Grice: Breaking The Code

Speranza

Comments on S. Assimakopoulos's "On Encoded Lexical Meaning:
Philosophical and Psychological Perspectives"

Grice would never speak of the code, never mind think of breaking. Yet, scholars like a keyword, and Grice will be forever featured as a proponent of the "INFERENTIAL MODEL OF COMMUNICATION", whatever that is. The idea is that there is a code, and that there is games with the code, or inferences with the code. It may even be propounded, in an exaggerated way, that Herbert Paul Grice is a non-code theorist!

The past few years have seen quite a bit of speculation over relevance
theorists’ commitment to Fodorian semantics as a means to account for
the notion of encoded lexical meaning that they put forth in their
framework. In this paper, I take on the issue, arguing that this view of
lexical semantics compromises Relevance Theory’s aim of psychological
plausibility, since it effectively binds it with the ‘literal first’ hypothesis
that has been deemed unrealistic from a psycholinguistic viewpoint.
After discussing the incompatibility of Fodor’s philosophical account
with the perspective that relevance theorists adopt, I briefly suggest
ways in which further behavioural research on the
semantics/pragmatics distinction could help advance more cognitivelyoriented
accounts of encoded lexical meaning.
Keywords: lexical pragmatics, Fodorian semantics, relevance theory,
“literal-first” hypothesis, polysemy.
(While preparing this paper the author  has benefited immensely from discussions with Lewis Bott, Ronnie
Cann, Robyn Carston, Billy Clark, Felicity Deamer, Albert Gatt, Alison Hall, Anne Reboul, Nicola
Spotorno, John Tomlinson Jr, Daniel Wedgwood and Deirdre Wilson. Of course none of the above is
to be held responsible for the views expressed in this paper, which I would after all expect many of
them to find objectionable. I am also particularly indebted to Francesca Ervas for her kind invitation to
this special issue, editorial guidance, and, above all, incredible patience with me. The research
presented herein has been financially supported by a travel grant provided by the ESF Euro-XPrag
Network for collaborative research with John Tomlinson Jr, Lewis Bott and Julio Santiago de Torres,
under the title “Varieties of lexical adjustment”, as well as the Junta de Andalucía project entitled
“Contemporary Naturalism and Pragmatism” (P08.HUM.04099).

Ever since its inception in the seminal work of the Oxford philosopher Herbert Paul Grice
(1989), contemporary linguistic pragmatics has been typically viewed as an
add-on to the more fundamental study of semantics, which deals with linguistic
meaning in abstraction of its context of use.

In this view, the recognition of
speaker intentions, albeit instrumental for the recovery of ‘what is implicated’
by the utterance of a sentence, has been supposed to play little, if any role in
the delineation of its semantic content, i.e. what Grice is customarily taken to
have originally referred to as ‘what is said’ by it.
This way of distinguishing between semantic content and pragmatic import
may still reflect the received view in the work of most contemporary
researchers who are interested in linguistic meaning, but it has been repeatedly
attacked in the past few decades. On a number of occasions, various scholars1
have pursued the argument that, even over and above the obvious contextsensitivity
of indexical expressions (such as I, you, here, now, etc), ‘what is
said’ by a sentence cannot always be identified in isolation from the context of
its utterance. Among them, relevance theorists have right from the beginning
(at least as far back as Wilson & Sperber, 1981) insisted that the linguistically
encoded meaning of an utterance, that is, its underlying sentence’s semantics,
falls short of determining the proposition explicitly expressed by it, and that
the hearer has to undertake processes of pragmatic inference in order to work
this proposition out. The thorough investigation of this claim primarily by
Robyn Carston in a number of publications, culminating in her Thoughts and
Utterances (2002), has eventually led Relevance Theory (henceforth RT) to
take up a radical version of it, according to which, “linguistically encoded
meaning never fully determines the intended proposition expressed” (Carston
2002, p. 49, emphasis in original).
The espousal of this position, which has been dubbed the linguistic
underdeterminacy thesis , has obviously placed RT in direct opposition to the
traditional way of carving the semantics/pragmatics distinction at the
propositional level. Quite predictably, this has in turn sparked the reaction of
various philosophers of language, who have criticized this deviation from the
long-established way of studying linguistic meaning, counter-proposing
1 See, for example, Searle (1978), Travis (1981, 1997), Sperber & Wilson (1986/1995), Carston
(1988, 1999, 2002, 2004, 2008), Atlas (1989, 2005), Recanati (1989, 2001, 2002, 2004,
2010), Bach (1994a, 1994b, 1999), Levinson (2000), Jaszczolt (2005).

