THE ℜEALIS MODEL OF HUMAN INTERPRETERS AND ITS

APPLICATION IN COMPUTATIONAL LINGUISTICS

Gábor Alberti, Márton Károly and Judit Kleiber

Department of Linguistics, University of Pécs, 6 Ifjúság Street, Pécs, Hungary

Keywords: Automated Discourse Analysis, Interpretation Modelling, Referents.

Abstract: As we strive for sophisticated machine translation and reliable information extraction, we have launched a

subproject pertaining to modelling human interpreters. The model is based on ℜeALIS, a new “post-

Montagovian” discourse-semantic theory concerning the formal interpretation of sentences constituting

coherent discourses, with a lifelong model of lexical, interpersonal and cultural / encyclopedic knowledge of

interpreters in its center including their reciprocal knowledge on each other. After the introduction of

ℜeALIS, we provide linguistic data in order to show that intelligent language processing requires a realistic

model of human interpreters. Then we put down some principles of the implementation (in progress) and

demonstrate how to apply our model in computational linguistics.

1 ℜEALIS: THE THEORY IN THE

BACKGROUND

ℜeALIS, REciprocal And Lifelong Interpretation

System, is a new “post-Montagovian” (Kamp et al.

2005) theory concerning the formal interpretation of

sentences constituting coherent discourses (Asher–

Lascarides 2003), with a lifelong model (Alberti

2000) of lexical, interpersonal and cultural/

encyclopedic knowledge of interpreters in its center

including their reciprocal knowledge on each other

(Alberti 2004).

The decisive theoretical feature of ℜeALIS lies

in a peculiar reconciliation of three objectives which

are all worth accomplishing in formal semantics but

could not be reconciled so far. The first aim

concerns the exact formal basis itself, which is often

mentioned as Montague’s Thesis: human languages

can be described as interpreted formal systems (we

thus does not agree with the viewpoint of Cognitive

Grammar: “That no attempt has yet been made to

formalize Cognitive Grammar reflects the judgment

that the cost of the requisite simplifications and

distortions would greatly outweigh any putative

benefits” (Langacker 200: 423)). The second aim

concerns compositionality, practically postulating

the existence of a homomorphism from syntax to

semantics. In Montague’s interpretation systems a

traditional logical representation played the role of

an intermediate level between the syntactic

representation and the world model, but Montague

argued that this intermediate level of representation

can, and should, be eliminated. The post-

Montagovian history of formal semantics, however,

seems to have proven the opposite, some principle of

“discourse representationalism”: “some level of

[intermediate] representation is indispensable in

modeling the interpretation of natural language”

(Dekker 2000).

The Thesis of ℜeALIS is that the two

fundamental Montagovian objectives can be

reconciled with the principle of “discourse

representationalism” – by embedding discourse

representations in the world model, getting rid of an

intermediate level of representation in this way

while preserving its content and relevant structural

characteristics. This idea can be carried out in the

larger-scale framework of embedding discourse

representations in the world model not directly but

as parts of the representations of interpreters’ minds,

i.e. that of their (permanently changing) information

states.

The frame of the mathematical definition of

ℜeALIS (whose 40 page long complete version is

available here: http://lingua.btk.pte.hu/realispapers)

is summarized here. As interpreters’ mind

representation is part of the

WORLD MODEL, the

definition of this model ℜ = 〈U, W

, W〉 is a quite

complex structure where

468

Alberti G., Károly M. and Kleiber J. (2010).

THE REALIS MODEL OF HUMAN INTERPRETERS AND ITS APPLICATION IN COMPUTATIONAL LINGUISTICS.

In Proceedings of the 5th International Conference on Software and Data Technologies, pages 468-474

DOI: 10.5220/0003035004680474

 SciTePress

¾ U is a countably infinite set: the

UNIVERSE

¾ W

= 〈U

, T, S, I, D, Ω, A〉: the EXTERNAL

WORLD

¾ W is a partial function from I×T

where W[i,t] is

a quintuple 〈U[i], σ[i,t]

, α[i,t]

, λ[i,t]

, κ[i,t]

〉:

the

INTERNAL-WORLD FUNCTION.

