The Dutch not only have what must be the greatest number of linguists per capita in the world,

they also have a very long and rich tradition of combining linguistics, logic, and philosophy of
language. So it should not be a surprise that it is an interdisciplinary collaboration of Dutch
scholars that has produced the first comprehensive introduction to logic, language, and meaning
that includes on the one hand a very fine introduction to logic, starting from the beginning, and
on the other hand brings up at every point connections to the study of meaning in natural
language, and thus serves as an excellent introduction and logical background to many of the
central concerns of semantics and the philosophy of language as well. This book is
pedagogically beautifully designed, with the central developments very carefully introduced and
richly augmented with examples and exercises, and with a wealth of related optional material
that can be included or omitted for different kinds of courses (or self-teaching) for which the
book could very well be used: I could imagine tailoring very fine but slightly different courses
from it for inclusion in a linguistics curriculum, a philosophy curriculum, a cognitive science
curriculum, or an AI/computational linguistics program. It would be less suitable for a logic
course within a mathematics department, since there is less emphasis on proofs and
metamathematics than in a more mathematically oriented logic book. There is certainly no lack
of rigor, however; I think the authors have done a superb job of combining pedagogical
user-friendliness with the greatest attention to rigor where it matters. One very noticeable
difference from familiar introductory logic texts is the inclusion of accessible introductions to
many nonstandard topics in logic, ranging from approaches to presupposition and many-valued
logics to issues in the foundations of model theory, and a wide range of more advanced (but still
very accessible) topics in volume 2. The book thereby gives the student an invaluable
perspective on the field of logic as an active area of growth, development, and controversy, and
not simply a repository of a single set of eternal axioms and theorems. Volume 2 provides an
OUtstanding introduction to the interdisciplinary concerns of logic and semantics, including a
good introduction to the basics of Montague grammar and model-theoretic semantics more
generally. x Foreword I first became acquainted with this book in its Dutch version during a
sabbatical leave in the Netherlands in 1982-83; it made me very glad to have learned Dutch, to
be able to appreciate what a wonderful book it was, but at the same time sorry not to be able to
use it immediately back home. I started lobbying then for it to be translated into English, and I'm
delighted that this has become a reality. I hope English-speaking teachers and students will
appreciate the book as much as I anticipate they will. The authors are top scholars and leaders
in their fields, and I believe they have created a text that will give beginning students the best
possible entry into the subject matter treated here. BARBARA H. PARTEE Logic, Language,
and Meaning consists of two volumes which may be read independently of each other: volume I,
An Introduction to Logic, and volume 2, Intensional Logic and Logical Grammar. Together they
comprise a survey of modern logic from the perspective of the analysis of natural language.
They represent the combined efforts of two logicians, two philosophers, and one linguist. An
attempt has been made to integrate the contributions of these different disciplines into a single
consistent whole. This enterprise was inspired by a conviction shared by all of the authors,
namely, that logic and language arc inseparable, particularly when it comes to the analysis of
meaning. Combined research into logic and language is a philosophical tradition which can be
traced back as far as Aristotle. The advent of mathematical logic on the one hand and
structuralist linguistics on the other were to give rise to a period of separate development, but as
these disciplines have matured, their mutual relevance has again become apparent. A new
interdisciplinary region has emerged around the borders of philosophy, logic, and linguistics, and
Logic, Language, and Meaning is an introduction to this field. Thus volume 1 establishes a
sound basis in classical propositional and predicate logic. Volume 2 extends this basis with a
survey of a number of richer logical systems, such as intensional logic and the theory of types,
and it demonstrates the application of these in a logical grammar. Logic is introduced from a
linguistic perspective in volume 1, although an attempt has been made to keep things
interesting for readers who just want to learn logic (perhaps with the exception of those with a
purely mathematical interest in the subject). Thus some subjects have been included which are
not to be found in other introductory texts, such as many-valued logic, secondorder logic, and
the relation between logic and mathematical linguistics. Also, a first attempt is made at a logical
pragmaticS. Other and more traditional subjects, like the theory of definite descriptions and the
role of research into the foundations of mathematics, have also been dealt with. Volume 2
assumes a familiarity with propositional and predicate logic, but not necessarily a familiarity with
volume 1. The first half of it is about different systems of intensional logic and the theory of
types. The interaction between the origins of these systems in logic and philosophy and the part
they have to play in the development of intensional theories of meaning is a com- xii Preface
mon thematic thread running through these chapters. In the course of the exposition, the careful
reader will gradually obtain a familiarity with logic and philosophy which is adequate for a proper
understanding of logical grammar. Montague grammar, the best-known form of logical grammar,
is described in detail and put to work on a fragment of the English language. Following this,
attention is paid to some more recent developments in logical grammar, such as the theory of
generalized quantification and discourse representation theory. One important objective of this
book is to introduce readers to the tremendous diversity to be found in the field of formal logic.
