See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/238683285

Lexical Pragmatics

Article in Journal of Semantics · May 1998

DOI: 10.1093/jos/15.2.115

CITATIONS READS

149 3,559

1 author:

Reinhard Blutner
University of Amsterdam
84 PUBLICATIONS 1,419 CITATIONS

SEE PROFILE

All content following this page was uploaded by Reinhard Blutner on 03 July 2014.

The user has requested enhancement of the downloaded file.

 Two case studies in lexical pragmatics 1

TWO CASE STUDIES IN LEXICAL

PRAGMATICS

Reinhard Blutner, Humboldt University, Berlin, Germany

Torgrim Solstad, University of Oslo, Oslo, Norway

1 INTRODUCTION

Lexical Pragmatics is a particular account of the division of labor between lexical

semantics and pragmatics (e.g. Blutner 1998). It combines the idea of (radical)
semantic underspecification in the lexicon with a theory of pragmatic strengthening
(based on conversational implicatures). In the core of this approach is a precise
treatment of Atlas & Levinson's (1981) Q- and I-principles and the formalization of
the balance between informativeness and efficiency in natural language processing
(Horn's 1984 division of pragmatic labor). In a roughly simplified formulation, the I-
principle seeks to select the most coherent interpretation, and the Q-principle acts as
a blocking mechanism which blocks all the outputs which can be grasped more
economically by an alternative linguistic input. Recently, these mechanisms have
been implemented within a bidirectional version of optimality theory (OT) which
aims to integrate expressive and interpretive optimization (Blutner 1999).
The aim of this paper is to apply this framework to two different kinds of
examples. First, we aim to provide a concise treatment of the phenomenon of
negative strengthening as arising in connection with gradable adjectives. Second, we
want to resolve some puzzles of dimensional designation of spatial objects. In the
first case the optimality theoretic treatment can be seen as a reformulation and
revivification of earlier approaches by Horn (1989) and Levinson (2000). In the
second case the treatment is really new and crucially deviates from earlier ways of
solution.
The paper is organized as follows. Section 2 introduces the bidirectional optimality
framework and illustrates how the Gricean framework of conversational implicature
can be reformulated by means of this technique. In Section 3 we give a concise
2

introduction in the phenomenon of negative strengthening, and in the subsequent

section we account for the basic phenomena by using the bidirectional optimality
framework. In section 5 some puzzles of dimensional designation of spatial objects
are outlined. Finally, in section 6, the bidirectional optimality framework is applied to
solve these puzzles, and in section 7 some preliminary conclusions are drawn.

2 BIDIRECTIONAL OT AND PREFERENCES FOR INTERPRETATION

Recently, de Hoop & de Swart (1998), Hendriks & de Hoop (to appear), and de Hoop
(2000) have applied OT to sentence interpretation. They argue that there is a
fundamental difference between the form of OT as used in phonology, morphology
and syntax on the one hand and its form as used in semantics on the other hand.
Whereas in the former case OT takes the point of view of the speaker (expressive
perspective), in the latter case the point of view of the hearer is taken (interpretive
perspective).
One obvious reason for this difference is that ambiguity, polysemy, and other
forms of flexibility are much more obvious and manifested much broader in the area
of interpretation than in the realm of syntax. The assumption that OT in sentence
interpretation takes the point of view of the hearer is mainly motivated by this
observation. Using the interpretive perspective, a mechanism for preferred
interpretations is constituted that provides insights into different phenomena of
interpretations, such as the determination of quantificational structure (Hendriks & de
Hoop, to appear), nominal and temporal anaphorization (de Hoop & de Swart 1998),
the interpretational effects of scrambling (de Hoop 2000), the projection mechanism
of presupposition (Zeevat 1999 a,b; Blutner 1999; Geurts, to appear).
However, Blutner (1999) argues that this design of OT is inappropriate and too
weak in a number of cases. This is due to the fact that the abstract generative
mechanism (called Gen in the OT literature) can pair different forms with one and the
same interpretation. The existence of such alternative forms may raise blocking
effects which strongly affect what is selected as the preferred interpretation. The
phenomenon of blocking requires us to take into consideration what the speaker could
have said. As a consequence, we have to go from a one-dimensional, to a two-
dimensional (bidirectional) search for optimality.
This bidirectional view was independently motivated by a reduction of Grice's
maxims of conversation to two principles: the Q-principle and the I-principle (Atlas &
Levinson 1981; Horn 1984, who writes R instead of I). The I/R-principle can be seen
as the force of unification minimizing the Speaker's effort, and the Q-principle can be
seen as the force of diversification minimizing the Auditor‘s effort. The Q-principle
corresponds to the first part of Grice's quantity maxim (make your contribution as
informative as required), while it can be argued that the countervailing I/R-principle
collects the second part of the quantity maxim (do not make your contribution more
informative than is required), the maxim of relation and possibly all the manner
maxims.
 Two case studies in lexical pragmatics 3