The external world consists of the following

components:

¾ U

is the EXTERNAL UNIVERSE (U

⊂ U), whose

elements are called

ENTITIES

¾ T = 〈T, Θ〉 is a structured set of

TEMPORAL

INTERVALS

¾ S = 〈S, Ξ〉 is a structured set of

SPATIAL ENTITIES

¾ I = 〈I, Υ〉 is a structured set of

INTERPRETERS

¾ D = 〈D, Δ〉 is a structured set of

LINGUISTIC

SIGNS

(practically performed morph-like

entities and bigger chunks of discourses)

¾ where T⊂U

, S⊂U

, I⊂U

, D⊂U

¾ Ω ⊂ T×U

* is the set of CORE RELATIONS (with

time intervals as the first argument of all core

relations)

¾ A is the

INFORMATION STRUCTURE of the

external world (which is nothing else but

relation structure Ω reformulated as a standard

simple information structure, as is defined in

Seligman–Moss (1997:245); its basic elements

are called the

INFONS OF THE EXTERNAL WORLD

The above mentioned internal-world function W is

defined as follows:

¾ The relation structure W[i,t] is called the

INTERNAL WORLD (or INFORMATION STATE) of

interpreter i at moment t

¾ U[i] ⊂ U is an infinite set: interpreter i’s

INTERNAL UNIVERSE (or the set of i’s

REFERENTS, or INTERNAL ENTITIES); U[i’] and

U[i”] are disjoint sets if i’ and i” are two

different interpreters

¾ what changes during an interpreter i’s lifespan is

not her referent set U[i] but only the four

relations among the (peg-like) referents, listed

below, which are called i’s

INTERNAL

FUNCTIONS

¾ σ[i,t]

: Π×U[i] → U[i] is a partial function: the

EVENTUALITY FUNCTION (where Π is a complex

label characterizing argument types of

predicates)

¾ α[i,t]

: Ψ×U[i] → U[i]∪U

is another partial

function: the

ANCHORING FUNCTION (α

practically identifies referents, and Ψ contains

complex labels referring to the legitimizing

grammatical factors)

¾ λ[i,t]

: Λ×U[i] → U[i] is a third partial

function: the

LEVEL FUNCTION (where elements

of Λ are called

LEVEL LABELS); the level

function is intended to capture the “box

hierarchy” among referents in complex Kampian

DRS boxes (Kamp et al. 2005) enriched with

some rhetorical hierarchy in the style of SDRT

(Asher–Lascarides 2003)

¾ κ[i,t]

: Κ → U[i] is also a partial function: the

CURSOR, which points to certain temporary

reference points prominently relevant to the

interpreter such as “Now”, “Here”, “Ego”,

“Then”, “There”, “You”

¾ The temporary states of these four internal

functions above an interpreter’s internal universe

serve as her “agent model” in the process of

(static and dynamic) interpretation.

Suppose the information structure A of the external

world (defined above as a part of model

ℜ

〈

〉

)

) contains the following infon: ι = 〈

PERCEIVE, t, i,

j, d, s〉, where i and j are interpreters, t is a point of

time, s is a spatial entity, d is a discourse (chunk),

and

PERCEIVE is a distinguished core relation (i.e. an

element of Ω). The

INTERPRETATION of this

“perceived” discourse d can be defined in our model

relative to an external world W

and internal world

W[i,t].

The

DYNAMIC INTERPRETATION of discourse d is

essentially a mapping from W[i,t], which is a

temporary information state of interpreter i, to

another (potential) information state of the same

interpreter that is an extension of W[i,t]; which

practically means that the above mentioned four

internal functions (σ, α, λ, κ) are to be developed

monotonically by simultaneous recursion,

expressing the addition of the information stored by

discourse d to that stored in W[i,t].

The new value of eventuality function σ chiefly

depends on the lexical items retrieved from the

interpreter’s internal mental lexicon as a result of the

perception and recognition of the words /

morphemes of the interpreter’s mother tongue in

discourse d. This process of the identification of

lexical items can be regarded as the first phase of the

dynamic interpretation of (a sentence of) d. In our

ℜeALIS framework, extending function σ

corresponds to the process of accumulating DRS

condition rows containing referents which are all –

still – regarded as different from each other.

It will be the next phase of dynamic

interpretation to anchor these referents to each other

(by function α) on the basis of different grammatical

relations which can be established due to the

recognized order of morphs / words in discourse d

and the case, agreement and other markers it

contains. In our approach two referents will never

have been identified (or deleted), they will only be

THE REALIS MODEL OF HUMAN INTERPRETERS AND ITS APPLICATION IN COMPUTATIONAL

LINGUISTICS

469

anchored to each other; but this anchoring

essentially corresponds to the identification of

referents in DRSs.