minimalist theories of semantic content that defend the context-independence
of the proposition explicitly expressed by an utterance (e.g. Borg, 2004;
Cappelen & Lepore, 2005). However, even minimalists themselves have at
times recognized that “if communication can be shown to proceed without
hearers processing the literal meaning of the sentence, i.e. without grasping
minimal propositions, then the claim that minimal propositions have a unique
role to play in actual communicative exchanges is undermined” (Borg, 2007,
p. 353). In this respect, the linguistic underdeterminacy thesis seems to be
vindicated by the existence of a quite substantial body of psycholinguistic
research which has shown that the comprehension of figurative language does
not necessarily involve the prior processing of the surface literal meaning (for
representative overviews, see, Gibbs, 1994; Glucksberg, 2001).
Moving from propositional to lexical meaning, it is this very psychological
implausibility of the so-called ‘literal first ’ hypothesis that has also motivated,
as Deirdre Wilson herself notes (2011, p. 15), one of the latest developments
within RT, that is, its account of ad hoc concept construction. Based on
experimental research which has shown that our categorisation behaviour is
highly context-dependent (e.g. Barsalou, 1987, 1992), relevance theorists
have advanced the argument that the concept communicated through the use of
some particular lexical item can be distinct from the concept it encodes, and
thus requires a spontaneous process of pragmatic enrichment to be reached at
during interpretation.

Even though the experimental investigation of the particular proposals that
relevance theorists have put forth with respect to lexical pragmatics is still very
limited, I will attempt in this short paper to show that their theoretical
proposals have far-reaching implications for the discussion of word meaning,
to the extent that they could even challenge the current view of encoded lexical
meaning within RT itself. To this end, I will start off with a brief overview of the
framework’s assumptions that are relevant to the present discussion and will
then move on to assess the account of lexical pragmatics put forth by relevance
theorists and the tension it creates for traditional approaches to lexical
semantics and, more specifically, the Fodorian one that relevance theorists
have adopted right from the beginning. Wrapping up this paper, I will consider
how the present argumentation can motivate new directions for behavioural
research on the semantics/pragmatics interface.

Right from its emergence, RT has aimed at providing a thoroughgoing
cognitive account of utterance interpretation. For this reason, relevance
theorists draw the semantics/pragmatics distinction in terms of the different
kinds of mental processing they take each of these types of meaning to be the
output of, instead of discussing them at a theoretically abstract level, as most
philosophers have so far tended to. From their perspective, semantic content is
provided via decoding, which is performed by an autonomous linguistic mental
module, while pragmatically derived meanings are taken to be generated by an
inferential processor, which is in turn dedicated to the comprehension of
deliberately communicated stimuli and effectively integrates the output of
decoding with readily available contextual assumptions in the interest of
calculating a reasonable hypothesis about the original speaker-intended
meaning. Without getting into too much detail, which is after all unnecessary
for my current purposes, RT predicts that a hearer will automatically
comprehend a deliberately communicated utterance by following a path of least
effort, according to which, he will assess interpretive hypotheses in order of
accessibility until his expectations for an interpretation that will uncover the
speaker’s intended meaning are satisfied (or, in the case of miscommunication,
abandoned).
In the current setting, the crucial aspect of the RT account is that it does
not take inference to work on the overall output of decoding during the
comprehension of a single utterance, as a traditional Gricean approach would
have it; rather, the two modules work simultaneously, with the decoding one
feeding input to the inferential every step of the way during the processing of
the linguistically encoded stimulus. Obviously, the replacement of Fregeanstyle
thoughts with subjective and context-dependent propositions – what
relevance theorists call explicatures – makes it tempting to assume, as semantic
minimalists have on occasion, that the framework encompasses some radically
contextualist notion of semantics in its premises. On closer inspection,
however, it turns out that this is not the case. As Daniel Wedgwood (2007)
extensively discusses, the sole difference between the minimalist’s way of
describing semantic content and the relevance-theoretic notion of ‘encoded
meaning’ lies in the contention of the former that sentences do actually encode
full propositions; other than that, encoded meaning is equally ‘properly’
On Encoded Lexical Meaning: Philosophical and Psychological Perspectives 21
semantic, in the traditional sense, for relevance theorists too. Considering the
following passage from Relevance, this conclusion seems to be warranted:
By definition the semantic representation of a sentence, as assigned to it by a
generative grammar, can take no account of such non-linguistic properties as,
for example, the time and place of utterance, the identity of the speaker, the
speaker’s intentions, and so on. The semantic representation of a sentence
deals with a sort of common core of meaning shared by every utterance of it.
(Sperber & Wilson, 1986/1995, p. 9)
This identification of encoded meaning as essentially context-independent
within RT also follows from the adherence of relevance theorists to Fodorian
semantics. As Carston asserts (2002, p. 58), RT follows Fodor in assuming
that ‘real’ semantics deals with the explication in truth-conditional terms of the
relation between our mental representations and that which they represent,
while linguistic semantics is merely ‘translational’, in the sense that publiclanguage
forms inherit their meaning directly from the Mentalese forms they
encode. Therefore, since in the resulting picture the only actual bearer of
semantic content is Mentalese, which consists in concepts, and, as Jerry Fodor
has it, a concept is an unanalysable, monolithic atom that is individuated by
some property of the real-world entity to which it is nomologically locked, it
becomes virtually impossible for encoded semantic contents to vary across
different contexts.
Therefore, even though relevance theorists maintain that sentences cannot
be attributed any fully propositional semantics, they cannot but accept that
lexical items2 do encode context-independent meanings; that is, the ones that
they directly inherit from their associated atomic concepts.3 And this is indeed
what they seem to have had in mind ever since they first entertained the idea
that “words in a language can be used to convey not only the concepts they
encode, but also indefinitely many other related concepts to which they might
point in a given context” (Sperber & Wilson, 1998, p. 197), since, in the
standard RT picture, in order for such ad hoc concepts to be constructed, the
corresponding encoded concepts crucially need to be used as a starting point.
In order to briefly illustrate the rationale behind the RT account of ad hoc
concept construction now, let’s consider the following examples:
2 Much like most relevant discussions, the present one deals exclusively with monomorphemic ‘openclass’
lexical items, i.e. words that have some descriptive content (unlike indexicals, connectives and
the like).
3 For a recently presented, yet still speculative, alternative view, see Carston (2012).