In a slightly different formulation, the I/R-principle seeks to select the most

coherent interpretation, and the Q-principle acts as a blocking mechanism which
blocks all the outputs which can be grasped more economically by an alternative
linguistic input (Blutner 1998). This formulation makes it quite clear that the Gricean
framework can be conceived of as a bidirectional optimality framework which
integrates expressive and interpretive optimality. Whereas the I/R-principle compares
different possible interpretations for the same syntactic expression, the Q-principle
compares different possible syntactic expressions that the speaker could have used to
communicate the same meaning. The important feature of this formulation within
bidirectional OT is that although it compares alternative syntactic inputs to one
another, it still helps to select the optimal meaning among the various possible outputs
of the single actual syntactic input given, by acting as a blocking mechanism.
A strong version of bidirectional OT can be formulated as given in (1). We here
relate pairs (f, m) of possible (syntactic) forms f and meanings (= semantic
interpretations) m, by means of an ordering relation >, being more efficient.

(1) Bidirectional OT (Strong Version)

A form-meaning pair (f, m) is optimal iff it is realized by Gen and it satisfies

both the Q- and the I-principle, where:

(Q) (f, m) satisfies the Q-principle iff there is no other pair (f', m) realized
by Gen such that (f', m) > (f, m)

(I) (f, m) satisfies the I-principle iff there is no other pair (f, m' ) realized
by Gen such that (f, m') > (f, m)

We will now give a very schematic example in order to illustrate some characteristics
of the bidirectional OT. Assume that we have two forms f1 and f2 which are
semantically equivalent. This means that Gen associates the same meanings with
them, say m1 and m2. We stipulate that the form f1 is less complex (marked) then the
form f1 and that the interpretation m1 is less complex (marked) then the interpretation
m1 . From these differences of markedness with regard to the levels of syntactic
representation / semantic interpretations, the following ordering relation between
representation-meaning pairs can be derived:

(2) a. (f1 , m1) > (f2, m1)

b. (f1 , m2) > (f2, m2)
c. (f1 , m1) > (f1, m2)
d. (f2 , m1) > (f2, m2)

Using Dekker’s & van Rooy’s (1999) notation, the following bidirectional OT
diagram can be construed, nicely representing the preferences between the pairs. More
importantly, such diagrams give an intuitive visualization for the optimal pairs of
4

(strong) bidirectional OT: they are simply the hallows if we follow the arcs. (It should
be noted that Dekker & van Rooy (1999) give bidirectional OT a game theoretic
interpretation where the optimal pairs can be characterized as so-called Nash
Equilibria). The optimal pairs are marked with the symbol  in the diagram.

(3)

f1 

f2  
m1 m2

The scenario just installed describes the case of total blocking where some forms
(e.g., *furiosity, *fallacity) do not exist because others do (fury, fallacy). However,
blocking is not always total but may be partial. This means that not all the
interpretations of a form must be blocked if another form exist. According to
Kiparsky (1982) partial blocking is realized in the case where the special (less
productive) affix occurs in some restricted meaning and the general (more productive)
affix picks up the remaining meaning (consider examples like refrigerant -
refrigerator, informant - informer, contestant - contester). McCawley (1978) collects
a number of further examples demonstrating the phenomenon of partial blocking
outside the domain of derivational and inflectional processes. For example, he
observes that the distribution of productive causatives (in English, Japanese, German,
and other languages) is restricted by the existence of a corresponding lexical
causative. Whereas lexical causatives (e.g. (4a)) tend to be restricted in their
distribution to the stereotypical causative situation (direct, unmediated causation
through physical action), productive (periphrastic) causatives tend to pick up more
marked situations of mediated, indirect causation. For example, (4b) could have been
used appropriately when Black Bart caused the sheriff's gun to backfire by stuffing it
with cotton.

(4) a. Black Bart killed the sheriff

b. Black Bart caused the sheriff to die

Typical cases of total and partial blocking are not only found in morphology, but in
syntax and semantics as well (cf. Atlas & Levinson 1981, Horn 1984, Williams 1997).
The general tendency of partial blocking seems to be that "unmarked forms tend to be
used for unmarked situations and marked forms for marked situations" (Horn 1984:
26) – a tendency that Horn (1984: 22) terms "the division of pragmatic labor".
There are two principal possibilities to avoid total blocking within the bidirectional
OT framework. The first possibility is to make some stipulations concerning Gen in
order to exclude equivalent semantic forms. The second is to weaken the notion of
 Two case studies in lexical pragmatics 5

