LEVELS OF REPRESENTATION IN NATURAI, LANGUAGE BASED INFORMATION 
SYSTEMS AND THEIR RELATION TO THE METHODOI,OGY OF COMPUTATIONAL LINGUISTICS 
G. ZIFONUN, INSTITUT FUER DEUTSCHE SPRACHE, 
D-6800 MANNHEIM, FEDERAL REPUBI.IC GERMANY 
Summar L 
In this paper the methodological ba- 
sis of the 'computational linguistics ap- 
proach' for representing the meaning of 
natural language sentences is investiga- 
ted. Its adherance to principles of for- 
mal linguistics and formal philosophy of 
language like the 'separation of levels 
of syntactic and semantic analysis', and 
the "Fregean" principle may be contrast- 
ed with the 'artificial intelligence ap- 
proach'. A "Montague" style method of 
mapping the syntax of natural language 
onto the syntax of the 'semantic lan- 
guage' used as the means of internal re- 
presentation in the information system 
PLIDIS is presented. Rules for defining 
subsequent levels of representation like 
'syntax-interpretative level', 'redundancy' 
free level' are given. 
Introduction 
The present paper presents ideas 
concerning a methodology of the 'seman- 
tics in computational linguistics' (COL- 
semantics). 
There is the following hypothesis 
underlying: 
In the field of COL-semantics algo- 
rithms and computer programs are devel- 
oped which deliver structures of lingui- 
stic analysis and representation that 
can be compared with those of formal lin- 
guistic semantics and satisfy the ade- 
quacy criteria of certain linguistic the- 
ories. They therefore are suitable in- 
struments for developing and testing such 
theories. 
COL-semantics hence proceeds in a 
way different from the semantic process- 
ing as it is found in the framework of 
artificial intelligence (AI-semantics). 
AI-semantics is not so much linked to 
the semantics of formal linguistics or 
logic but rather to cognitive psycholo- 
gy, problem solving theory and the the- 
ory of knowledge representation which has 
been recently put forward within AI it- 
self. 1 Between both branches of semantic 
processing of natural language that are 
realized in computer systems there there- 
fore exists a difference in aims, the- 
ories and methods. 
Starting from a brief sketch ot the 
aims and theories of both approaches one 
essential methodological principle of 
COL-semantics will be elaborated in the 
second chapter of the paper. In the third 
chapter COL-semantic methods will be ex- 
emplified by a concrete application, the 
production of semantic representations in 
an information system. Stress will notbe 
laid on the question of w h a t a COL- 
semantic representation should look like 
but h o w levels of a semantic represen- 
tation can be systematically relatedwith 
natural language and with each other. 
Aims and theoretical concept__ E 
of COL-semantics and AI-semantics 
The difference of aims and methods 
can only be outlined here as far as it is 
relevant with respect to the methodologi- 
ca1 divergence which will be dealt with 
in detail: Aim of AI-semantics is the si- 
mulation of the human language understand- 
ing and/or language generating process 
that is to be understood as a manifesta- 
tion of intelligent human problem solv- 
ing behaviour. Aim of COL-semantics is 
the algorithmic generation of descrip- 
tive structures (of a generativ-semantic, 
interpretative, logico-semantic or other 
type) out of a given natural language in- 
put. Both purposes can be partial aims 
or intermediate steps within a larger 
project like 'simulation of dialogue be- 
haviour', 'natural language information 
or question answering system'. 
Thus the AI-approach leads to a the- 
ory where the object of explanation (or 