(1) John has a temperature.
(2) The fridge is empty.
It should be pretty straightforward that in order to come up with the
proposition explicitly expressed by the utterances (1) and (2) in certain, most
likely familiar, contexts, the hearer would have to somehow adjust the encoded
meaning of ‘temperature’ and ‘empty’. In (1), the word ‘temperature’ could be
easily attributed the interpretation of ‘a high temperature’ rather than its actual
denotational content provided by the concept TEMPERATURE, as this can be
taken to be used in utterances like ‘Celcius is a scale for temperature
measurement’. Similarly, in (2) the fridge might not be interpreted as being
totally empty, but rather as being insufficiently filled with the goods that are
needed by a household on a daily basis. In this case, the encoded concept
EMPTY would again need to be adjusted so that the ‘not entirely empty, but
insufficiently full’ interpretation can be yielded.
According to the RT account, the construction of ad hoc concepts, like
TEMPERATURE* and EMPTY* (to use their common notation), is the
outcome of two pragmatic processes that can either individually or in unison
contextually adjust the meaning that a lexical item’s encoded concept carries.
The first one, which is dubbed narrowing, results in meanings that are typically
more specific than the encoded ones, such as the one of TEMPERATURE* in
(1), while the second, broadening, respectively generates looser word
interpretations. Apart from general approximation cases, like the one of
EMPTY* exemplified above, concept broadening is also assumed to mediate
the interpretation of hyperboles and metaphors, as well as category extensions
(e.g. when a brand name, like ‘typex’ is used as an umbrella term for all
products with a common function), neologisms and word coinages.4
As I have already noted in the previous section, the RT account of lexical
pragmatics was originally motivated by Lawrence Barsalou’s behavioural
research on conceptual categorisation, a point that in itself gives the account a
quite high degree of psychological plausibility. However, it has often been
noted5 that this plausibility is compromised when it is coupled with Fodor’s
philosophical semantics, which relevance theorists have adopted for the
4 For detailed overviews of the RT account of lexical pragmatics, see Carston (2002: chapter 5),
Wilson (2004), Wilson & Carston (2007).
5 See, for example, Vicente (2005), Burton-Roberts (2007), Groefsema (2007), Assimakopoulos
(2008), Reboul (2008), Vicente & Martinez Manrique (2010).
On Encoded Lexical Meaning: Philosophical and Psychological Perspectives 23
purposes of describing what gets decoded and fed into the inferential
processor during interpretation. As Anne Reboul puts it, “the notion of an ad
hoc concept has rendered visible a long-standing tension in Relevance Theory,
viz. that between the adoption of an (atomistic and externalist) view of concepts
(such as Fodor’s) and the description that Relevance Theory effectively gives to
concepts” (2008, p. 523). In the following section, I will turn to this tension
and will attempt to show that the argument that relevance theorists have used
against its existence runs into trouble when psychological considerations enter
the picture.
2. Ad hoc concepts and the literal-first hypothesis
Based on descriptions similar to the one provided above, albeit much more
detailed of course, Vicente (2005) and Groefsema (2007) have justifiably
noted that the formation of communicated lexical meanings by means of
broadening and narrowing suggests that encoded concepts must have some
kind of internal structure, as it is only by way of manipulating such a structure
that the construction of a speaker-intended ad hoc concept can be made
possible. Naturally, if this is the case, encoded concepts cannot be conceived of
as Fodorian atoms to begin with, since it is by definition impossible to either
‘narrow down’ or ‘loosen up’ a non-decomposable atom. This criticism seems
to be further motivated by the description of concepts within RT itself,
according to which, logical, lexical and encyclopaedic information is standardly
assumed to be stored in different entries of a conceptual address. Given the
explicit recognition of such different types of information associated with a
concept and their implementation in various discussions of ad hoc concept
construction, like, for example, when Carston suggests that, in narrowing, an
encyclopaedic property of a lexically encoded concept can be ‘elevated’ to a
logical (or content-constitutive) status (2002, p. 339), it certainly becomes all
the more tempting to assume that encoded concepts must have some more
substantive content than a monolithic atom would normally allow for in order
for ad hoc concepts to be constructed on their basis.
The way in which RT addresses this criticism, however, can be easily
located in Carston’s parallelism (2010a, pp. 174–175, fn.6) of the relevancetheoretic
notion of a conceptual address with what Fodor has recently called a
‘mental file’:

When you are introduced to John […] you assign him a Mentalese name and
you open a mental file, and the same Mentalese expression (M(John)) serves
both as John’s Mentalese name and as the name of the file that contains your
information about John; […] according to this story, we think in file names;
tokens of file names serve both as the constituents of our thoughts and as the
Mentalese expressions that we use to refer to things we think about. (Fodor,
2008, pp. 94–95, emphasis in original)
Taking into consideration Fodor’s description, it becomes clear that from
the corresponding RT perspective, the various kinds of information that are
thought to be associated with a concept do not form part of its semantic
content per se, and thus play no role whatsoever in the decoding process as far
as relevance theorists are concerned. Consequently, as a theoretical construct,
an ad hoc concept would not appear to pose any particular problems for the
way in which RT views semantics, as it is essentially the output of inference,
with the input from decoding being solely the respective atomic concept (i.e.
the Fodorian mental file name).
This line of argument, which Carston (2010a, 2010b) has followed in
response to a slightly different, but comparable critique that Vicente &
Martínez Manrique (2010) have put forth, is certainly reasonable when it
comes to the deflation of the argument that relevance theorists would be better
off employing a decompositional picture of lexical semantics rather than
Fodor’s atomistic account in their framework; yet, when the overall RT aim of
developing a cognitively realistic account of communication is taken into
consideration, it seems to be binding relevance theorists with a view that, as we
have seen, they otherwise explicitly seek to distance themselves from, i.e. the
‘literal-first’ hypothesis.
Given the current RT account of ad hoc concept construction, according to
which, the inferential enrichment of encoded concepts is standardly treated as
an optional, top-down process, if decoding provides the inferential processor
with the content of a word’s encoded concept, this content cannot but be the
first interpretive hypothesis that the hearer will test for relevance during the
comprehension procedure. Consider, for example, the meaning
communicated by the word ‘temperature’ during the interpretation of ‘John
has a temperature’ in the aforementioned context in which John has a fever.
Here, according to the current RT view, the output of the decoding of
‘temperature’ would be the concept TEMPERATURE, which carries the real,
context-insensitive, and hence literal semantic content associated with the
On Encoded Lexical Meaning: Philosophical and Psychological Perspectives 25
word. For the inferential processor to construct the relevant ad hoc concept
TEMPERATURE*, it will need to do so after testing the content of this
encoded concept as a plausible hypothesis about the speaker-intended
meaning, since, being by definition an unanalyzable atom, TEMPERATURE
will necessarily be wholly employed in the process. But if the comprehension of
figurative language is equivalent to that of literal meaning in processing terms,
as the relevant experimental evidence suggests, the priority of this encoded
meaning over the pragmatically enriched one is seriously compromised.6 Given
the mechanics of the cognitive systems that RT posits, even Wilson’s recent
suggestion that “the concept encoded by a word is activated during
comprehension, but not necessarily deployed (2011, p. 16, emphasis in
original) seems unsatisfactory, since again RT currently has no way of
accounting for the activation of a concept without its initial incorporation in
(and, if deemed unsatisfactory, potential discarding from) the mental
representation that the inferential processor calculates as an utterance’s basic
explicature. In general, if the encoded denotational content of any concept can
be bypassed during comprehension, it follows that the inferential processor has
some way of discarding ‘irrelevant’ lexical meanings before actually assessing
them as intended interpretations. But since the inferential processor’s task is
precisely to carry out this assessment in the first place, it has no way of knowing
beforehand which encoded concept it will eventually keep intact and which it
will need to enrich into an ad hoc concept.
If this line of reasoning is on the right track, it reveals a challenge that RT
would need to tackle in order to satisfy its overarching aim of psychological
plausibility. And while the relevant literature has focused almost exclusively on
the implications that the account of ad hoc concept construction carries for the
discussion of a concept’s internal composition (or lack thereof), I believe that
an equally important question that needs to be addressed is how well the
philosophical discussion of lexical meaning that RT clings to can fit its
6 A potential counter-argument that has been brought to my attention is that the experimental
evidence against the literal-first hypothesis only carries implications for the discussion of
propositional and not for that of lexical meaning. I think this is highly debatable since the metaphors
used in the relevant literature often consist of a topic followed directly by the metaphor vehicle, as in
the case of ‘Her surgeon was a butcher’ or ‘My job is a jail’. According to the current RT account of
lexical pragmatics then, it is only the concepts BUTCHER and JAIL that would need to get enriched
for the figurative interpretation to become available, which effectively means that metaphor
interpretation pertains more to the discussion of communicated meaning at the lexical rather than the
propositional level (in at least such cases).
psychological orientation. As I will now turn to argue the two perspectives are
quite hard to reconcile, and for good reason.
3. Philosophy and the psychology of encoded
lexical meaning (all too briefly)
As we have already seen, much like Fodor, relevance theorists take on the
commonplace assumption that semantic theory aims at providing an account of
linguistic meaning at a level of abstraction from its actual use; ‘a sort of
common core of meaning shared by every utterance of a sentence’. Given the
RT on-line processing picture, however, it is only natural to expect that the
inferential processor will have already enriched bits and pieces of an utterance
by the time it has been fully heard in an actual communicative setting. In this
respect, even though the decoding of this utterance will generate a
concatenation of context-independent conceptual representations, by the time
an utterance’s explicitly expressed meaning is constructed, it will inevitably
present various degrees of deviation from the type proposition that Fodor’s
semantic theory puts forth. Following this rationale, Carston recently observed
that even if a sentence’s encoded meaning did somehow turn out to typically
express a full proposition like the minimalist holds, the repercussions of this
discovery would not be “a devastating blow for the central tenets of RT”, as
“the propositions concerned would usually be very weak/general or absurdly
strong, often either truisms or obvious falsehoods”, which would “almost never
be the sort of contents that speakers want to communicate” (2010b, p. 268).
Indeed, considering the particular cognitive processing that mediates the
comprehension of linguistic stimuli against the traditional philosophical
context of studying semantics, Carston’s remark appears to be on the right
track, but from the very same psychological perspective, a pressing question
also arises: if, without any contextual input, the thoughts that ‘there are cats’ or
that ‘it’s raining’ that Fodor alludes to in his discussions are never ‘the sort of
contents that speakers want to communicate’, is there any principled reason
for which we need to accept that they are thoughts that we ever even entertain?
And if the answer to this question is negative, as I think it is, what is the reason
for which we need to maintain that these semantic contents are actual thoughts
- rather than artificial examples pertaining to an abstract model of thought - to
begin with?