The third phase in this simplified description of

the process of dynamic interpretation concerns the

third internal function, λ, the level function. This

function is responsible for the expression of intra-

and inter-sentential scope hierarchy (Reyle 1993) /

information structure (Szabolcsi 1997) / rhetorical

structure (Asher–Lascarides 2003), including the

embedding of sentences, one after the other, in the

currently given information state by means of

rhetorical relations more or less in the way

suggested in SDRT.

It is to be mentioned that the information-state

changing dynamic interpretation and the truth-value

calculating static interpretation are mutually based

upon each other. On the one hand, static

interpretation operates on the representation of

sentences (of discourses) which is nothing else but

the output result of dynamic interpretation. On the

other hand, however, the above discussed phases of

dynamic interpretation (and chiefly the third phase)

include subprocesses requiring static interpretation:

certain presuppositions are to be verified (Kamp et

al. 2005).

The interpreter’s fourth internal function, cursor

κ, plays certain roles during the whole process of

dynamic interpretation. Aspect, for instance, can be

captured in our approach as the resetting or retaining

of the temporal cursor value as a result of the

interpretation of a sentence (→ non-progressive /

progressive aspect, respectively). It can be said in

general that the input cursor values have a

considerable effect on the embedding of the “new

information” carried by a sentence in the

interpreter’s current information state and then this

embedding will affect the output cursor values.

YNAMIC INTERPRETATION in a ℜeALIS model

ℜ=〈U, W

, W〉, thus, is a partial function Dyn which

maps a (potential) information state W° to a

discourse d and an information state W[i,t] (of an

interpreter i):

Dyn(d) : 〈ℜ,W[i,t]〉 a 〈W°, e°

, U°〉,

where U°, shown up in the output triple, is the

COST

of the given dynamic interpretation (coming from

presuppositions legitimized by accommodation

instead of verification), and e°

is the eventuality that

the output cursor points to (this is the eventuality to

be regarded as representing the content of discourse

d). Function Dyn(d) is partial: where there is no

output value, the discourse is claimed to be ill-

formed in the given context. Due to the application

of cost, ill-formedness is practically a gradual

category in ℜeALIS.

The

STATIC INTERPRETATION of a discourse d is

nothing else but the static interpretation of the

eventuality referent that represents it. The recursive

definition of static interpretation is finally based

upon anchoring internal entities of interpreters to

external entities in the external universe, and

advances from smaller units of (the sentences of) the

discourse towards more complex units.

2 SENTENCES AND

DISCOURSES

Let us take the problem of translation. For example,

a Hungarian text (1a) can only be translated by

someone who has the Then (1b-c) and Now (1d)

cursor values while being aware of the world around

him/her.

Example 1: Knowledge about the world and the

temporal cursors.

a. Megjelen-t az elnök, de nem válaszol-t a

kérdés-ek-re.

appear-Past the president but he/she not answer-Past the

questions-Pl-onto

b. The President appeared but he/she answered no

questions.

c. Der Präsident/Die Präsident-in stell-te sich ein,

aber er/sie hat keine Frage-n be-antwort-et.

the president / the president-Female to.place-Impf Refl in,

but he/she has no question-s prefix-answer-Perf

d. The President will appear but he/she is not going

to answer any questions.

In example (2a) below, the temporal cursor will

jump (forward), just like the spatial cursor: the

fridge is taken to be the one at Peter’s home (2b).

When describing states, we shall keep the position of

the temporal (and the spatial) cursor (2c). Besides

fitting patterns onto the real world, the worldlet

concerning the intention of the actor and that of the

expectation of the speaker also play a role. The

temporal cursor can stand not only in the cumulative

phase, but also in the preparation phase of the given

eventuality type (2d); the event mentioned in the

discourse will never happen but it is there in the

worldlet concerning the belief of the Patient.

Example 2: Progressive aspect, imperfective

paradox, result state – the interpreter’s cursors and

worldlets.

a. Peter travelled home. He drank a beer.

b. Peter travelled home. He wanted to drink a beer

but the fridge was empty.

ICSOFT 2010 - 5th International Conference on Software and Data Technologies

470

c. Peter was travelling home. He was drinking a beer.