(strong) optimality in a way that allows us to derive Horn’s division of pragmatic

labor by means of the evaluation procedure.
Blutner (1998,1999) argues that the second option is much more practicable and
theoretically interesting. A weak version of two-dimensional OT was proposed,
according to which the two dimensions of optimization are mutually related:

(5) Bidirectional OT (Weak Version)

A form-meaning pair (f, m) is super-optimal iff it is realized by Gen and it

satisfies both the Q- and the I-principle, where:

(Q) (f, m) satisfies the Q-principle iff there is no other pair (f', m) realized
by Gen which satisfies the I-principle such that (f', m) > (f, m)

(I) (f, m) satisfies the I-principle iff there is no other pair (f, m' ) realized
by Gen which satisfies the Q-principle such that (f, m') > (f, m)

A more transparent formulation of super-optimality has been proposed by Jäger

(2000):

(6) Bidirectional OT (Weak Version, Jäger’s variant)

A form-meaning pair (f, m) is super-optimal iff it is realized by Gen and it

satisfies the following two conditions:

(Q) there is no other super-optimal pair (f', m) : (f', m) > (f, m)

(I) there is no other super-optimal pair (f, m' ) : (f, m') > (f, m)

Under the assumption that > is transitive and well-founded, Jäger (2000) observes that
both versions of weak bidirection coincide; that is a representation-meaning pair is
super-optimal in the sense of definition (5) if and only if it is super-optimal in the
sense of definition (6).
The important difference between the weak and strong notions of optimality is that
the weak one accepts super-optimal form-meaning pairs that would not be optimal
according to the strong version. It typically allows marked expressions to have an
optimal interpretation, although both the expression and the situations they describe
have a more efficient counterpart. Consider again the situation illustrated in (3), but
now apply the weak versions of bidirectional optimization (to make things more
concrete we can take f1 to be the lexical causative form (4a), f2 the periphrastic form
(4b), m1 direct (stereotypic) causation and m2 indirect causation).
6

(7)
f1 

f2  ()
m1 m2

We have seen that the strong version cannot explain why the marked form f2 has an
interpretation as well. The weak version, however, can explain this fact. Moreover, it
explains that the marked form f2 gets the atypical interpretation m2. The form f2 gets
the interpretation m2, because this form-meaning pair is super-optimal: (i) the
alternative form f1 doesn't get the atypical interpretation m2, and (ii) we prefer to refer
to the typical situation m1 by using f1 instead of f2. In this way, the weak version
accounts for the pattern called "the division of pragmatic labor". It is not difficult to
see that this pattern can be generalized to systems where more than two forms are
associated by Gen with more than two interpretations. In the general case, we start
with determining the optimal pairs. Then we drop the rows and columns
corresponding to the optimal pair(s) and apply the same procedure for the reduced
tableau.
The additional solutions are due to the flexibility and ability to learn which the
weak formulation alluded to. The strong view is sufficient when it is enough to find
one prominent solution. The weak view allows us to find out other solutions as well.
In section 4 we will make use of this more general solution concept to explain the
effects of negative strengthening, and in section 6 we will use it in order to explain the
patterns of dimensional designation for spatial objects.

3 THE PHENOMENON OF NEGATIVE STRENGTHENING

Negation in natural language is a rich source of a variety of non-logical inferences

(see Horn 1989). Standard examples are scalar implicatures (Not all of the students
came > Some of them came). Others are collected under the term negative
strengthening. For an excellent discussion of the phenomenon of negative
strengthening we refer to Horn (1989) and Levinson (2000).
One instance of the phenomenon of negative strengthening arises in connection
with gradable adjectives which typically occur as antonyms, such as {good, bad},
{large, small}, {happy, unhappy}. Semantically, the elements of antonym pairs are
contraries, that is they are mutually inconsistent but do not exhaust the whole
spectrum, permitting a non-empty middle ground.
What are the effects of negating gradable adjectives? For the sake of explicitness
let’s consider the gradable antonyms happy and unhappy, and assume three possible
states of happiness – iconized by -/, , and .. Not unexpectedly, we want to
 Two case studies in lexical pragmatics 7

take happy as referring to the first state, unhappy as referring to the second state, and
neither happy nor unhappy as referring to the third state.
Let’s consider first the effect of negating positive adjectives, starting with a
sentence like (8a). Obviously, the preferred interpretation of this sentence is (8c); this
corresponds to a logical strengthening of the content of (8a) which is paraphrased in
(8b). The discourse (8d) shows that the effect of strengthening (8c) is defeasible.
This indicates that the inferential notion that underlies the phenomenon of
strengthening ought to be non-monotonic.