Turning to lexical semantic content, this problem becomes even more
obvious. As Carston herself discusses (2002, p. 360),
Focusing on the word ‘happy’, let’s consider the concept that it is supposed to
encode, a concept which is to provide communicative access to a wide range of
other more specific concepts […]. The idea is that the lexically encoded
concept HAPPY is distinct from all of these; it is more general and abstract than
any of them, but provides the basis, in appropriate contexts, for processes of
pragmatic enrichment so that addressees can come to grasp one of the more
specific concepts and incorporate it into their representation of the speaker’s
thought. But what is not at all clear is whether we ever actually have (hence
sometimes try to communicate) thoughts in which this very general lexicalized
concept features as a constituent, or indeed what the property of being HAPPY
is, as opposed to being HAPPY* or HAPPY**, etc.
Clearly, this worry is not exclusive to ‘happy’, but rather seems to present
itself when the encoded meaning of any gradable adjective, where no absolute
denotational property exists, or even that of commonly used verbs like ‘open’
or ‘stop’ are put into scrutiny.7 Even when we turn to nouns, the postulation by
relevance theorists of ad hoc concepts in the mind, raises important questions
regarding their implementation in the individual’s everyday thinking too. So,
when the doctor thinks that John has a temperature in our familiar by now
context, is she thinking that he has a TEMPERATURE or rather a
TEMPERATURE*? Similarly, when Mary thinks that she wants to meet a
bachelor, to use another well-worn example from the RT literature, does she
implement in her mental processing the concept BACHELOR, whose
denotation includes all male individuals who are not married, or the narrower
concept BACHELOR*, whose denotation includes those male individuals who
are not married, but who would also be eligible candidates for her to marry
(obviously not the Pope or some very old or gay man)? If, as I take it, the
answer to these questions points to the ad hoc rather than the encoded
concept, it follows that these concepts coexist in our conceptual repertoire
alongside their Fodorian realist counterparts TEMPERATURE and
BACHELOR. And even if relevance theorists argue that “most occasional
representations of a property (or an object, event or state) do not stabilise into
a concept” (Sperber & Wilson, 1998, p. 198), they would still have to accept
that at least some ad hoc meanings, which are very often used in everyday
7 For the arguments here, see Sperber & Wilson (1998) and Carston (2012) respectively.
28 Humana.Mente – Issue 23 – December 2012
communication, like TEMPERATURE* and BACHELOR* (possibly also
BACHELOR**, a stereotypical bachelor who is untidy, or even
BACHELOR***, a man who seeks ephemeral relationships and so on and so
forth), eventually get to be stored in the mental lexicon; and this time these
seemingly ad hoc concepts would effectively be ‘semantic’, that is, decodable
rather than inferred.
This rampant encoded polysemy , as Vicente and Martínez Manrique
(2010) have aptly called it, would be problematic from a philosophical
viewpoint, as it effectively violates Grice’s Modified Occam’s Razor. And
although Carston has noted that such a proliferation of word senses in the
lexicon would not necessarily be problematic for RT, since “within a theory of
utterance interpretation conceived as a matter of on-line cognitive processes, it
might well be more economical to retrieve a clutch of stored senses and choose
among them, than to construct an interpretation out of a single sense and
contextual information” (2002, p. 219, fn.50), when combined with the
preceding discussion of the encoded meaning of words like ‘happy’ or ‘open’,
this remark certainly raises questions regarding the relevance theorists’ need
to postulate single, general concepts that encode the meaning of such words in
the first place.
Fodor has based his account of semantics on the presupposition that the
content of a natural language sentence or a lexical item is entirely isomorphic
to some determinate thought or atomic concept that they correspondingly
encode. In this respect, apart from the few cases of homonymy, as in the two
distinct meanings of the word ‘bank’, a lexical item carries a single meaning
that is referentially derived. That is largely because of the issues that he has
sought to address in the first place; issues for which context-sensitivity has
traditionally been thought of as problematic, such as compositionality, the
assignment of satisfaction conditions to semantically evaluable expressions,
intentional explanation and so on and so forth. But this isomorphism does not
work when psychological considerations enter the picture, a point that has
been made by Sperber and Wilson themselves from at least as far back as
(1998). When it comes to actual verbal communication, even Fodor agrees
that “language is strikingly elliptical and inexplicit about the thoughts it
expresses” (2001, p. 11), but this does not compromise his account, since it
has little, if anything, to do with the actual processing of linguistic stimuli per
se; it is “an account of the metaphysical character of the (primitive) semantic
properties and relations” rather than “a specification of the semantic
On Encoded Lexical Meaning: Philosophical and Psychological Perspectives 29
properties of the expressions in a language” (Fodor, 2008, p. 18, fn.34). In
fact, like most philosophers, Fodor has emphasized time and again that
confusing psychology with semantics “is a very bad idea” and that “there is, as
a matter of principle, no such thing as a psychological theory of meaning ”,
since semantics is by definition about “constitutive relations between
representations and the world” (Fodor, 2008, p. 88, emphasis in original).
Relevance theorists are in all likelihood equally aware of the more general
dichotomy of interest between philosophers and psychologists who study
linguistic meaning as they seem to acknowledge that “it is far from obvious that
the label ‘concept’ refers to the same entity for both parties (and very clear that
the term ‘semantic’ does not) so that little conciliatory progress is likely to be
made until these differences are mapped out and resolved” (Carston, 2010a, p.
175, fn.8). Even so, they choose to ignore semantics from their research
agenda. For instance, in a recent paper, Carston indirectly responded to the
criticisms that RT has been receiving regarding what minimalists perceive to be
its semantic commitments by noting that discussions concerning semantics are
not central to what the theory is all about and suggesting that the label ‘radical
pragmaticism’ fits the theory’s orientation much better than ‘radical
contextualism’. As she argues, “it is us, the users of language, who are sensitive
to context, and, as rational communicating/interpreting agents, we are able, by
exploiting this sensitivity in each other, to get linguistic expressions to do a lot
more than simply express their standing linguistic meaning” (2010b, p. 266).
In this way, Carston distinguishes the study of the cognitive processing that
underlies linguistic communication from that of semantic content, or ‘standing
linguistic meaning’ as she calls it. Since RT’s concern has always been to
account for the ways in which the dedicated inferential process enriches and
complements the semantic representation of linguistic strings, the argument
goes, it should have nothing more to say about the nature of these
representations other than that they are structured strings of Fodorian-style
atomic concepts, which in turn need considerable contextual enrichment to
reach full propositional status.
As we have seen, however, the implementation of Fodorian-style concepts
as actual processing units that lexical decoding feeds into the inferential
processor during utterance interpretation appears to be creating problems in
its own right; from a psychological perspective, RT’s inability to escape the
literal-first hypothesis is a case in point, while from a philosophical one, the
proliferation of word senses creates an uncontrollable system, where it
30 Humana.Mente – Issue 23 – December 2012
becomes very difficult to keep track of how many senses a word effectively
encodes. Against this background, relevance theorists might need to explore
alternative ways of accounting for encoded lexical meaning, since abstract
philosophical models of lexical semantics of the type that Fodor offers do not
fit the bill, nor are they supposed to. Therefore, a more psychologicallyoriented
approach to the question of what an expression’s encoded linguistic
meaning effectively is seems needed and, to this effect, behavioural research
from the domain of psycholinguistics would undoubtedly have a pivotal role to
play.
4. Experimental prospects