The interpreter has numerous “famous” referents in

his/her cultural/encyclopedic knowledge (stored in

an appropriate hierarchy structured by functions α

and λ, demonstrated in Section 1), which can be

invoked by a name (3a). Although a name can refer

to another entity (3b). A rich ontology concerning

the world is also available (3c). The interpreter also

stores non-logical relations (3d), to be applied while

building stories again and again —typically based on

already-built similar stories (3d-e)) and while

searching for contacts between temporal, eventual

and normal referents. Discourse (3e) is ambiguous:

if the eventuality referent e” belonging to John’s

pushing Peter is taken to stand in a Narrative relation

with referent e’ of Peter’s falling, the temporal

referent t” belonging to e” follows t’ (belonging to

e’) chronologically, whilst if e” is construed as the

Reason of e’ then t” precedes t’ (Asher and

Lascarides 2005). A topic cursor can also play a role

while building discourses (3f): it is made explicit in

Hungarian by the lack or presence of a pronoun

which participant is taken to be the topic of a

sentence relative to the preceding sentence.

Example 3: Different sorts of knowledge

a. Mozart had a powerful influence on the work of

Beethoven. Beethoven knew much of Mozart's work.

b. J. G. Leopold Mozart (November 14, 1719 – May

28, 1787) was a composer, conductor, teacher, and

violinist. Mozart is best known today as the father

and teacher of Wolfgang Amadeus Mozart.

c. a I have a half-St. Bernard and half-Scottish

Shepherd, a Dalmatian and a parrot. The two dogs

often frighten the poor bird.

d. Peter married yesterday. The priest spoke very

harshly.

e. Peter fell. John pushed him.

f. Péter-nek van egy unokahúg-a.

Kedvel-i őt. / Az kedvel-i őt.

Peter-DAT is a niece-PossSg3

Like-Sg3def him/her / That like-Sg3def him/her

’Peter has a niece. He likes her. / She likes him.’

Our last set of examples concerns the rich and

explicit Hungarian system of operators to be

interpreted logically (Kiss 2001). Based on his/her

background knowledge and the “relevant set” as a

part of it, one can infer the presence and place of

some unnamed participants from the operator and

the named participants of the discourse.

Example 4: Operators and claims about the relevant

participants not mentioned in the discourse

a. Tizenkét unokatestvér-em van, de csak Annát és Beá-t

hív-t-am meg a születésnap-i parti-m-ra.

twelve cousin-PossSg1 is but only Ann-ACC and Bea-ACC

invite-Past-Sg1def Perf the birthday-DerAdj party-PossSg1-onto

‘I have twelve cousins but I invited only Ann and

Beatrice to my birthday party.’

b. Lát-om, a ↑nővér-em-et (bezzeg) meg-hív-t-ad!

see-Sg1def the sister-PossSg1-ACC (contr.top.) Perf-invite-Past-Sg2def

‘But, as for my sister, I see that you invited her!’

c. Meg-hív-hat-t-ál volna mindannyiunk-at!

Perf-invite-may-Past-Sg2 PastCond all.of.us-ACC

‘You could have invited all of us.’

d. Meglep, hogy a nővér-em-et is meg-hív-t-ad.

surprise that the sister-PossSg1-ACC also Perf-invite-Past-Sg2def

‘It surprises me that you invited my sister, too.’

Table 1 summarizes the logical implicature of the

Hungarian operators (apart from topic (3f), whose

interpretation is not of logical nature). Let ‘every’

(4c) be our starting-point: this operator practically

retrieves the set of participants mentioned earlier as

the ‘relevant set’ e.g. ‘all of us’), and it is claimed

that what is predicated is predicated of each member

of the relevant set. Operator ‘also’ (4d) refers to the

existence of an unnamed participant satisfying what

is predicated (at least according to the speaker). The

contrastive topic (4b) refers to the existence of an

unnamed participant not satisfying what is

predicated, whilst the focus (4a) refers to the fact

that each unnamed participant is such that he/she/it

does not satisfy what is predicated.

Table 1: The system of Hungarian operator meanings

= R \ R

, where R

: mentioned participants, R: every

participant which could have played the role played by the

mentioned participants).

P(x)

¬P(x)

∃

x ∈ R

operator ‘also’ (4d) contrastive topic

(4b)

∀

x ∈ R

operator ‘every’

(4c)

focus (4a)

3 PRINCIPLES OF THE

IMPLEMENTATION

The most basic data elements represented in the

implementation are the referents, which are assigned

to the grammatical components during the process of

the semantic analysis. In this article, which

demonstrates a “work in progress”, we shall only put

down the principles of a potential implementation

(in a greatly simplified manner), focusing onto the

four basic functions (σ, α, λ and κ; see Section 1)

THE REALIS MODEL OF HUMAN INTERPRETERS AND ITS APPLICATION IN COMPUTATIONAL

LINGUISTICS

471

and mention some possible applications of such

discourse-analyzing systems.