They will become acquainted with many different logics-that is, combinations of formal
languages, semantic interpretations, and notions of logical consequence-each with its own field
of application. It is often the case in science that one is only able to see which of one's theories
will explain what, and how they might be modified or replaced when one gets down and
examines the phenomena up close. In this field too, it is the precise, formal analysis of patterns
and theories of reasoning which leads to the development of alternatives. Here formal precision
and creativity go hand in hand. lt is the authors' hope that the reader will develop an active
understanding of the matters presented, will come to see formal methods as llexiblc methods for
answering semantic questions, and will eventually be in a position to apply them as such. To this
end, many exercises have been included. These should help to make the two volumes suitable
as texts for courses, the breadth and depth of which could be quite diverse. Solutions to the
exercises have also been included, in order to facilitate individual study. A number of exercises
are slightly more difficult and are marked by 0 . These exercises do not have to be mastered
before proceeding with the text. In order to underline their common vision, the authors of these
two volumes have merged their identities into that of L. T. F. Gamut. Gamut works (or at least
did work at the time of writing) at" three difjerent universities in the Netherlands: Johan van
Benthem as a logician at the University of Groningen; Jeroen Groenendijk as a philosopher,
Dick de Jongh as a logician, and Martin Stokhof as a philosopher at the University of
Amsterdam; and Henk Verkuyl as a linguist at the University of Utrecht. This work did not
appear out of the blue. Parts of it had been in circulation as lecture notes for students. The
exercises, in particular, derive from a pool built up through the years by the authors and their
colleagues. The authors wish to express their thanks to all who have contributed in any way to
this book. Special thanks are due to Piet Rodenburg, who helped write it in the early stages, to
Michael Morreau for his translation of volume 1 and parts of volume 2, and to Babette Greiner
for her translation of most of volume 2. Summary of Volume 1 In chapter I, logic is introduced as
the theory of reasoning. Some systematic remarks are made concerning the connection
between logic and meaning, and J I f.:JUCC. a short historical survey is given of the relationship
between logic, philosophy, and linguistics. Furthermore, the role of formal languages and how
they are put to use is discussed. Chapter 2 treats propositional logic, stressing its semantic side.
After the exposition of the usual truth table method, the interpretation of connectives as truth
functions is given. In connection with this and also for later use, the concept of a function is
introduced. Chapter 2 concludes with a section in which the syntax of propositional languages is
developed in a way more akin to the syntax of natural language. The purpose of this
section-which different argument schema is obtained if the conjunction or in ( 11) is replaced
by another conjunction, say, if: (12) AifB NotA B This schema is not valid. One of the
substitutions for A and B is, for example, (l 0), and that is not a valid argument. That
expressions other than the conjunctions can lead to arguments being valid becomes apparent if
we examine example (5) in more depth. Considerations similar to those for ( 1) lead to the
following argument schema for (5): (13) All P are Q a is P a is Q In this schema the letters P and
Q stand for expressions which refer to properties, and a stands for an expression which rerers
to an individual or an entity, that is, to a material or an abstract object. It will be clear that every
substitution for a, P, and Q results in a valid argument; (5) is one example of these. The validity
of this schema derives from, among other things, the meaning of the quantifying expression all.
Other examples of quantifying expressions to be found in argument schemata are some and no.
Logic, as the science of reasoning, investigates the validity of arguments by investigating the
validity of argument schemata. For argument schemata are abstractions which remove all those
elements of concrete arguments which have no bearing on their validity. As we have seen,
argument schemata can be formed from a variety of expressions and syntac~ic constructions.