(8) a. I’m not happy /

b. It isn’t the case that I’m happy (Entailment) /.
c. I'm unhappy (Implicature) /
d. I'm not happy and not unhappy (Defeasibility)

Following Levinson (2000), the effect of negative strengthening for positive

adjectives can be illustrated in the following way:

HAPPY | INDIFFERENT | UNHAPPY

coded range of happy

coded range of not happy

implicated range

Figure 1: Negative strengthening as implicated contraries

It describes the effect of negative strengthening as implicating contraries from

contradictions.
The illustrated shape of negative strengthening is restricted to the positive
(unmarked) element of an antonym pair. When considering negative adjectives,
deviations from this pattern may be found. The deviations are rather obvious for
adjectives with incorporated affixal negation. This leads us to the well-known case of
double negation (Litotes):

(9) a. I’m not unhappy /

8

b. It isn’t the case that I’m unhappy (Entailment) -.

c. I’m neither happy nor unhappy (Implicature) .
d. I'm rather happy (but not quite as happy as (proper Implicature)
using the expression “happy” would suggest)

e. I'm not unhappy, in fact I’m happy (Defeasibility)

Admitting only three states on the happiness scale allows only a rather rough
approximation of the interpretational effects. The simplest approximation describes
negative strengthening as a preference for the middle ground. This is what (9c)
expresses. A more appropriate formulation of the effect is given in (9d). For the sake
of precision, we had to introduce intermediate states between - and .(on the
scale of happiness). In the following diagram a fairly adequate illustration of the
basic pattern is presented (as described in Horn 1989, Levinson 2000.)

UNHAPPY | INDIFFERENT | HAPPY

coded range of unhappy

coded range of not unhappy

implicated range

Figure 2: Litotes: when two negatives don't make a positive

As in the case discussed before the effect of negative strengthening proves defeasible,
a fact that requires the underlying inferential notion to be non-monotonic.

4 NEGATIVE STRENGTHENING AND BIDIRECTIONAL OT

In the previous section a concise description of the phenomenon of negative

strengthening was given. Now we will bring this phenomenon into play in order to
illustrate the general mechanism of pragmatic strengthening, which is formulated by
using the method of bidirectional optimization.
In the analysis of Horn (1989) and Levinson (2000) there are some types of
negative strengthening that are obviously attributable to the I/R-principle. A clear
case is the negation of positive adjectives, which was described in connection with
 Two case studies in lexical pragmatics 9

example (8). Here the I/R principle leads to a pragmatic strengthening effect
excluding the middle ground and inferring the contrary.
The situation is not so clear in the case of adjectives with incorporated affixal
negation such as in example (9). Whereas Horn (1984, 1989) seems to attribute the
observed effect of negative strengthening to the interaction between Q and R,
Levinson stipulates a third pragmatic principle, the M(anner)-principle: “what’s said
in a abnormal way, isn’t normal; or marked message indicates marked situation.”
(Levinson 2000: 33). Obviously, this principle expresses the second half of Horn’s
division of pragmatic labor. In our opinion, Levinson (2000) tries to turn a plausible
heuristic classification scheme based on the three principles Q, I, and M into a
general theory by stipulating a ranking Q > M > I. Accepting the heuristic
classification schema, we see problems for this theory, which is burdened with too
many stipulations. Not unlike Horn’s conception, we would like to see the M-
principle as an epiphenomenon that results from the interaction of Zipf’s two
“economy principles” (Q and R in Horn’s terminology).
Let’s see now how bidirectional OT accounts for the effects of negative
strengthening. The bidirectional tableau (10) shows the competing candidate forms in
the left column. (Take the candidate entries as shortcuts for complete sentences; for
example take happy as abbreviating I’m happy, etc.). The other columns are for the
three possible states of happiness considered in this simplified analysis. The gray
areas in the tableau indicate which form-interpretation pairs are excluded by the
compositional mode of truth-functional semantics, which is described by Gen. For
example, I’m not unhappy is assumed to exclude the state iconized by /.

(10)
happy  
not unhappy ()a a
F

not happy  h f

unhappy 

- . /
The preferences between the form-interpretation pairs are due to markedness
constraints for forms and markedness constraints for interpretations, respectively.
With regard to the forms, we simply assume that the number of negation
morphemes is the crucial indicator. The corresponding preferences are indicated by
the vertical arrows. (Note that not happy and unhappy aren’t discriminated in terms
of markedness – a rough simplification, of course.)
With regard to the states, we assume that they are decreasing in markedness
towards both ends of the scale, assigning maximal markedness to the middle ground.
Although this assumption seems not implausible from a psycholinguistic perspective,
10

