity they are Refined Processes.
    We obtain the following inclusive series of situation frames:

    STATES ⊆ EVENTS ⊆ ORD. PROCESSES ⊆ REF. PROCESSES

     In Logic, Modal and Temporal enhancements capture the nuances concerning the participants and the (temporal) anchors respectively. In the past CASK Project framework, we proposed to see linguistic modalities as effects (in cause-effect relationships) or conclusions (in premises-conclusion sequents). The above described characteristics of 'situation frames' lead directly to the definition of Aspectualities as another possible important enhancement of Logics and Linguistics.

     Together with the operations of Control (such as repetition, modification and composition), the semantic frames theory is a very useful concept for defining Aspectuality. Mutatis mutandis, without the concept of situation frames, it would be hardly conceivable why humans use gramiticised or lexical means for expressing aspectual categories in natural languages.


Links to the References:
     
         1.  "Roles and Anchors of Semantic Situations"

         2.  “Roles, Anchors and Other Things we Talk About : Associative Semantics and Meta-Informative Centering Theory”

Last edited by Wlodarczyk A (08/08/12 13:08)

Offline

#2 24/09/12 15:31

Wlodarczyk A
Coordinator

Re: Situation Frames - Semantics and Ontology

We received a message with the following question concerning the relationship between semantic arguments and centres of attention in the MIC theory. Namely, the question was about those sentences where two objects (direct and indirect) co-occur, those where adverbial and/or prepositional complements cooccur with direct objects. Are they cases of double Local CA? Are they to be regarded as members of a single Local CA?

Our anwser is:

Linguistic expressions contain linearly ordered components of diverse sorts. Thus, the representation of the semantic-and-pragmatic contents of a linguistic unit (utterance, paragraph, text) is not as simple as a tree-like structure (hierarchy). Neither is it a semi-lattice-like structure (heterarchy), though both are useful for this kind of partial (or local) representation.

However, we claim that both the meta-informative (as in base utterances) and the meta-meta-informative (as in extended utterances) status of communication provide the basis for a constituency (hierarchical) relationship between homogeneous (Subject-Predicate) and heterogeneous (Topic-Comment or Focus-Background) parts of utterances. In both cases, the leading concept is the attention-driven phrase (ADP) which we are going to introduce as new terminology and which is defined as the distinctive part of the representation as expressed by the given segment of an utterance. Hence, in this respect, the MIC theory is less a theory of predication than one of "generalised subjectivation/topicalisation".

It follows from the above explanation that in this theory, which continues to use the linguistic expression ('surface') though it uses diverse representation devices, whatever is expressed as an object (by the accusative case or by the first post-verbal position) is a LOCAL CA which corresponds the lateral ADP.

It is conceivable therefore that a semantic situation may be communicated as driven by no more than two kinds of CA (one global and one local) - at one and the same meta-informative (analogously at one and the same meta-meta-informative) level - even if there are more than two inferential semantic roles or anchors in the representation of the given utterance. Thus, there can be neither double local nor two-member single local CA. Nevertheless, when the analyst feels that there might be two local ADPs, there is no doubt that one of them will reflect either the topicalised or focalised phrase (the CA of the meta-meta-informative level), i.e. neither a "second" subject nor a "second" object.

Examples:

(1) John opened the door with his key. --- is a base utterance: "the door" is the Object, "with his key" is not meta-informatively centred.

(2a) John gave Mary a book. -- is a base utterance: "Mary" is an Object. The phrase "book" is not meta-informatively centred (it is not an object because it is not in the prime post verbal position).

(2b) John gave a book to Mary. -- is an extended utterance: "a book" is an Object and "to Mary" is a Focus of the beneficiary role.

Offline

Board footer

Powered by FluxBB