Given the detailed account of lexical pragmatics that relevance theorists have
recently developed, it would certainly be interesting in its own right to see the
extent to which the experimental research that has challenged the ‘literal first’
hypothesis with respect to the processing of figurative language could also be
applied to the study of the processing of other types of pragmatically enriched
lexical interpretations, like narrowings, approximations, category extensions,
neologisms and word coinages. As Wilson and Carston note, some preliminary
data on examples from the last two categories (Clark & Clark, 1979; Clark &
Gerrig, 1983) suggest that they are “no harder to understand than regular
uses” (2007, p. 237), but it remains to be seen how fast the interpretation of
neologisms and word coinages takes place in comparison to the processing of
literal meaning. Turning to cases of lexical narrowing, like the one presented
in (1), and approximation, like the one in (2), I think it would be quite
counterintuitive to expect any delays in their processing, but again there is, to
my knowledge, a complete lack of experimental evidence to support this
intuition.
From the current discussion’s perspective, experimentation on different
varieties of lexical meaning adjustment would be able to give us a clearer idea
of how different types of enriched interpretations of a word relate to its literal
meaning. If no significant difference is documented, this would strengthen not
only the assumption that literal lexical meaning is in no way exceptional but
also the need to come up with more psychologically-oriented accounts of the
mental lexicon and more essentially its particular contribution during verbal
communication. One of the few experimental studies on lexical pragmatics
from an RT perspective that has surfaced in the recent years, for example,
suggests that certain context-independent properties of a word’s meaning
On Encoded Lexical Meaning: Philosophical and Psychological Perspectives 31
remain activated even after a metaphorical (Rubio-Fernández, 2007) or
narrower interpretation (Rubio-Fernández, 2008) have been reached.
However, even though this set of experiments has been extensively quoted in
the RT literature, no full explanation has been provided about what these
properties are and how they relate to the equally context-independent atomic
concept that the same word encodes. Foreseeing that a ‘core’ relevance
theorist’s response would be that they are part of the information that is
attached to the word’s conceptual address and thus not part of its content (as
per the discussion of Fodorian mental files above), it still is curious why these
particular properties are obligatorily activated during spontaneous
interpretation, which in itself gives the impression that they are decoded rather
than inferred.
It goes without saying of course that the suggested behavioural research
would not be without its limitations either. For instance, it can certainly be
argued that it is dangerous to rely too much on time-reaction measurements,
which most of the relevant experiments have implemented either way, since
they cannot, on their own provide any direct evidence for the contention that
we actually interpret literal and non-literal language by using the same types of
mental processes.8 To this effect, using more advanced experimental
techniques, and also potentially looking into what neurolinguistics would have
to say could provide us with much more solid conclusions. Regardless of any
such limitation, however, I am convinced that the tension between the
philosophical analysis of lexical meaning and its psychological consideration
that the RT discussion of ad hoc concept construction seems to have
involuntarily revealed can open up new and exciting prospects not only for
theoretical analysis, but also for experimental research on the
semantics/pragmatics interface, at a time when the field seems to be dealing
almost exclusively with tropes, scalar implicatures and presupposition
projection.
8 For a discussion along these lines from a psycholinguistic perspective, see McElree & Nordlie
(1999), a paper suggested to me by John Tomlinson Jr.
32 Humana.Mente – Issue 23 – December 2012

