Content uploaded by Erkki Luuk
Author content
All content in this area was uploaded by Erkki Luuk on Oct 19, 2018
Content may be subject to copyright.
Syntax–Semantics Interface
Erkki Luuk, University of Tartu, Tartu, Estonia; and Stockholm University, Stockholm, Sweden
Ó2015 Elsevier Ltd. All rights reserved.
Abstract
I introduce the topic by pointing out some common problems and disagreements in defining the notions of syntax and
semantics. Avoiding theory-specificity and adhering to no particular linguistic school, the article makes minimal assumptions
about the notions to arrive at a description of syntax–semantics interface. Among the topics discussed are logic, truth,
meaning and their relevance to syntax and semantics, syntactic and semantic validity, and semantics’relations to core areas of
syntax (word order, grammatical relations, and constituency). The main claims are (1) a robust correlation between semantic
and syntactic validity and (2) that grammatical relations and constituency are largely determined by semantics, while certain
forms of word order are autonomous.
Introduction: Syntax/Semantics
The notions of syntax and semantics are widely used beyond
linguistics (e.g., in mathematical logic and computer science, not
to mention their colloquial usage) but the present article focuses
on their application on natural languages (Semantics). In
linguistics, the notions are by no means strictly defined, and
estimates as to what is what and where to draw the line between
the two vary considerably (Jackendoff, 2002: pp. 268–277).
Thus, at this stage, it is useful to think of the domain we are
interested in as syntax/semantics, and to keep in mind
that different theories partition it into syntax and semantics in
different ways. A (or perhaps the) crucial difference between
different schools is whether phrasal and sentential semantics is
allowed. In generative grammar (originating in Chomsky, 1964
[1957]), at least in its recent emanation (Berwick et al., 2013),
only lexical items (basically, words and morphemes) can have
semantics (see Generative Grammar; Lexicon). By contrast,
there are approaches focusing mainly on phrasal and sentential
semantics, e.g., frame and formal semantics and semantic
syntax (originating in Fillmore, 1976;Montague, 1970;Seuren,
1996, respectively). This picture is further complicated by the
fact that syntax and semantics are already very different by their
nature (see Sections Syntax; Semantics; and Syntax–Semantics
Interface: Preliminaries for details). While everybody regards
meaning as central to semantics, views on what is (central to)
syntax are much more variegated. Syntax has at least three
distinct focal parts –word order, grammatical relations, and
constituency –and their roles, contents, and interrelations are
widely debated by different theories (see Sections Constituency;
Grammatical Relations; and Word Order for details; (cf Word
Order; Grammatical Relations)). In some ramified approaches to
sentential semantics, it may be possible to derive grammatical
relations and hierarchical sentence structure (constituency) from
semantics, leaving only the cross-linguistically highly variable
word order for syntax. The gulf between what could be termed as
syntacto- and semantocentric views on syntax/semantics is also
evident in their claims about the universality of syntax or
semantics. For generative grammarians, syntax is universal and
semantics (except for the general principle of form-meaning
relation) highly language-specific, while a semantocentric view
is the direct opposite. In addition, the approaches have generally
opposing research agendas. As referred above, a semantocentric
agenda is largely shaped by questions like how much of
constituency, grammatical relations, and word order can be
derived from semantics and pragmatics. By contrast, in the
history of generative grammar, the role of semantics has gradu-
ally diminished and the function of syntax expanded (cf Berwick
et al., 2013;Chomsky, 1964[1957];Jackendoff, 2002).
Due to innumerable fundamental discrepancies between
different approaches and schools, as well as for some peculiari-
ties of natural language and its study, it seems impossible to give
an objective overview of syntax/semantics, much less of the
syntax–semantics interface. However, if we ever plan to get
anywhere with this we would better start somewhere.
Partitioning Syntax/Semantics
To arrive at a description of syntax–semantics interface, we must
start with a few axioms about syntax and semantics. Given the
state of theorizing in this field, one cannot hope for our
axioms to be both strong and universal (or uncontroversial);
nevertheless, if we want to progress we have to posit some. In
reading indiscriminately what has been written about syntax/
semantics in the last few hundred years, one may observe the
following general principle:
1. Syntax pertains to the ordering of elements, while semantics
deals with their meanings.
Usually, the ‘elements’are meaningful elements –thus, for
example, the ordering of phonemes does not belong to syntax.