we cannot provide independent evidence for it at the moment. In the tableau (10), the
corresponding preferences are indicated by the horizontal arrows.
Now it is a simple exercise to find out the optimal solutions – indicated by . One
optimal solution pairs the sentence I’m not happy with the interpretation /. This
solution corresponds to the effect of negative strengthening that is attributable to the
I/R-principle. The other two optimal solutions are reflecting the truth condition of I’m
happy/unhappy.
Most interesting, there is an additional super-optimal solution, indicated by (). It
pairs the sentence I’m not unhappy with the interpretation .. This corresponds to the
effect of negative strengthening in the case of Litotes, normally attributed to
Levinson’s (2000) M-principle or Horn’s division of pragmatic labor. As already
stressed, this solution comes out as a natural consequence of the weak form of
bidirection, which can be seen as a formal way of describing the interactions between
Q and I/R.
It’s an interesting exercise to introduce more than three states of happiness and to
verify that the proper shape of implicature as indicated in Figure 2 can be
approximated. More importantly, in the context of litotes it seems necessary to
account for the effect of gradient acceptability and continuous scales. Using a
stochastic evaluation procedure, Boersma (1998) did pioneering work in this field,
which should be exploited in the present case.
The other prominent class of examples that exhibit the effect of negative
strengthening concerns the phenomenon of neg-raising, i.e. the tendency for negative
main sentences with subordinate clauses to be read as negations of the subordinate
clause (cf. Horn 1998, Levinson 2000). It seems fruitful to analyze the phenomenon
using the same technique as described before.

5 SOME PUZZLES OF DIMENSIONAL DESIGNATION

The term ‘dimensional designation’ refers to the contextual interpretation of a group

of spatial adjectives such as long, high, broad, deep, thick and can be illustrated by
the following example:

(11) a. The windowsill is 1 m long, 30 cm wide and 3 cm thick

b. The windowsill is 1 m wide, 30 cm deep and 3 cm thick

In (11a) the adjective wide refers to the secondary dimension whereas in (11b) it
refers to the maximal (most salient) dimension. In order to explain the basic effects of
dimensional designation we need the right combination of lexical stipulations and
general principles of coherence, blocking and (perhaps) deblocking. In the following,
we want to illustrate how bidirectional OT solves this conceptual and methodological
problem.
There is a thorough literature describing the linguistic facts of dimensional
designation of spatial expressions (e.g. Bierwisch 1967, Lang 1989). It is not the aim
of this section to extend this literature or to find out new observations that challenge
 Two case studies in lexical pragmatics 11

the basic facts described there. As usual, the facts are described by using some
(semi-) formal representational system. Certainly, there are good reasons for further
improving the existing systems. However, that isn’t our job in the present article.
Let’s concentrate on a small sector of the known observations by using standard
representational systems. What is more important, we feel, is to provide a real
explanation of these facts and observation. Our aim is to demonstrate that the
framework of bidirectional optimization can be an appropriate tool for obtaining
explanatory adequacy. Hopefully, this tool will help us to get a real understanding of
the basic facts. A related point is a methodological one. It aims at the right
relationship between lexical stipulations and general principles of economy. This is
of great importance also for practical systems that should use lexical stipulations
sparingly.
Suppose a physical object that our brain tries to encode. Suppose further that we
can discriminate different dimensions (or axes) of spatial extent. It is the typical
function of a spatial adjective to refer to a particular dimension of that object (in a
particular contextual setting). The theoretical problem concerning the dimensional
designation of spatial objects is to provide a mechanism that allows a realization of
the mapping between dimensional adjectives on the one hand and the dimensions of
physical objects referred to on the other hand. For simplicity, we will concentrate on
two- and three-dimensional spatial objects where all axes are disintegrated (i.e. we
don’t considering objects like tree, ball, and wheel where two or more axes are
integrated into one dimension. Furthermore, we are considering only a very restricted
list of adjectives, namely the following: long, high, wide, deep, thick.
The facts we are considering aim at two different but interrelated phenomena:
interpretational preferences and blocking. We start with the first aspect, preferences
in interpretation. As an example, consider the following question in the context of a
visually presented rectangle:

(12) How long and how wide is this rectangle?

b
a
Obviously, there is a strong preference to refer to the maximal axis a with long (and to
the secondary axis b with wide). However, as noticed in Lang (1989: 349), there are
exceptions to the rule that long designates the maximal axis:

(13) a. The seed drill is wider than long

b. Our new double bed is 2 m long and 3 m wide
c. The velvet remnant is 1,3 m wide but only 0,5 m long.