simulation) is "rational human behaviour ''2 
or more specifically human language be- 
haviour as a rational psychic process, 
whereas in the theory of linguistic se- 
mantics language is being objectified as 
a generated structure or a system which 
can be considered independently from the 
associated mental processes. In lingui- 
stic semantics and also in COL-semantics 
meta-linguistic notions which refer to 
language as a system like 'synonymy', 
'equivalence' and (particularly in the 
formal linguistics based on logic)'truth' 
and 'entailment' are crucial; in AI-se- 
mantics however we have the 'behaviour' 
oriented conce~ts of 'inferencing','dis- 
ambiguating', "reasoning', 'planning' 
etc o 
A methodological principle of 
COL-semantics 
A distinctive feature of lingui- 
stics, especially logico-linguistic the- 
ories, is the separation of different 
"expression" and "content" levels of ana- 
lysis and representation and the speci- 
-202 
fication of mapping rules between them 
(surface structure versus deep structure, 
syntactic structure versus semantic struc- 
ture). In Montague grammar this differen- 
tiation between a well defined syntactic 
level and an also well defined semantic 
level of description is a methodological- 
ly necessary consequence of the "Fregean" 
principle. The Fregean principle states 
that the meaning of an expression can be 
determined on the basis of the meanings 
of its logically simple constituent ex- 
pressions and the syntactic structure of 
the whole expression. This principle has 
been revived by Montague and has ~eenre- 
alized in his theory of language in such 
a way that the syntactic and the seman- 
tic structure of a natural language ex- 
pression are respectively represented as 
expressions of formal systems (syntax and 
meaning algebras) between which systems 
there exist well defined formal relation- 
ships (homomorphisms). 
When this concept is transferred to 
the operationalizing of linguistic analy- 
sis in a computer system it will be ex- 
cluded to conceive the mapping from nat- 
ural language into semantic representa- 
tion as a simple integrated pass, where 
in the course of parsing a sentence the 
valid semantic interpretation is assigned 
to each occurring item or group of items 
and where the possibilities of inference 
and association with stored background 
knowledge are flocally f realized without 
ever generating a full syntactic analysis. 
Saving an explicit level of syntactic re- 
presentation seems to be compatible with 
the Fregean principle only under the con- 
dition that the algorithm incorporates a 
grammar (in the technical sense of a con- 
sistent set of generating or accepting 
syntactic rules), but for reasons of op- 
timization directly associates or applies 
semantic 'values' or 'rules' in process- 
ing the corresponding syntactic 'nodes' 
or 'rules '4, or even allows a semantic 
control of rule selection without leaving 
the parsing mode. This condition however 
is mostly not maintained in AI parsing ap- 
proaches where the one step processing is 
understood as a cognitively adequate ana- 
logue of human linguistic information pro- 
cessing and where even the terminal and 
non terminal symbols of the "grammar" are 
interpreted as semantic categories.5 
Syntactic and semantic represen- 
tation in an information system 
The way of processing natural lan- 
guage according to the principles of COL- 
semantics shall be demonstrated by the 
linguistic component of a natural language 
information system. The description is o- 
riented at the application area and the 
structure of the system PLIDIS (informa- 