By default, each part of the sentence has its own

referents which do not depend on any referents

belonging to other parts of the sentence (the function

α will search for dependencies). Most of these

referents (belonging to certain interpreters in the

richly structured worldlet system defined in Section

1: r

, r

,...) refer to entities existing in the external

world, while other referents can be eventual (e

temporal (t

) or spatial (s

An eventual referent must be assigned to all

verbs, nouns, adjectives and adverbs (parts of speech

which can, in principle, play a predicative role in the

sentence). For example, the noun banker and the

adjective clever both can be treated as eventualities

“being a banker” and “being clever”. Function σ,

taking the label Π into account (which, as mentioned

in Section 1, contains information on the already-

analyzed syntactic structures), assigns the

argumental (r

), temporal and spatial referents to the

eventual referent of the regent and forms the

structure pred(e, t, s, r

, r

, …) (one should note that

only a small part of the referents play a significant

role in the actual discourse).

The anchoring function α assigns the referents to

each other in different worldlets, thereby declaring

some referents as identical, which induces an

equivalence relation. This process, however, should

be aided by an ontology (apart from the information

collected during the syntactic analysis) as mentioned

in section 2. The ontology can have an arbitrary

structure – for example, if we take the basics of

psycholinguistics into account, the semantic web of

the mental lexicon could be modeled by a neuron

network (and, in general, RL could be used to handle

lexical semantics). The mental lexicon has multiple

dimensions on its own, and these dimensions can be

regarded as multiple relation types in the network –

or, even more precisely, a network of networks. If

we have no ontology at all, only the referents of

literally identical syntactic entities (and, in some

cases, pronouns) can be anchored to each other.

The ontology is also used when we try to

determine whether the discourse being analyzed is

coherent or not. In (3d), the computerized equivalent

of the semantic web should be used to determine

coherence (by using a metric to measure the distance

between the concepts (semantic categories)

WEDDING and PRIEST). If the (semantic) distance

is sufficiently low, the discourse can be regarded as

coherent. Of course, if we use RL technology to

create our ontology, the semantic proximity between

two (or more) concepts must be pre-taught.

Summarizing the above, the actual identity of

referents can be determined by using the relation

types (such as synonimity, semantic priming etc.)

and distance metrics in the network.

The set of temporal referents (t

) can be regarded

as a partially ordered set (see, say, (3e)). As we

mentioned above, referents can be identified by

using a distance metric on the semantic web, thereby

fuzzifying the process of determining discourse

coherence. Many temporal adverbs are fuzzy by

their nature (e.g. nowadays, a long time ago, shortly)

and they are not even culturally independent. But

even if we only handle well-defined temporal

referents, the system still has some important

applications (see Section 4 below).

The level function λ (practically) assigns the

referents to certain worldlets of interpreters (such as

those of their beliefs, desires, intentions, dreams).

The entities of the external (real) world (model)

always exist and they are “seen” and referenced by

all interpreters. But during the syntactic analysis,

level-changing words (mostly adverbs or particles)

are found, expressing modality. “If only Mary had a

car! Me, too, could drive it occasionally.”

The phrase “if only” refers to the speaker’s

desire, rendering all referents in its scope to a

different level, also a different worldlet (expressing a

desire of the speaker). Entities on this level do not

exist in the real world but the speaker must refer to

them in order to express his/her desire.

Certain values of the cursor κ can be regarded as

quasi-constant since they do not depend on the

actual discourse flow directly and, in most cases,

they do not need to be set during the analysis.

“Then” is set at the beginning of a story and, for

example, “You” can point to the user, who (as an

agent) must always be considered as an active

participant in the discourses being analyzed. Many

applications of the system are based on this. We will

show some of them in the next section.

4 APPLICATONS AND PLANS

As we mentioned above, the aim of our ℜeALIS

model is automatic discourse analysis. To do this,

we need to implement all the above-described tasks

and functions. But why should an interpreter based

on ℜeALIS be implemented?

First, expert systems can be created by

implementing the ℜeALIS model. This depends on

the ontology on which the function α is based. The

ontologies need not to include everything or be over-

ICSOFT 2010 - 5th International Conference on Software and Data Technologies

472

complicated. For example, ontologies concerning a

special field combined with the ℜeALIS model

(which is responsible for the syntactic and semantic

analysis) could form an expert system or a decision-

supporting system together. Questions could be

asked or predicates could be stated to the program in

a natural language – the system extracts information

and prints it in a readable form (preferably also in a

natural language, for example, if we want to do

machine translation backed by ℜeALIS).