However, this is by no means uncontested. According to
Berwick et al. (2013), syntax is also responsible for the
ordering of phonemes. Admittedly, this is something with
which no other approach that I know of concurs. Besides,
when viewed like this, the scope of syntax would be much
wider than that of semantics, which deals only with
meaningful elements by our definition. Thus we are tempted to
set the additional condition:
2. The elements in (1) are meaningful.
From (1) and (2), one can deduce an important dependency:
3. Syntax depends on semantics but the converse does not hold.
900 International Encyclopedia of the Social & Behavioral Sciences, 2nd edition, Volume 23 http://dx.doi.org/10.1016/B978-0-08-097086-8.57035-4
International Encyclopedia of the Social & Behavioral Sciences, Second Edition, 2015, 900–905
Author's personal copy
As a test that the converse does not hold, take any known
word in natural language, and observe that it has meaning
irrespective of its context. This follows from the definition of
elementary word: a minimal unit of language understood (but
not necessary used) outside context (Luuk, 2010). As another
test, consider the sentence:
4. He teached them,
which is morphosyntactically (see Section Syntax) invalid but
semantically valid (i.e., readily understood). We end this
section with an observation which, although not particularly
illuminating or important, is still worth mentioning: Both the
objects of study and the disciplines studying them are called
syntax and semantics, so we should alert ourselves to this
potential difference in these words’meanings.
Syntax
The context, i.e., the ordering meant (and studied) by syntax, is
fundamentally determined by the condition of temporal repre-
sentation over serial channel. In spoken languages this manifests
itself as a sequence of sounds, in written language asthe scanning
sequence in reading, in signed language as a sequence of manual
signs. There is a subtle point as to whether the order of
morphemes (the smallest meaningful units in natural language)
belongs to the domain of syntax. On one hand, many
morphemes are words, but on the other hand, there is an entire
discipline (and level of language) –morphology –generally
viewed as distinct from syntax dealing with them (see
Morphology in Linguistics). In fact, quite a few researchers
would be willing to dismiss syntax and morphology as
independent disciplines altogether and speak of morphosyntax
instead (cf Hengeveld and Mackenzie, 2008). From a logical
viewpoint, as a study of orderings of linguistic signs,
morphosyntax would make perfect sense. (In this article,
linguistic sign and linguistic expression mean the same.)
However, unfortunately ‘morphosyntax’has been used in
another sense already, and for historical (as well as synchronic,
typological, and terminological reasons) syntax has, at least for
the last 100 years, vastly overshadowed morphology. For
a major typological reason, there are languages like Classical
Chinese that admittedly lack mor-
phology (in the sense of lacking morpheme compounding, not
in the sense of lacking morphemes, which would be impossible
for any language). On the other hand, no language could lack
syntax because no language lacks words, the main building
blocks of syntax (note that word compounding is never an
issue, and probably lacking in a vast number of languages).
Thus the reason for this superiority of syntax is an important
definitional asymmetry: syntax is mainly about word ordering,
while morphology is about morpheme compounding and
ordering. This last ‘and’makes morphology’sdomainmuch
more restrictive, essentially equating it with the intersection of
morpheme compounding and morpheme ordering. The
domain of syntax is more elementary, although in the analysis
of levels (of linguistic representation, semantics, etc.) mor-
phemes are more elementary than words.
The vast majority of syntactic rules are language-specific (see
Section Syntax–Semantics Interface: Preliminaries). However,
some cross-linguistically universal synchronic (e.g., (1), (7),
and head-dependent relation –see Section Constituency) and
diachronic principles remain. In general, the syntax of any
language at a particular moment is partly coincidental and
partly determined by the phase of grammaticalization cycle
that the language is currently in (cf Grammaticalization). By
coincidental I mean the result of a complex interplay of
diverse historical factors such as migration, occupation,
technology, war, language contact, language prestige, true
accident, etc. that we will never disentangle (cf Language
Contact).
Semantics
When one considers levels of linguistic representation –
phoneme, morpheme, word, phrase, sentence, and discourse –
semantics pertains uniformly to all of them except phoneme.
Put differently, all these five levels have meaning, thus
semantics by (1). However, the meaning constructed (or
extracted) on one level may in some respect differ from that
on the others. Semantics is uniform in the sense that it has
only one general rule:
5. Form-meaning correspondence.
Note that this view of semantics does not address the defini-
tion of meaning, which must be specified separately (see Section
Predicates, Arguments,and Meaning). The universality of rule (5)
(essentially, a definition of sign) stretches far beyond language, as
(5) describes all natural and artificial communication systems.