REFERENCES

Assimakopoulos, S. (2008). Logical structure and relevance . University of Edinburgh,
PhD dissertation.
Atlas, J.D. (1989). Philosophy without ambiguity . Oxford: Oxford University Press.
Atlas, J.D. (2005). Logic, meaning, and conversation: semantic underdeterminancy,
implicature, and their interface . Oxford: Oxford University Press.
Bach, K. (1994a). Conversational impliciture, Mind & Language , 9, 124-162.
Bach, K. (1994b). Semantic slack: what is said and more, in S. Tsohatzidis (Ed.),
Foundations of speech act theory: philosophical and linguistic perspectives .
London: Routledge, 267-291.
Bach, K. (1999). The semantics-pragmatics distinction: what it is and why it matters,
in K. Turner (Ed.), The semantics/pragmatics interface from different points
of view, Amsterdam: Elsevier, 65-84.
Barsalou, L. (1987). The instability of graded structure: implications for the nature of
concepts, in U. Neisser (Ed.) Concepts and conceptual development:
ecological and intellectual factors in categorization. Cambridge: Cambridge
University Press, 101-140.
Barsalou, L. (1992). Frames, concepts, and conceptual fields, in E. Kittay & A. Lehrer
(Eds.), Frames, fields, and contrasts: new essays in semantic and lexical
organization. Hillsdale, NJ: Lawrence Erlbaum Associates, 21-74.
Borg, E. (2004). Minimal semantics. Oxford: Oxford University Press.
Burton-Roberts, N. (2007). Varieties of semantics and encoding: negation,
narrowing/loosening and numericals. In N. Burton-Roberts (Ed.), Pragmatics.
Basingstoke: Palgrave Macmillan, 90-114.
Cappelen, H., Lepore E. (2005) Insensitive semantics: a defense of semantic
minimalism and speech act pluralism. Malden: Blackwell.
Carston, R. (1988). Implicature, explicature, and truth-theoretic semantics. In R.
Kempson (Ed.), Mental representations: the interface between language and
reality. Cambridge: Cambridge University Press, 155-181.
On Encoded Lexical Meaning: Philosophical and Psychological Perspectives 33
Carston, R. (1999). The semantics/pragmatics distinction: a view from relevance
theory. In K. Turner (Ed.), The semantics/pragmatics interface from different
points of view. Amsterdam: Elsevier, 85-125.
Carston, R. (2002). Thoughts and utterances: the pragmatics of explicit
communication. Oxford: Blackwell.
Carston, R. (2004). Explicature and semantics. In S. Davis & B.S. Gillon (Eds.),
Semantics: a reader. Oxford: Oxford University Press, 817-845.
Carston, R. (2008). Linguistic communication and the semantics/pragmatics
distinction, in Synthese, 165, 321-345.
Carston, R. (2010a). Lexical pragmatics, ad hoc concepts and metaphor: from a
relevance theory perspective, Italian Journal of Linguistics, 22, 153-180.
Carston, R. (2010b). Explicit communication and ‘free’ pragmatic enrichment. In B.
Soria & E. Romero (Eds.), Explicit communication: essays on Robyn Carston’s
pragmatics. Basingstoke: Palgrave Macmillan, 217-287.
Carston, R. (2012). Word meaning and concept expressed, The Linguistic Review,
29, 607-623
Clark, E,. Clark, H. (1979). When nouns surface as verbs, Language, 55, 767-811.
Clark, H., Gerrig, R. (1983). Understanding old words with new meanings, Journal of
Verbal Learning and Verbal Behavior , 22, 591-608.
Fodor, J.A. (2001) ‘Language, thought and compositionality’, Mind & Language , 16,
1-15.
Fodor, J.A. (2008). LOT 2: the language of thought revisited. Oxford: Oxford
University Press.
Gibbs, R.W.Jr. (1994). The poetics of mind: figurative thought, language and
understanding. Cambridge: Cambridge University Press.
Glucksberg, S. (2001). Understanding figurative language: from metaphors to idioms.
Oxford: Oxford University Press.