tion system for controlling industrial 
water pollution, developed at the Insti- 
tut fuer deutsche Sprache, Mannheim). 6 
Giving only the over all structure of the 
system we have the following processings 
and levels: 
morphological analysis of natural 
language input ~ syntactic analysis (le- 
vel of syntactic representation) ~ trans- 
duction into formal representation lan- 
guage (level of semantic representation) 
interpretation (evaluation) against the 
database ~ answer generation 
The formal representation language 
is the language KS an extended first or- 
der predicate calculus, where the fea- 
tures going beyond predicate calculus are 
many sorted domain of individuals, lambda- 
abstraction and extended term building. 7 
In the following two aspects of the se- 
mantic representation will be treated: 
- the mapping between syntactical- 
ly analyzed natural language expressions 
and their KS counterparts will be inves- 
tigated 
- a differentiation between three 
levels of semantic representation will be 
accounted for: (level l) syntax-interpre- 
tative level, (level 2) canonical level, 
(level 3) database-related level, 
All three levels follow the same 
syntax, i.e. the syntax of KS and have 
the same compositional model theoretic se- 
mantics; they differ in their non logical 
constant symbols. 
_Mapping_natural language into the 
kemantic representation l~i!g~age KS 
In analogy with Montague's "theory 
of translation" in "Universal Grammar"we 
assume that the syntactic structures of 
natural language (NL, here German) and the 
semantic language (here KS) are similar, 
i.e. there exists a translation function 
f, such that the following holds: 
(l.l.) Given the categories of a 
categorial grammar of NL, f is mapping 
from these categories on the syntactic ca- 
tegories of KS. I.e. If m, ~I, ..., ~n are 
basic categories of German, then f(~), f (~I),..., f(#n) 
are syntactic categories of K$. 
If ~/~I/.../~n is a derived category (func- 
tor category) of NL, then f(~)/f(~1)/.../ f(~n) 
is a derived category of KS. 
(1.2.) If a is-an expression of ca- 
tegory 6 in NL (a6), then f(a) is an expres- 
sion of category f(6) in KS (f(a)f(6)). 
(1.3.) The concatenation of an ex- 
pression of the derived category m/~I/.../ 
#nwithexpressions of category ~1,...,#nre- 
sulting in an expression of category 
--203-- 
~/#I/.../~n ~ ~I ~ ... ~ Fn ~ 
is rendered in KS by the construction of 
a list 
\[f~/~/.../~n) Z(~) ... #(Fn)\] 
with the category ~'(~)(concatenation and 
list construction are defined for cate- 
gories instead of expressions in order to 
zmprove readability). 
Thus the 'transduction grammar' NL- 
KS is the triple 
< GNL, GKS, ~ > 
We now specify a minimal categoria\[ 
grammar of German GNL. A particular of 
GNL is the analysis of verbs as m-ary pre- 
dicates, i.e. in the categorial frame- 
worK, as functions from m NP into S 8 and 
the analogue treatment of nouns as func- 
tot categories 9 taking their attributes 
as arguments. 
Basic categories of NL 
S category of sentences 
O-N category of "saturated" common noun 
phrases 
NP category of noun phrases (singular 
terms) 
NPR category of proper nouns 
(If MNp is the set of noun phrases, 
MNp R the set of proper nouns 
MNPR C MNp 
holds.) 
derived categories of NL 
S/NP/.../NP category of m-ary verbs 
I J m times 
O-N/NP/.../NP category of common noun 
I I phrases taking n attri- 
n times butes 
NP/NP category of prepositions 
NP/O-N category of articles (deter- 
miners) 
syntactic rules (expansion of (1.3.), 
NL-part) 
(1) NP/NP ~ NP ~ NP 
(2) NP/O-N ~ O-N --~ NP 
(3) O-N/~P/.../N~ ~ NPI ~ ... NPn -"~" O-N 
i i 
n tlmes 