One possible application of the ℜeALIS model is

to use it as a legal expert system to aid lawyers.

Backed by a legal ontology, temporal referents (t

)

could even be handled in a simplified way (see

above), because temporal adverbs tend to be much

less fuzzy in legal texts than in general stories. For

example, confessions and evidences given by

participants of a court case could be analyzed

according to the current laws (which are integrated

into the ontology) to facilitate judgement, or even

laypersons could use the system if they consider

taking legal action.

Machine translation, too, can be based on

ℜeALIS. It was already implemented, although in a

greatly simplified manner, in its predecessor (Alberti

et al. 2004) which was able to translate simple

Hungarian sentences into grammatically correct

English. In the process of translating entire

discourses, however, references always play a key

role, as illustrated in Section 2. Translating certain

pronouns is only possible after having analyzed

large parts of the discourse while recording the

position of the topic cursor (3f). The same applies

for the tense and aspect system of certain languages.

Here, the precise handling of the temporal cursor

seems to be far more problematic than in the case of

doing a “mere” discourse analysis (e.g. when

functioning as an expert system). If the actual

position of the temporal cursor can not be exactly

reproduced in the target language, the translation

process is especially difficult, if at all possible.

5 CONCLUSIONS

We have based a subproject pertaining to the

modeling of human interpreters (of our project

whose chief aims are machine translation and

information extraction) upon ℜeALIS, REciprocal

And Lifelong Interpretation System (Section 1),

because a large scale of linguistic data (Section 2)

shows that intelligent language processing requires a

realistic model of human interpreters. Then we put

down some principles of the implementation (in

progress) and sketch how to apply our model in

computational linguistics (Sections 3-4).

The initial state of the model of our ideal

interpreter’s mind can practically be regarded as an

enormous, unstructured set of peg-like referents in

Landman’s style (1986), which is then permanently

being enriched, due to the input of linguistic

information (to be worked up in different ways),

with an intricate structure “spanned” by four

functions, σ, α, λ and κ, responsible for,

respectively, the assignment of eventuality referents

to statements about temporal, spatial and “normal”

referents (σ), the identification of co-referring ones

(α), the decision of a scopal/modal relation system

among the referents (λ), and the highlighting of

those playing some distinguished role at a certain

moment of working up a discourse (κ).

ACKNOWLEDGEMENTS

We are grateful to OTKA 60595 for their financial

help.

REFERENCES

Alberti, G. 2000: Lifelong Discourse Representation

Structures, Gothenburg Papers in Computational

Linguistics 00–5, 13–20.

Alberti, G. 2004: ReAL Interpretation System, in L.

Hunyadi – Gy. Rákosi – E. Tóth eds.: Preliminary

Papers of the Eighth Symposium on Logic and

Language, Univ. of Debrecen, 1–12.

Alberti, G., Kleiber, J. Viszket, A. 2004: GeLexi project:

Sentence Parsing Based on a GEnerative LEXIcon, in

Acta Cybernetica 16, 587–600.

Asher, N. – A. Lascarides 2003: Logics of Conversation,

Cambridge Univ. Press.

Dekker, P. 2000: Coreference and Representationalism, in

von Heusinger – Egli eds.: Reference and Anaphoric

Relations.

Kamp, H. – J. van Genabith – U. Reyle 2005: Discourse

Representation Theory, http://www.ims.uni-

stuttgart.de/~hans.

Kiss, É. K. 2001: Hungarian Syntax. Cambridge Univ.

Press, Cambridge.

Landman, F. 1986: Towards a Theory of Information,

Foris, Dordrecht.

Langacker, R. W. 200: Cognitive Grammar, Geeraerts, D.

– H. Cuyckens eds.: The Oxford Handbook of

Cognitive Linguistics, Oxford Univ. Press, 421–462.

Reyle, U. 1993: Dealing with Ambiguities by

Underspecification, J. of Semantics 10, 123–179.

THE REALIS MODEL OF HUMAN INTERPRETERS AND ITS APPLICATION IN COMPUTATIONAL

LINGUISTICS

473

Seligman, J. – L. S. Moss 1997: Situation Theory, in van

Benthem, J. – A. ter Meulen eds: Handbook of Logic

and Language, 239–30.

Szabolcsi, A. ed. 1997: Ways of Scope Taking, SLAP 65,

Kluwer, Dordrecht.

ICSOFT 2010 - 5th International Conference on Software and Data Technologies

474