*****************************************************************
Grice, H. P. (1938). Negation. The Grice Papers. UC/Berkeley, Bancroft Library.
Grice, H. P. (1989). Studies in the ways of words . Cambridge, Mass.: Harvard University
Press.
************************************************************************
Groefsema, M. (2007). Concepts and word meaning in relevance theory. In N. Burton-
Roberts (Ed.), Pragmatics. Basingstoke: Palgrave Macmillan, 136-157.
Horn, L. R. A brief history of negation.
Jaszczolt, K.M. (2005). Default semantics: foundations of a compositional theory of
acts of communication. Oxford: Oxford University Press.
Levinson, S.C. (2000). Presumptive meanings. Cambridge, Mass.: The MIT Press.
McElree, B., Nordlie, J. (1999). Literal and figurative interpretations are computed in
equal time, Psychonomic Bulletin & Review, 6, 486-494.
Reboul, A. (2008). Review of Noel Burton-Roberts, Pragmatics, Journal of
Linguistics, 44, 519-524.
Recanati, F. (1989). The pragmatics of what is said, Mind & Language, 4, 295-329.
Recanati, F. (2001). What is said, in Synthese, 128, 75-91.
Recanati, F. (2002). Unarticulated constituents, Linguistics and Philosophy , 25, 299-
345.
Recanati, F. (2004). Literal meaning. Cambridge: Cambridge University Press.
Recanati, F. (2010). Truth-conditional pragmatics. Oxford: Oxford University Press.
Rubio-Fernández, P. (2007). Suppression in metaphor interpretation: differences
between meaning selection and meaning construction, Journal of Semantics,
24, 345-371.
Rubio-Fernández, P. (2008). Concept narrowing: the role of context-independent
information in concepts, Journal of Semantics, 25, 381-409.
Searle, J.R. (1978). Literal meaning, Erkenntnis, 13, 207-224.
Speranza, Join the Grice Club.
Sperber, D., Wilson, D. (1986). Relevance: communication and cognition. Oxford:
Blackwell. (2nd edition with postface: 1995).
Sperber, D., & Wilson, D. (1998). The mapping between the mental and the public
lexicon, in P. Carruthers & J. Boucher (Eds.), Language and thought:
interdisciplinary themes . Cambridge: Cambridge University Press, 184-200.
Travis, C. (1981). The true and the false: the domain of the pragmatic . Amsterdam:
John Benjamins.
Travis, C. (1997). Pragmatics. In B. Hale. & C. Wright (Eds.), A companion to the
philosophy of language . Oxford: Blackwell, 87-107.
Vicente, A., Martínez-Manrique, F. (2010). On relevance theory's atomistic
commitments. In B. Soria & E. Romero (Eds.), Explicit communication: essays
on Robyn Carston’s pragmatics. Basingstoke: Palgrave Macmillan, 42-48.
On Encoded Lexical Meaning: Philosophical and Psychological Perspectives 35
Vicente, B. (2005). Meaning in relevance theory and the semantics/pragmatics
distinction. In S. Coulson & B. Tomaszczyk (Eds.), The literal and non-literal
in language and thought. Frankfurt: Peter Lang, 179-200.
Wedgwood, D. (2007). Shared assumptions: semantic minimalism and relevance
theory, Journal of Linguistics, 43, 647-681.
Wilson, D. (2004). Relevance and lexical pragmatics, UCL Working Papers in
Linguistics, 16, 343-360.
Wilson, D. (2011). The conceptual-procedural distinction: past, present and future.
In V. Escandell-Vidal, M. Leonetti & A. Ahern (Eds.), Procedural meaning:
problems and perspectives. Bingley: Emerald, 3-31.
Wilson, D., Carston, R. (2007). A unitary approach to lexical pragmatics: Relevance,
inference and ad hoc concepts. In N. Burton-Roberts (Ed.), Pragmatics.
Basingstoke: Palgrave Macmillan, 230-259.
Wilson, D., Sperber, D. (1981). On Grice’s theory of conversation. In P. Werth (Ed.),
Conversation and discourse: structure and interpretation. London: Croom
Helm, 155-178