Obviously, in cases where a non-maximal axis is designated, this axis is the most
salient for other reasons than spatial extent (salient direction of movement / salient
inherent orientation / prototypical designation). As a consequence, we should not
characterize the adjective long as referring to a maximal dimension. Instead, we
should take the lexical entry for long to be a candidate for a radical underspecification
12

and we should look for a mechanism of pragmatic strengthening that conforms to

internal competition and aims to assign the most salient axis in the given context.
A more complex example which is appropriate to make a similar point was already
presented at the beginning of this section and is repeated here for convenience:

(11) a. The windowsill is 1 m long, 30 cm wide and 3 cm thick

b. The windowsill is 1 m wide, 30 cm deep and 3 cm thick

The puzzling fact is that in contexts where we try to conceptualize an inherent

observer (example b) the adjective wide designates the maximal dimension.
However, in observer-free context (example a) there is a strong preference to
designate the secondary (next to maximal) dimension. This makes clear that the
adjective wide is another candidate for radical underspecification. The puzzling point
(to be resolved in the next section) is how to manage the extreme context-dependency
of this adjective.
Spatial adjectives provide an excellent area for studying the phenomenon of
blocking. The examples are legion, and so we can restrict ourselves to a short list. In
this list, the (a) examples block the corresponding (b) examples.

(14) a. The tower is 10 m high [vertical dimension]

b. ??The tower is 10 m long

(15) a. The pencil is 20 cm long [maximal dimension]

b. ?The pencil is 20 cm high (in particular contexts possible !)

(16) a. The tunnel is 2 km long [maximal dimension]

b. ?The tunnel is 2 km deep (in particular contexts possible !)

(17) a. The well is 10 m deep [vertical, observer direction]

b. ??The well is 10 m long / high

The examples (15b) and (16b) can be taken to illustrate the phenomenon of
deblocking: in particular contexts, the anomalies may disappear. As is discussed
elsewhere (e.g. Blutner 1998), the phenomenon of blocking / deblocking excludes a
classical treatment of such examples as simple violations of definite conditions.
Another domain where the effects of interpretational preferences, blocking and
deblocking come to the surface is the field of spatial prepositions (see Solstad 2000).
However, reasons of space force us to drop this extension here.
 Two case studies in lexical pragmatics 13

6 OPTIMAL DIMENSIONAL DESIGNATION

It is a common assumption of all theories of form perception that the description of

spatial relationships between the various parts of an object involves a frame of
reference or mental coordinate system. According to seminal work of Jackendoff
(1996) there are eight frames of reference to consider, four which are intrinsic
(geometric frame, motion frame, canonical orientation, canonical encounter) and
which determine the axis proper to an object, and four which are environmental
(gravitational frame, geographical frame, contextual frame, observer frame). For the
present discussion we can restrict ourselves to three frames of reference:

♦ geometric reference frame determining the axes proper to an object: intrinsic

♦ scene-based / gravitational frame

environmental
♦ observer frame

Following Jackendoff (1996) we assume a domain-specific and informationally

encapsulated module of spatial representation (SR). This module is distinct from
conceptual structure (CS). The usual assumption is that CS encodes propositional
representation and SR image schema or mental models (SR is geometric, but not
imagistic). Spatial information is encoded in SR, rather than in CS.
For convenience, we adopt Lang’s (1989) object scheme as a structural skeleton of
SR. In short, an object scheme represents that part of SR which is relevant for
specifying the spatial dimensions of the object. At first place, thus, an object scheme
contains a specification of the involved coordinate system(s). So, the object scheme
of the term brick may contain three perpendicular axes a, b, c which are ordered
intrinsically (a is the axis with maximal extent, b the secondary and c the axis with
minimal extent). Furthermore, the axis c is extrinsically specified as the vertical axis.

(18) <a b c> ‘intrinsic’ c

Vert ‘gravitational’ b
a
Similar object schemes apply for tombstone, beam (girder), board, and windowsill.
As we have seen in example (11), the term windowsill (and analogously, the other
terms) may trigger an additional object scheme which involves the observer
coordinate:

(19) <a b c> ‘intrinsic’

Vert ‘gravitational’
Obs ‘observer’
14

Following Jackendoff (1996), we can see the lexicon as the interface between the
language module and the modules CS and SR. For spatial dimensional adjectives it is
plausible to assume that an association with SR is most crucial. In the following we
assume that dimensional adjectives are discriminated by two factors:

♦ the reference frame they trigger, e.g.

- intrinsic: long, wide
- gravitational: high
- observer: deep, wide

♦ specificity, e.g.
{long, deep, high} > wide

The assumption that wide is the most unspecific adjective considered here derives
from the fact that it is related to two different frames (intrinsic & observer) whereas
the other adjectives refer to one frame only.
In subject-predicate expressions the reference frame triggered by the predicate
must be present in the SR of the subject term. This is our basic assumption that
determines the Generator. Roughly spoken, it is a realization of all the potential
pairings of a dimensional adjective with the designated axes of an object scheme
given by the predicate term. The only condition is that the reference frame triggered
by the adjective agrees with the designated axis.
In the case of our earlier example (12) the generator leaves this correlation
completely underspecified: Each of the two adjectives long and wide can be paired
with either of the two axes a and b. This fact is due to the lexical entries of long and
wide which only contain the information that both axes are intrinsic axes. The correct
correlations a-long and b-wide are realized by the basic mechanism of bidirectional
optimization (weak version). The discussion is completely analogous to that of the
schematic example (7). The two axes are intrinsically ordered by salience: a > b, and
the two adjectives are ordered by specificity: long > wide. From these two orderings
of the inputs / outputs the ordering of the adjective-axes pairs can be derived. This is
shown in (20):