(4) S/~P/.../NF ~ N Pi~.?NPm -~ S 
m tlmes 
application of y to the basic categories: 
#(S) = FORMEL 
Z(O-N) = LAMBDAABSTRAKT 
~(NP) = TERM 
Z(NPR) = KONSTANTE, with MKONSTANTE 
MTERM 
to the derived categories: 
?(S/NP/.../NP )=,P(S)/?(NP)/.../#~(NP) = 
k.......>,,,__J FORMEL/TERM/.../TERM 
m tlmes for short: PRAED stel m 
~(O-N/NP/.../NP/= ,f(O-N)/~(NP)/.../~(NP): 
I I LAMBDAABSTRAKT/TERM/... 
V /TERM where LAMBDAAB- n times 
STRAKT itself i a func- 
tor category in KS: 
LAMBDAABSTRAKT : FORMEL 
/TERM 
#(NP/NP) = #(NP)/f(NP) = TERM/TERM 
#(NP /O-N) = #(NP)/f'(O-N) : TERM/LAMBDA- 
ABSTRAKTp for short: 
QUANT 
syntactic rules of KS (expansion of (1.3.) 
KS part) 
(I-KS) \[TERM/TERM TERM\] ~ TERM 
(Z-KS) \[TERM/LAMBDAABSIRAKT LAMBDA- 
ABSTRAKT\] ~ TERM 
for short : 
\[QUANT LAMBDAABSTRAKT \] ~ TERM 
(3-KS) \[ LAMBDAABSTRAKT/TERM/.../TERM 
L__ I \i- 
n times 
TERMi ... FERMn\] -~ LAMBDAABSTRAKT 
where an expression 
aLAMBDAABSTRAKT : aFORMEL/TER M 
is wrltten as 
\[LAMBDA x a ×\]. 
In a Lambdaabstrakt 
\[LAMBDA x \[al bl ... bn\]x\] 
al has the function of a n+1-ary predi- 
cate (PRAED), seen from the viewpoint of 
predicate calculus, such that we can re- 
write 
\[LAMBDA x \[al hl ... bn\]x\] as 
\[LAMBDA x \[al bl ... bn x\]\]. 
(4-K5) \[FORMEL/TERM/.../TERM 
l I 
m times 
TERMI • • • TERM m\] -~ FORMEL 
for short : 
\[PRAED stel m TERMI ... TERMm\] 
-~ FORMEL 
By applying the function ~ we have 
got a grammar GKS for our semantic lan- 
guage KS in an inductive way. We now give 
the following lexical correspondence rules 
for some non logical expressions of NL, 
taken from the application area of PLIDIS. 
204-- 
NL word NL cate- 
gory 
Probe (a) O-N/ 
("sample NP 
of sewage 
water") (b) O-N PROBE1 
enthalten S/NP/NP ENTHALT 
vorliegen S/NP/NP/NP VORLIEG 
der, die, NP/O-N JOTA 
das 
ein NP/O-N EIN 
bei NP/NP 'ID' 
(identity: 
\[ID aTERM\] 
= aTERM 
an NP/NP 'ID' 
in NP/NP "ID" 
Arsen NPR AS1 
Lauxmann NPR G-L 
Gehalt O-N/NP ENTHALTi 
KS transla- KS category 
tion 
PROBE LAMBDAABSTRAKT/ 
TERM 
LAMBDAABSTRAKT 
PRAED stel 2 
PRAED stel 3 
QUANT 
QUANT 
TERM/TERM 
TERM/TERN 
TERM/TERM 
KONSTANTE 
KONSTANTE 
LAMBDAABSTRAKT/ 
TERM 
With the given syntactic and lexical rules 
we can generate the following level I represen- 
tations of two natural language sentences: 
Enthielt die Probe bei tauxmann Arsen ? 
Did contain the sample from Lauxmonn arsenic ? 
(of polluted (name of a 
water) firm ) 
S/NP/NP NP/O-N O-NINP NP/NP NPR NPR 
ENTHALT .10TA PROBE 'ID' G-L AS1 
PRAED ste\[ 2 QUANT LAMBDAABSTRAKT/ TERM/ TERN TERM 
TERM TERM 
NP 
\[',o' ~-L\]TERM o ~'~T~M 
I 
O-N 
\[PROBE G- L\]LN, iBDAABS.nRAKi _= 
I 
NP ~OTA \[LA~OA x\[PROBE ~-k\] x\]ITER~ 
I 
I 
S 
\[ENTHALT~AEO stel 2 \[IOTA \[LAMBDA ×\[PROBe ~-L\] x\]\] TERM ASITER~ \]FORMEL 
(figure 1 ) 
O P 
o 
O A 
> 
> 
o 
~7 
z 
~D 
z 
~ z 
Z 
a. ~ - = 
• Z 
Y 
I 
N 
51 m 
,i 
(i Z h 
N 
° 
N 
z 
Y 
L ~ -~ . ~= 
= - .% 
(f±gure 2) 
Meaning postulates for generating 
~anonical representatlons 
Both sentences have received differ- 
ent representations on level I, they are 
nevertheless synonymous at least as far 
as the context of information seeking is 
concerned. 
An important principle in COL-se- 
mantics is the notion of structural (not 
lexical) synonymy. The following intui- 
tively valid synonymy postulates (meaning 
postulates) can be formulated. 
--205-- 
(1) A NL noun phrase containing n (n _> o) 
attributes (category O-N/NP/.../NP) 
I I -EYiih-gs 
is synonymous with an NP containing 
n+\] attributes, where the n+\]st at- 
tribute is an unspecified "place 
holder" attribute, under the precon- 
dition that the central noun of the 