Usually they are not all considered together but are taken in groups. So, for example, we can
concentrate on those argument schemata which can be formed solely from sentences,
grammatical conjunctions, like or and if . .. then. and negation. Or we can single out arguments
containing quantifying expressions. But before going into this any further, let us briefly consider
the relationship between logic and meaning. 1.2 Logic and Meaning As we have pointed out, the
meanings of certain kinds of expressions play an essential part in determining the validity of
schemata in which they appear. So to the extent that logic is concerned with the validity of
argument schemata, it is also concerned with the meanings of expressions. Take the
conjunction or. for example; its meaning is partly responsible for the validity of argument
schema (ll). In investigating the validity of schemata in which this conjunction plays a part, then,
we are also investigating its meaning. And if we were Introduction 5 to succeed in determining
exactly which of these schemata are valid and which are not, which is the business of logic, we
would to some extent have succeeded in determining what or means. And the same applies, of
course, to all of the other expressions which can play a part in the validity of argument
schemata, like the other conjunctions, negation, and quantifying expressions. But would we,
having characterized all the valid argument schemata in which a given expression occurs, have
succeeded in determining all of that expression's meaning? That is a matter which we shall
return to in chapter 2 and chapter 6. In the meantime we will just say that at least a large and
important part of the meaning of an expression can be determined in this manner. Knowing what
the word and means obviously involves knowing that the conclusion A (and the conclusion B)
may be drawn from the expression A and B. An investigation into the validity of arguments
involves an investigation into one particular relation between the meanings of sentences, the
relation of logical consequence, and thus at the same time into the meanings of particular
expressions. Earlier we said that valid arguments are those with conclusions which are logical
consequences of their premises. So a characterization of valid arguments is a characterization
of what sentences follow from what others. The relation of logical consequence, which as we
shall see can be defined in terms of the even simpler semantic notion of truth, can therefore in
turn be used to characterize other relationships between the meanings of sentences and other
kinds of expressions. It is the connection between logic and meaning that makes logic
interesting from a linguistic standpoint. And the contribution which logic can make to linguistics
is not limited to giving precise descriptions of the meanings of the grammatical conjunctions,
negation, quantifying expressions, and so on. It should be noted that logic gives semantic
interpretations of syntactic operations. By that we mean the following. When we investigate
what arguments are valid on the basis of the meanings of the grammatical conjunctions and
negation, we are not interested in the actual meanings of the sentences conjoined by those
conjunctions. We do not consider actual arguments like ( 1) and (10), but argument schemata
like (11) and (12). But when we consider them we still say at least something about the
meanings of sentences, in that we must at some stage say what kinds of entities the meanings
of sentences are and how the meanings of composite sentences depend on the meanings of
their composite parts. The nature of the concept 'meaning of a sentence' must, in other words,
be made more precise, and a semantic interpretation must be given to the syntactic operations
by which some sentences may be obtained from others. So we do not go into the actual
meanings of particular predicating expressions, but we do determine the nature of their
meanings and give a semantic interpretation to the syntactic rules by means of which sentences
may be obtained from predicating and quantifying expressions. Logic thus gives a precise
content to the principle which states that the meaning of a composite expression must be built
up from the meanings of its comp here was a considerable sensitivity to the great diversity of
language, and explanations were sought for each different kind of expression. The wellknown
supposition theory can be seen as an attempt at a semantic analysis of the terms and their
combinations found in language. As such, the supposition theory declined together with the rest
of Scholasticism. Some distinctions drawn at the time, however, are still with us today. So, for
example, the distinction between suppositio forma/is and suppositio materialis is now known as
the use/mention distinction. This distinction is apparent in the difference between sentences like
(28) and (29): (28) Amsterdam is the capital city of the Netherlands. (29) Amsterdam has nine
letters. The Scholastics said that the term Amsterdam in (28) has suppositioformalis, that is, that
it is used in order to refer to that Dutch city. But in (29) the term has suppositio materialis: it
refers to the word Amsterdam; the term is mentioned. In this book we make a typographic
distinction between use and mention, writing (29) as (30): (30) Amsterdam has nine letters. The
theory of the distribution of terms, which until recently was memorized by many generations of
students, is another relic of the Middle Ages. In the universal affirmative sentence All A are B.
the term A is 'distributed': the sentence says something about the whole concept A. The term B.