Semantics is also uniform in the sense that it is cross-linguistically
universal (as testified by the possibility of translation –cf
Haspelmath, 2007). Succinctly speaking, semantics is
a function that, given a form as input, returns its meaning(s) –
assuming there are any. If there is more than one meaningful
input, the relation will attempt to compose the meanings of its
arguments. For example, as the sentence is a combination of
meaningful units, its meaning is generally compositional (as
usual in language, there are few exceptions, like it rains,whereit
has clearly no meaning whatever –although it as a pronoun has
a meaning). On the other hand, the meaning of a morpheme is
usually elementary (of which there is a plethora of examples,
e.g., the noun stem cat or the noun class (or gender) markings
for feminine and masculine). Several other morpheme types
(e.g., most cases, adpositions, tense-aspect-mood) mark
relations between more than one input, and have thus
compositional meanings involving predicates with two to four
arguments. Four arguments seem to be (close to) the attested
grammatical maximum, probably due to the constraints on
short-term memory (Cowan, 2001).
Predicates, Arguments, and Meaning
Mathematically, predicate is a function from a set A to a special
two-element set. In the context of language (and mathematical
logic), the two elements are usually taken to be true and false
(cf Logic and Linguistics). Assuming that the values are true
and false, and since function is a subcase of relation,
predicates describe the truth or falsity of relations. In the
Syntax–Semantics Interface 901
International Encyclopedia of the Social & Behavioral Sciences, Second Edition, 2015, 900–905
Author's personal copy
simplest case, the relation could be, e.g., belonging to a subset B
of A. If this is the case, the predicate P that represents the relation
is true of an argument x (usually written P(x) or Px) iff x is an
element of B and false (:Px) otherwise. In this case we say
that P is the characteristic (or indicator) function of B. Thus,
predicates are a convenient way of addressing truth, falsity,
sets, relations, functions, and (other) predicates. Given all this,
it might occur that they are also a convenient way of
addressing meaning. In the traditional model-theoretic sense,
the meaning of a sentence is its truth-conditions (Davidson,
1967). In a more refined (but still entirely model-theoretic)
sense, the meaning of a sentence is the proposition expressed
by the sentence as the set of circumstances in which it is true
(Soames, 1992: p. 29). Admittedly, I have a difficult time
distinguishing these definitions, but they are purportedly
different (Soames, 1992), so it makes sense to mention both.
In either case, meaning boils down to a set of circumstances.
Now circumstances are relations, and relations can be
expressed as predicates –which is all the more appropriate
since the circumstances are required to be true (and it is
difficult to get around with truth without predicates). The
frequent argument that only (declarative) sentences can have
meanings is immaterial, because one can take, e.g., cat as
a one-place predicate (which is true if used in reference to the
right type of creature). In a suitable context, even a single word
can produce a reference. Also, one should not be too daunted
by the standard counterargument to this type of theory of
meaning, concerning the truth-conditions of sentences that are
always true (such as it rains or it does not rain or both and
1¼1). It is namely claimed that all such sentences have
exactly the same truth-conditions, whereas in fact there are
ways to filter out the relevant circumstances (but we will not
delve further into this issue here). In sum, we seem to have
amathematicallyprecisewayofaddressingmeaningasa(set
of) predicate(s). Note that this does not tell anything about
how or whether these predicates are implemented in the brain,
so there is no conflict with representation or concept-based
theories of meaning (of which there are many –e.g.,
Gärdenfors, 1998;Jackendoff, 2002;Langacker, 1990;Luuk,
2013a). After all, predicates are somehow implemented as
concepts in our consciousness. If so, why could not they be
implemented on some lower conceptual level? (See Hurford
(2003a,b) for a claim that predicates and arguments can be
distinguished on a perceptual level already.)
Syntax–Semantics Interface: Preliminaries
Both semantics and syntax are rule systems but the rules
involved are very different. As we saw above (see Section
Semantics), there is only one general rule in semantics, holding
far beyond natural language. Differently from animal commu-
nication systems, there is usually no one-to-one correspondence
between linguistic form and meaning (Ferrer-i-Cancho et al.,
2005). This (possibly) unique property of natural language
facilitates expressivity, creativity, ambiguity, and making
mistakes. Syntax, on the other hand, is a relatively complex
6. combinatorial system of dependency, agreement, and linear
ordering rules,
the sole purpose of which appears to be mapping meanings to
a serial channel (by (1)). Differently from (5), (6) and even
the weaker (7) are uniquely human. In addition, the vast
majority of syntactic rules are language-specific, and only
a tiny fraction of them is known, because there are (or
recently were) 6000–7000 languages in the world. The
staggering typological diversity of syntactic rules is a
testament to the degree of freedom the meaning mappings
have. Indeed, it seems as if anything is allowed as far as the
rules strike a (sub)optimal balance between minimum effort
and maximum precision. This is a complex task, with trade-
offs in at least three dimensions: both for the speaker and
hearer individually and between them (producing vs
processing effort). The sole criteria for such a balance seem
to be communicative and propagative success (by the latter
is meant the propagation rate of a syntactic structure in the
community).