NP systematically admitslOn+\] attri- 
butes : 
eine Probe is synonymous eine Probe bei 
with einem Betrieb 
('a sample ('a sample of an 
of sewage industrial plant' ) 
water ' ) 
The application of this principle 
may be iterated. 
(2 There are verb classes the elements 
of which have no descriptive meaning 
("non-content verbs"), in German the 
so called "Funktionsverben", the 
copula segn and others). In such ca- 
ses the NP as object or subject of 
the verb is the content bearer or 
'principal' NP, e.e. it becomes the 
predicate of the proposition. Such a 
sentence is synonymous with a cor- 
responding sentence containing a con- 
tent verb equivalent in meaning to 
the content bearing NP. For example: 
Arsengehalt liegt in 
der Probe vor. 
('There exists an 
arsenic content in 
the sample.') 
is synonymous 
with 
Die Probe enth~it Arsen. 
('The sample contains 
arsenic.') 
In such a non-content verb proposi- 
tion a noun phrase with a place hol- 
der attribute can also function as a 
"second order" principal NP, i.e. its 
unspecified attribute can be replaced 
by a "filler" NP, occurring as argu- 
ment of the non-content verb: 
Arsengehalt liegt bei Lauxmann in der Probe 
vor. is synonymous with 
Die Probe bei Lauxmann enthZlt Arsen. 
Both postulates shall be applied for 
transducing the level \] representations 
of NL sentences into level 2 representa- 
tions. We first give a definition of 
'principal term', i.e. the KS construc- 
tion corresponding to a 'principal NP'. 
(Def.) A principal term in a formula con- 
taining as PRAED the translation of anon 
content verb is a term that is capable, 
according to its semantic and syntactic 
structure, to embed other argument terms 
o~ the translation of the non content 
verb as its arguments. 
The operationalized version of the 
two principles is now after having shift- 
ed them onto the KS level: 
(1: maximality principle)When a NL-expres- 
sion has n analysis (n ~ 2J in level \] 
which only differ in the number of argu- 
ments, then the level 2 representation 
consists of the 'maximal' level I expres- 
sion, i.e. the expression containing the 
largest number of arguments. Any failing 
arguments are to be substituted by (ex- 
istentially bound~ variables. 
(2: transformation principle) 
(2.1.) When the PRAED of a formula is the 
translation of a non-content verb, at 
least one of its arguments must be a prin- 
cipal term. 
(2.2.) A formula containing the transla- 
tion of a non content verb must be trans- 
formed into an expression which contains 
the PRAED of a principal term as predi- 
cate iff there is an unambiguous mapping 
of the arguments of the translation of 
the non-content verb 
a) into arguments of a 
principal term 
or 
b) into a princapal term 
such that a well-formed formula of leve\] 
2 is obtained. 
We now state that PROBE and ENTHALT 
are 'maximal' expressions and PROBEI and 
ENTHALTI must be mapped into them re- 
spectively and that further holds: 
VORLIEG is the translation of the non- 
content verb vorliegen 
PROBE is the PRAED of a second order 
principal term with respect to 
a 'plant' argument 
ENTHALT is the PRAED of a principal term 
with respect to a 'sample' ar- 
gument 
Then the two examples of level I are 
mapped into a single representation on 
level 2: 
\[ENTHALT\[JOTA\[LAMBDA x\[PROBE G-L XJ\]\]AS1\] 
The reduction of synonymous structures in 
the canonical level of representation 
meets the criteria of economy as they are 
necessary in a computer system. II As we 
have tried to show, however,it can be 
based upon general linguistic principles 
and need not be imputed to the field of 
"world semantics". On the other side ad- 