on the other hand, is not distributed: the sentence does not necessarily say something about all
B 's, but only about the A's among them. It should be noted that supposition theory was also
supposed to deal with problems arising from sentences with'more than one quantifier. As we
have mentioned, such sentences were not included in Aristotle's theory of syllogisms. But as
early as the thirteenth century, William of Shyreswood considered the validity of inferences like
(31 ): (31) There is someone who is seen by (suppositio determinata) everyone Everyone sees
someone. (suppositio confusa tantum) Note that the inverse inference is not valid. lt is surprising
how Aristotle and some medieval philosophers were happy to make use of the (invalid) inverse
of (31) when this suited their metaphysical purposes. So, for example, the conclusion that there
is one cause which is the cause of all events was drawn from the premise that every event has
a cause. The Scholastics did not succeed in giving a satisfactory account of sentences with
more than one quantifying expression. ln fact, it wasn't until 1879 and the publication of Frege's
Begriffsschrift that the problems of multiple quantification were definitively solved. Introduction
13 In the Middle Ages, linguistics was primarily searching for a rational basis for the rules of
grammar. It was not enough for such rules to 'work' in the analysis of literary texts; what was
important was how they are related to, or how they reflect, the nature of thought. The
philosophical grammarians who considered language from this viewpoint are known as the
Modists. Besides descriptive grammars for practical purposes, speculative grammars (speculum
'mirror') were developed. The ideal of a universal grammar gained popularity. After all, if human
thought is the same everywhere, then the ideal grammar must be too. The different languages
are, according to this view of grammar, all variations on and approximations to this one ideal
theme. That logic was thought indispensable for the grammarian is apparent from the following
quote from Albertus Magnus (thirteenth century): "A (logically) unschooled grammarian is to a
grammarian schooled in logic as an idiot to a wise man." And logic increasingly began to be
occupied with the linguistic aspects of reasoning, as is apparent from William of Shyreswood's
opinion that grammar teaches us to speak correctly, rhetoric teaches us to speak elegantly, and
logic teaches us to speak the truth. For the philosophical grammarians, logic was not so much a
scientia rationalis, a science of concepts, as a scientia sermocinalis, a science of discourse
concerned with terms. One of the fruits of this interest in terms and their semantics is the
distinction between categorematic terms like man or sick, which refer to something, and
syncategorematic terms like every or no, which are supposed not to have their own reference,
but which are from a logical point of view essential to the meanings of sentences and the logical
relations obtaining between them. But as the Middle Ages progressed, the development of logic
gradually seemed to grind to a halt. In 1789 Immanuel Kant wrote in the preface to the second
edition of his Critique of Pure Reason that logic hadn't lost any ground since Aristotle, but that
the subject hadn't gained any ground either, and that there was every indication that it would go
no farther. But Kant was mistaken. As much as one hundred years before, working at the same
time and in the same isolation as the Port-Royal school, the mathematician and philosopher p is
said to be the first member (or conjunct) of the conjunction (p 1\ q), and q is the second. The
same applies to implications, disjunctions, and equivalences, though the first and second
members of an implication are sometimes referred to as its antecedent and its consequent,
respectively, while the two members of a disjunction are referred to as its disjuncts. Our choice
of connectives is in a certain sense arbitrary. Some are obviously important. We shall discuss all
five separately and consider the extent to which the corresponding expressions from natural
language can be regarded as truth-functional. We shall also discuss a few other possible
connectives which have not been included in this list. The syntactic rules of propositional
languages allow us to link up, by means of connectives, not only propositional letters (which are
also referred to as atomic formulas), but also composite formulas. The terminology is just the
same: if cf> and 1/1 are formulas, then •cf> is said to be the negation of cf>, (cf> 1\ 1/J) is the
conjunction of cf> and 1/J, etc.; •cf> refers naturally enough to the string of symbols obtained by
prefixing the string cf> with a •; (cf> 1\ 1/J) refers to the string of symbols consisting or a left
bracket, followed by the string ¢, followed b; the connective /\, followed by the string 1/J, and
closing with a right bracket. The brackets serve to remove any ambiguities. Otherwise a
sentence like p v q 1\ r could have either of two different meanings. It could be (a) the
disjunction of p, on the one hand, and the conjunction of q and ron the other; or (b) the
conjunction of the disjunction of p and q, on the one hand, and ron the other. That these have
distinct meanings can easily be seen from examples like (5) and ( 6): (5) McX has been elected,