In a sense, “mapping meaning to a serial channel”is an
understatement –if a mere mapping would be involved in
syntax, a sequence (a concatenation) of signs would do.
Granted, perhaps not any sequence but, even if we constrain
the sequences, the core of full syntax does more than just
concatenate signs –it also compounds and embeds them. The
difference between concatenation and compounding is that in
the latter the signs are necessarily interpreted together. The
difference between compounding and embedding is that only
the latter allows for center insertion and (potentially) more
than two-level nesting. Note that embedding has a broader
definition (a meaningful unit in another meaningful unit)
here than in the traditional syntactic literature. For example,
a morpheme in a word is an instance of embedding in this
broader sense.
As mentioned above, differently from the principle of
semantics (5), the principle of syntax:
7. a combination of meaningful elements by (morpho-)
syntactic rules,
is uniquely human (Hurford, 2004;Jackendoff, 1999;Ujhelyi,
1998). This claim is not particularly surprising, given that
(morpho)syntactic rule as such is uniquely human (if we
adhere to definitions (1) and (2) –if phonological
sequences are included in syntax, birds may have it).
Thus syntax must have emerged relatively recently by the
evolutionary scale. In fact, we have also some evidence
to base our conjectures on how it evolved. For example,
it can be shown that syntax evolved from concatenation to
compounding to embedding, whilst retaining all the
preceding procedures (Luuk and Luuk, 2014). As virtually all
modern analyses concur, protolanguage started with few
isolated signs (cf Bickerton, 2007;Jackendoff, 1999;
Johansson, 2006). Only after that could certain rules be
applied to them. Despite possible claims to the contrary
(Bickerton, 1998;Chomsky, 2010), it is implausible that
a complex rule system like syntax emerged suddenly full-
fledged as a result of some mutation or otherwise. A
gradualist scenario is much more likely (Pinker and Bloom,
1990). To be precise, the claims (at least that of Chomsky)
are about the abrupt emergence of language faculty, the core
of which would be a capacity for syntax. This is something
completely different from an abrupt emergence of full syntax
902 Syntax–Semantics Interface
International Encyclopedia of the Social & Behavioral Sciences, Second Edition, 2015, 900–905
Author's personal copy
in a linguistic community, which, as said, is implausible (and
probably also impossible).
Syntactic and Semantic Validity
It is frequently claimed that syntax and semantics are two
entirely different systems, domains or dimensions, which have
little (if anything) in common except for the need for a common
interface. This may be so –since (and as long as) the notions are
not strictly defined, not much seems to hinge on the truth or
falsity of this claim anyway. Another frequently advanced
claim is that there is a double dissociation between syntactic
and semantic validity. If this were the case, we would have
rather substantive evidence for syntax and semantics being
autonomous systems of linguistic representation. As the claim
about dissociation seems quite well-articulated, we might be
even able to test it. So let us examine this claim in more detail.
We mentioned above the principle of syntax: a combination
of meaningful elements by (morpho)syntactic rules. It has been
allegedly shown that this function can be separated from
semantics. The commonest example is (Chomsky, 1964[1957]):
8. Colorless green ideas sleep furiously.
This is taken to be an example of a sentence that is syntac-
tically valid but meaningless (that is, presumably, semantically
void). However, the truth-conditions for this sentence (the
circumstances under which it could be true) are
9. If green could be colorless and ideas green, and if sleep could
be furious and ideas sleep (then (8) could be true).
If we accept the standard model-theoretic stance that truth-
conditions are (or yield) the meaning, the meaning of (8) is
somehow given by (9). The circumstance that the relations in
(8) happen to be false in our world is irrelevant. Whether or
not a truth-condition is satisfied in a (or in any) model is
immaterial for its existence. What is required is a structure
where it can be interpreted. Observe that the actual world is
such a structure since it has all the necessary predicates
(coinciding with the words in (8)). This structure allows for
the interpretation of the meaning (i.e., truth-conditions) of the
sentence (8) that is always false in the structure. After all, if it
did not, how would one know that the sentence is false in the
actual world?