mitting paraphrases as natural language 
input (as our examples are) improves the 
systems "cooperativeness" towards the 
user. In PLIDIS special aspects of the 
world model are accounted for in the le- 
vel 3 representations which mirror the 
relational structure of the data model 
to some extent. We can not go into the 
details of the relationship between level 
2 and level 3 ~or reasons of space. 
-206 
Comparison 
with other approaches 
Language processing systems that 
are oriented at Montague grammar or mo- 
del theoretic semantics are being devel- 
oped among others by Friedman et al., 
Sondheimer and the PHLIQAI group. A the- 
oretical discussion of the relationship 
between model theoretic semantics and AI- 
semantics can be found in Gunji and Sond- 
heimer cf. also Hobbs and Rosenschein 
St. Bien and Wilks (witha contrary vlew). 
The methodological ideas presented here 
are most closely related with the ap- 
proach of multi-level semantics pursued 
in PHLIQAi. But unlike the PHLIQAi ap- 
proach we regard the level(sJ of lingui- 
stic representation not only under the 
more formal aspect of syntax interpreta- 
tion but, as the last chapters show, we 
also take into account aspects of seman- 
tics of natural language word classes and 
structural synonymy. 
Notes 
1 There are certainly important inter- 
actions with empirial semantic work 
done in the last 10 years, soOrtony 
and Wilks stress the pervasive in- 
fluence of Fillmore. Like any other 
systematic distinction the one bet- 
ween formal llnguistic semantics and 
AI-semantics is somewhat simplifying: 
Within AI there are semantic approa- 
ches which are more or less oriented 
at formal logic, so the one of McCar- 
thy, Creary or Nash-Webber and Reiter 
and others. As typical AI-semantic ap- 
proaches we regard the ones of Schank 
and his colleagues, Wi±ks or Charniak 
(cf. for instance the articles in 
Charniak and Wilks). 
2 Hayes, 9 
3 Slightly exaggerating this tendency 
is formulated by Schank in Schank et 
al.):"Researchers in NPL (natural lan- 
guage processing in AI) have become 
less and less concerned with language 
issues per se. We are more interested 
in inferencing and memory models for 
example." (p. 1OO8) 
4 Such systems are presented for in- 
stance in Riesbeck, Norman and Rumel- 
hart, and even more programmatically 
in Schank et al., DeJong. Also in sys- 
tems conceived as data base interfaces 
like LIFER (Hendrix) and PLANES ~altz) 
"semantic"grammars are used. A theore- 
tical discussion on the role of syn- 
tax can be found in Schank et al. 
5 I.e. one has to check, whether in sys- 
tems containlng only "part grammars" 
or working with a syntactic "pre-pro- 
cessing" the syntactic rules which 
were effectively used, can be com- 
bined resulting in a coherent and con- 
sistent grammar. Questions of syntac- 
tic-semantic and purely semantic gram- 
mars underlying parsers are also dis- 
cussed from a theoretical point of 
view in Wahlster. 
The system PLIDIS is described in 
Kolvenbach, L6tscher and Lutz. 
The language KS ("Konstruktsprache") 
is described in Zifonun. 
Cresswell gives an analogous categorial 
description for verbs. Like in this 
minimal grammar in applying the rule 
of concatenation phenomena of word or- 
der are neglected. 
Keenan and Faltz introduce the cate- 
gory of "function noun" (in our frame- 
work O-N/NP) 
10 The vague condition of "systematical- 
ly admitting" is made concrete in 
PLIDIS by prescribing a semantic"sort" 
for each argument of a predicate. 
ii This reduction is done in PLIDIS with 
the help of meaning postulates which 
are interpreted by a theorem prover. 