(20) How long and how wide is this rectangle ?

b
a
The framework of bidirectional optimization then produces the right correlation which
can be seen as another reflection of the division of pragmatic labor: the salient axis
correlates with the more specific adjective.
Earlier solutions (e.g. Bierwisch 1967, Lang 1989) crucially deviate from the
present solution. They make use of features like MAX DIMENSION in case of long and
SECONDARY DIMENSION in case of wide. The model theory of these features defines a
kind of internal competition that simply stipulates the wanted result. The aim of the
present approach is to avoid such stipulations and to replace internal competition by
an external one. Needless to say, external competition is defined by the overall
 Two case studies in lexical pragmatics 15

framework of bidirectional optimization, and thus reflects claims that are motivated
independently (see Wilson (1998) for a general discussion of the relationship between
internal and external competition).
Let us finally present the analysis for a more complex example, (21).

(21) a. The brick is 24 cm long, 15 cm wide, 8 cm high

b. The brick is 24 cm wide, 15 cm deep, 8 cm high c
a b

Similar to example (11), we assume that the module SR realizes two different object
schemes for the term brick, one that doesn’t involve the observer–represented in (18),
and one that does – represented in (19). The tableau that corresponds to the first case
is the one in (22a). It involves the intrinsic frame and the gravitational frame. The
tableau that corresponds to the second case involves all three frames (intrinsic,
gravitational, observer). It is (22b).

(22) a. b.

Vert Obs
Vert
a b c a b c
high  high 

long    long   


wide  ()  wide  

deep  

wide  
16

The ranking of the different sub-tableaus is conform to the following general

assumption:

(23) If activated, the involved frames of reference are ranked as follows:

environmental > intrinsic

This assumption reflects the relative autonomy of the environmental frames relative to
the intrinsic frame. It is an easy exercise to determine the super-optimal solutions in
the tableaus (22a,b). In the first case, (21a), it comes out that the adjective long
designates the maximal axis a, the adjective wide the secondary axis b, and high the
vertical axis c. In the second case, (21b), it results that the observer-sensitive variant
of wide designates the maximal axis a. The adjective deep designates the observer
axis b, and high the vertical axis c. Notably, the use of the adjective long is blocked
if an observer axis is involved. The treatment of example (11) is analogous. However,
it involves a further dimension: substance (triggered by the adjective thick, cf. Lang
1989).

7 SUMMARY AND CONCLUSION

Investigating the interactions between the (mental) lexicon and pragmatics we have
pointed out that situated meanings of many words and simple phrases are
combinations of their lexical meanings proper and some superimposed conversational
implicatures. The basic approach of lexical pragmatics combines the idea of (lexical)
underspecification with a theory of pragmatic strengthening. The latter is formulated
in terms of a bidirectional OT formalizing Grice‘s idea of conversational implicature.
The mechanism of pragmatic strengthening crucially makes use of “non-
representational” parameters that are described by preferential relations, such as
information scales or salience orderings.
The main advantage of bidirectional OT is that it helps us to put in concrete terms
what the requisites are for explaining the peculiarities of negative strengthening,
dimensional designation and other potential phenomena that may be discussed. What
are the relevant cognitive scales? How do we measure morpho-syntactic markedness?
How do we measure the values of probabilistic parameters that control and organize
conceptual knowledge (salience, cue validity)?
An important challenge for the present view is the work done in relevance theory
(e.g. Sperber & Wilson 1986, Carston 1998, 2000, this volume). Although I prefer a
variant of Atlas’, Levinson’s and Horn’s framework, that doesn’t mean that I am
taking a stand against relevance theory. Rather, it seems desirable and possible to
integrate most insights from relevance theory into the present view. As a kind of
meta-framework, optimality theory can help to realize this integrative endeavor and
to bring the two camps closer to each other. Recently, van Rooy (2000a,b) made the
first important steps in this direction.
The general conclusion that can be drawn from the present analysis is that weak
bidirection can simplify the system of lexical stipulations rather radically. In the case
 Two case studies in lexical pragmatics 17

of negative strengthening, the interpretational effects of negating graded adjectives.

were treated by means of weak bidirection. This appropriately accounts for the
differences between positive and. negative adjectives, thus avoiding unmotivated
lexical stipulations. In the case of dimensional designation, on the other hand,
bidirectional OT helps to eliminate internal competition and to replace it by external
competition. Both analyses nicely illustrates Saussure’s view that the semantics of
natural language is partly determined by the inventory of lexical items.