or Wyman has been elected and a new era has begun. ~ (6) McX has been elected or Wyman
has been elected, and a new era has begun. Example (5) corresponds to (p v (q 1\ r)) and as a
whole is a disjunction, while (6) corresponds to ((p v q) 1\ r) and as a whole is a conjunction. We
shall return to these more complex formulas in §2.3. Now we shall expand upon the different
meanings of the various connectives. What concerns us is the way in which the truth value of a
composite sentence formed from one or two simpler sentences depends on the truth values of
the constituent sentences and the connective used. For each connective, this is prescribed in a
truth table. The discussion of sentence (!) shows that the truth table for the conjunction is as in
(7): (7) cf> 1/1 ( cf> 1\ 1/1) ~-+--~~~~--~ I 0 () I 1 0 0 0 () () Propositional Logic 31 Beside each
possible combination of the truth values of cf> and 1/J we see the resulting truth value of (cf> 1\
1/J). On the face of it, it might seem that the logical behavior of 1\ is wholly in accordance with
that of and in natural language. The agreement, however, is not perfect. If someone truthfully
states (8), then according to the truth table sentence (9) is also true. (8) Annie took off her socks
and climbed into bed. (9) Annie climbed into bed and took off her socks. But it is very likely that
the person in question would not be inclined to accept this, since placing one sentence after the
other suggests that this was also the order in which the described events happened. Similar
complications arise with all the other connectives. In chapter 6 we shall discuss whether this
kind of phenomenon can be explained in terms of conditions for usage. The left column of (I 0)
is a list of sentences which have the same truth conditions as the conjunction of the two
sentences to their right. (10) Zandvoort and Haarlem Zandvoort lies west of Haarlem lies west of
lie west of Amsterdam. Amsterdam. Amsterdam. John and Peter are mar- John is married to
Anne. Peter is married to Betty. ried to Anne. and Betty. respectively. Both the Liberals and the
The Liberals favored the The Socialists favored the Socialists favored the motion. motion.
motion. John is at home but he is John is at home. John is asleep. asleep. John is at home but
Peter John is at home. Peter is not at home. is not. Although it was ex- It was extremely cold.
John did not stay indoors. tremely cold, John did not stay indoors. Even though it was beau- It
was beautiful out of John stayed indoors. tiful out of doors, John doors. stayed indoors. So the
sentences in the left column all express a logical conjunction, although from a strictly linguistic
point of view we are not dealing with two sentences linked by placing an and between them.
Apparently the connotations which but, although, and though have do not alter the truth
conditions of the sentences in which the words occur. Note also that not every sentence in
which the word and figures is a conjunction. Here is an example that is not: (11) John and Peter
are friends. lt seems rather unnatural to regard this as a conjunction, say, of John is friends with
Peter and Peter is friends with John. And sentence (12) does not mean the same as the
conjunction (13): 32 Chapter Two (12) Cheech and Chong are fun at parties. (13) Cheech is fun
at parties and Chong is fun at parties. Perhaps they are only fun when they're together.
Negation is also a relatively simple matter. The truth table consists of just two rows; see (14):
(14) m·cp I 0 0 I There are more ways to express the negation of a sentence than by means of
not or it is not the case that. See, for example. (15): (15) Porcupines are unfriendly. John is
neither at home nor at school. No one is at home. John is never at home. John is not home yet.
John has never yet been at home. Porcupines are friendly. John is either at home or at school.
Someone is at home. John is sometimes at home. John is home already. John has on occasion
been at home. For the disjunction we give the truth table in (16): (16) _,_cfJ-+_tfi.:__-l----cp~v
.....:...tfi I 0 0 I 0 I 0 I I 0 This is the obvious truth table for and/or, and in natural language or
generally means and! or. This usage is said to be inclusive. If or is used in such a way as to
exclude the possibility that both disjuncts are true-this is also expressed by an either ... or ...
construction-then the or is said to be exclusive. Sometimes a separate connective oo is
introduced for the exclusive disjunction, the truth table for which is given in figure (17): 0 0 I 0 0 I
0 I 0 ln this book, or is understood to be inclusive unless stated otherwise, as is also usual in
mathematics. Propositional Logic 33 Actually it is not very easy to find a natural example of an
exclusive or. A sentence like ( 18) will not do: (18) It is raining or it isn't raining. ln sentence (18)
there would be no difference in the truth value whether the or were inclusive or not, since it
cannot both rain and not rain. What we need is an example of the form A orB in which there is a
real possibility that both A and B hold; this eventuality is excluded by the exclusive disjunction.
In natural language, this is usually expressed by placing extra emphasis on the or, or by means
of either ... or. For example: (19) Either we are going to see a film tonight, or we are going to the
beach this afternoon. Another construction which can be used to express an exclusive
disjunction is