References 

Bronnenberg, W.J.H.J./Bunt, H.C./Lands- 
bergen, S.P.J./Scha, R.J.H./Schoen- 
makers, W.J./van Utteren, E.P.C.:"The 
Question Answering System PHLIQAi", 
in: L. Bolc (ed.) "Natural Language 
Question Answering Systems" (Natural 
Communication with Computers), Mac- 
millan, London 1980, 217-305. 

Charniak, E./Wilks, Y. (eds.): "Computa- 
tional Semantics", ~ed. North Holland, 
Amsterdam 1976. 

Creary, L.G.: "Propositional Attitudes: 
Fregean Representation and Simulative 
Reasoning", Proc. 0th IdCAI Tokyo 1979, 
~76-182. 

Cresswell, M.J.: "Logics and Languages", 
Methuen, London 1973. 

DeJong, G.: "Predictional Substantiation: 
Two Processes that Comprise Under- 
standing", Proc. 6th IdCAI Tokyo 1979, 
217-222. 

Friedman, J./Moran, D.B./Warren, D.S.: 
"Explicit Finite Intensional Models 
for PTQ and An Interpretation System 
for Montague Grammar", American Jour- 
nal of Computational Linguistics Mi- 
crofiche 74, 1978~ 3-96. 

Gunji, T./Sondheimer, N.: "The Mutual Re- 
levance of Model-Theoretic Semantics 
and Artificial Intelligence", unpubl. 
paper, Department of Computer and In- 
formation Science The Ohio State Uni- 
versity, February 1979. 

Hayes, P.: "On the Difference between 
Psychology and Artificial Intelli- 
gence", AISB quarterly 34 July 1979, 
8-9. 

Hendrix, G.G.: "LIFER: A Natural Lan- 
guage Interface Facility", Tech. Note 
135, AI Center Stanford Research Inst., 
Menlo Park, California 1976. 

Hobbs, J.R./Rosenschein, S.J.: "Making 
Computational Sense of Montague's In- 
tensional Logic", Artificial Intelli- 
gence 9, \]978, 287-306. 

Keenan, E./Faltz, L.M.: "Logical Types 
for Natural Language", UCLA Occasio- 
nal Papers in Linguistics 3, Fall 
1978. 

Kolvenbach, M./L6tscher, A./Lutz, H. 
(eds.): "KOnstliche Intelligenz und 
natfirliche Sprache. Sprachverstehen 
und Probleml6sen mit dem Computer", 
Narr, Tfibingen 1979. 

McCarthy, J.: "First Order Theories of 
Individual Concepts and Propositions", 
in:D. Michie (ed.) Machine Intelli- 
gence 9, Edinburgh 1979. 

Montague, R.: "Formal Philosophy", ed. 
by R. Thomason, Yale University Press, 
New Haven and London 1974. 

Nash-Webber, B./Reiter, R.: "Anaphora and 
Logical Form: On Formal Representa- 
tions for Natural Language", Proc.5th 
IdCAI Cambridge Mass..1977, 121-131. 

Norman, D.A./Rumelhart, D.E. (eds.): "Ex- 
plorations in Cognition", Freeman, 
San Francisco 1975. 

Ortony, A./Wilks, Y.: "Cognitive Science 
versus Artificial Intelligence", AISB 
quarterly 34, April 1979, 20-22. 

Riesbeck, C.K.: "Conceptual Analysis", 
in: R.C. Schank (ed.) "Conceptual 
Information Processing", North Hol- 
land, Amsterdam 1975, 83-156. 

Schank, R.C./Lebowitz, M./Birnbaum, L.: 
"Parsing directly into Knowledge 
Structures", Proc. 6th IJCAI Tokyo 
1979, 772-777. 

Schank, R.C. et al.: "Panel on Natural 
Language Processing",Proc. 5th IJCAI 
Cambridge Mass. 1977, 1007-1013. 

Sondheimer, N.K./Gunji, T.: "Applying 
Model-Theoretic Semantics to Natural 
Language Understanding: Representa- 
tion and Question Answering", Proc. 
7th COLING Bergen 1978. 

St. Bien, J.: "Computational Lxplication 
of Intensionality", Preprints 6th 
COLING Ottawa 1976. 

Wahlster, W.: "ATN und semantisch-pragma- 
tische Analysesteuerung, in: T. Christal- 
ler/D. Metzing (ed.) "Augmented Tran- 
sition Network Grammatiken", vol I, 
Einhorn, Berlin 1979, 167-185. 

Waltz, D.L.: "An English Language Ques- 
tion Answering System for a Large Re- 
lational Database", CACM 21.7, July 
1978. 

Wilks, Y.: "Philosophy of Language", in: 
E. Charniak, Y. Wilks (eds.) Computa- 
tional Semantics, 2ed. North Holland, 
Amsterdam 1976, 205-233. 

Zifonun, G.: "Formale Repr~isentation na- 
tOrlichsprachlicher ~uBerungen", in: 
Kolvenbach, L6tscher, Lutz (eds.), 
Kfinstliche Intelligenz und natOrliche 
Sprache. Sprachverstehen und Problem- 
lSsen mit dem Computer, Narr, T~ibin- 
gen 1979, 93-134. 