REFERENCES

Atlas, J. and Levinson, S. (1981). It-clefts, informativeness and logical form. In:
Radical Pragmatics (P. Cole, ed.), pp 1-61. Academic Press, New York.
Bierwisch, M. (1967). Some semantic universals of German adjectivals. Foundations
of Language 3, 1-36.
Blutner, R. (1998). Lexical pragmatics. Journal of Semantics, 15, 115-162.
Blutner, R. (1999). Some aspects of optimality in natural language interpretation. In:
Papers on Optimality Theoretic Semantics (H. de Hoop & H. de Swart, eds.), Utrecht
Institute of Linguistics OTS, Uil OTS Working Paper, pp. 1-21. Also available from
http://www2.rz.hu-berlin.de/asg/blutner/pap.html
Boursma, P. (1998). Functional Phonology. Holland Academic Graphics, The Hague.
Carston, R. (1998). The semantics/pragmatics distinction: a view from relevance
theory’. UCL Working Papers in Linguistics, 10, 1-30.
Carston, R. (2000). Informativeness, relevances and scalar implicature’. Manuscript,
University College London.
Dekker, P. and van Rooy, R. (1999). Optimality theory and game theory: some
parallels. In: Papers on Optimality Theoretic Semantics (H. de Hoop & H. de Swart,
eds.), Utrecht Institute of Linguistics OTS, Uil OTS Working Paper, pp. 22-45.
Geurts, B. (to appear). Buoyancy and strength. Paper available from
http://www.kun.nl/phil/tfl/bart/
Hendriks, P. & Hoop, H. de (to appear). Optimal theoretic semantics. To appear in
Linguistics and Philosophy.
Hoop, H. de (2000). Optimal scrambling and interpretation. In: Interface Strategies
(H. Bennis, M. Everaert, and E. Reuland, eds.), pp. 153-168. KNAW, Amsterdam.
Hoop, H. de & Swart, H. de (1998). Temporal adjunct clauses in optimality theory.
Manuscript, OTS Utrecht.
Horn, L.R.. (1984). Toward a new taxonomy for pragmatic inference: Q-based and R-
based implicatures. In: Meaning, Form, and Use in Context (D. Schiffrin, ed), pp. 11-
42. Georgetown University Press, Washington.
Horn, L.R. (1989). A Natural History of Negation. University of Chicago Press,
Chicago.
Jackendoff, R. (1996). The architecture of the linguistic-spatial interface. In:
Language and Space (P. Bloom, M.A. Peterson, L. Nadel and M.F. Garrett, eds.),
MIT Press, Cambridge, Mass.
 18

Jäger, G. (2000), Some notes on the formal properties of bidirectional optimality

theory. In: Studies in Optimality Theory (R. Blutner and G. Jäger, eds.), pp. 41-63.
Linguistics in Potsdam, Potsdam..
Kiparsky, P. (1982). Word-formation and the lexicon. In: Proceedings of the 1982
Mid-America Linguistic Conference (F. Ingeman, ed.).
Lang, E. (1989). The semantics of dimensional designation of spatial objects. In:
Dimensional adjectives (M. Bierwisch and E. Lang eds.). Springer-Verlag, Berlin etc.
Levinson, S.C. (2000). Presumptive Meaning. The MIT Press, Cambridge, Mass.
McCawley, J.D. (1978). Conversational implicature and the lexicon. In: Syntax and
Semantics 9: Pragmatics (P. Cole, ed.), pp. 245-259. Academic Press, New York.
Rooy, R. van (2000a). Comparing questions and answers: a bit of logic, a bit of
language, and some bits of information. Manuscript ILLC, Amsterdam.
Rooy, R. van (2000b). Decision problems in pragmatics. Manuscript ILLC,
Amsterdam.
Solstad, T. (2000). The optimal lexicon and methods in lexical semantics. Project
outline.
Sperber, D. and Wilson, D. (1986). Relevance: Communication and Cognition. Black-
well, Oxford.
Williams, E. (1997). Blocking and anaphora. Linguistic Inquiry, 28, 577-628.
Wilson, C. (1998). Bidirectional optimization and the theory of anaphora. Manuscript,
Johns Hopkin University.
Zeevat, H. (1999a). Semantics in optimality theory. In: Papers on Optimality
Theoretic Semantics (H. de Hoop & H. de Swart, eds.), Utrecht Institute of
Linguistics OTS, Uil OTS Working Paper, pp. 76-87.
Zeevat, H. (1999b). Explaining presupposition triggers. Manuscript AC99, Univers-
ity of Amsterdam. Available from http://www.hum.uva.nl/computerlinguistiek/henk/

View publication stats