In sum, and in light of what we said about the asymmetric
dependency between syntax and semantics (3), one is led to the
view that all syntactically valid expressions are also semantically
valid but not vice versa. As a further test, consider the strings
10. askldfjkldjkl sdfaklj fkkdfj
11. the wug drunted geperfully
12. you runs
13. this is a house
(10–13) exhibit a gradation of general semantic/syntactic
validity (well-formedness, if you will) from zero to maximum.
In each of them, a robust correlation between syntactic and
semantic validity is evident. In (10), both are (near) zero.
Pseudo-expressions like (11) are sometimes erroneously
viewed as examples of syntax–semantics dissociation. In fact,
it is easy to see that everything syntactically valid about (11) is
also meaningful, i.e., valid semantically. Let us try to analyze
this string, starting from syntax. First, there is the noun phrase
(NP) The wug, signaled by the article as well as ‘word’order.
Second, the ‘verb’drunted –technically not an English verb
but readily interpreted as one owing to the past tense suffix
-ed and ‘word’worder. Finally, the ‘adverb’geperfully
modifying (as adverbs do) the ‘verb.’We do not know how
exactly it does that but a modification (as such) is evident.
Like the ‘verb,’the ‘adverb’is signaled by a suffix(-fully)and
‘word’order. Now turn to semantics. It is not difficult to see
that all we said about syntax reflects in the string’s meaning.
In fact, everything in (11) except for the ‘stems’wug,drunt,
and geper is meaningful. We have a definite (the) agent that
was engaged in an activity (in a loose sense) in a specificway
(although we do not know, which) in the past. Admittedly,
this is a lot of meaning (as compared to, e.g., (10)), and the
fact that it was extracted mostly from syntactic rules and
categories does not make it less valid. It merely shows that not
only stems but all morphemes and even specific word orders
(such as the x) have meaning, and that semantics probably
cannot be dissociated from syntax (while the converse is
possible by (3)). (12) is defective morphosyntactically (the
agreement failure) and semantically (2nd and 3rd person
clash), but its syntactic and semantic interpretations are
constrained much better than those of (10–11). In fact, only
three readings suggest themselves –you run,your runs, and (s)
he runs –so as compared to the previous examples the
expression seems more ambiguous than invalid. (13) is valid
syntactically and semantically. In sum, there is a robust
correlation between syntactic and semantic validity in all
examples (cf Kako and Wagner, 2001), corroborating the
claim or suspicion (whichever you prefer at this point) that
syntax is not an autonomous system of linguistic representation.
Constituency
Constituency refers to the whole–part relation in language
expressions. Although one may speak of phonological and
morphological constituents, the term ‘constituent’has
a predominately syntactic usage, being mainly reserved for
words, phrases, and sentences. In the context of syntax,
constituent structure thus refers to words, phrases, or sentences
in a hierarchical structure. There are numerous ways to model
constituent structure of natural language. This (besides
modeling grammatical relations –cf Section Grammatical
Relations) is a bread-and-butter work of theoretical linguists,
and most grammatical theories also stipulate their specific
constituent structures. Likewise, there is no shortage of
methods for describing them (e.g., trees, rewrite rules,
bracketing, predicate logic, frame semantics, lambda calculus,
categorial grammar, model theory, etc. (cf Logic and
Linguistics)). Far from being equivalent or complementary, the
methods frequently lead to very different views on syntax,
semantics, and their interface.
The single most important (i.e., widespread and universal)
morphosyntactic rule is head-dependent relation. The essence of
a head-dependent relation is that in combining two linguistic
units (e.g., in a word or phrase), one of them ‘projects’or starts
governing the other (although it may be not easy to determine
Syntax–Semantics Interface 903
International Encyclopedia of the Social & Behavioral Sciences, Second Edition, 2015, 900–905
Author's personal copy
which one). The one projecting is called the head and the other is
dependent. Since every head-dependent relation specifies two
whole–part relations and three constituents, head-dependent
relation yields constituency for free. Since agreement marks
syntactic constituents, head-dependent relation provides also
the foundation for agreement (cf Grammatical Agreement). In
natural language, head-dependent relation and constituency go
far beyond syntax, possibly even to phonology. An interesting
question for syntax–semantics interface is what determines
which elements become heads at phrase and sentence levels.
At least in some cases semantics has probably some role in it.
For example, the finite verb or flexible (see Luuk, 2010)isthe
head of the clause and the clause predicate, suggesting that the
head may be determined by the semantic predicate-argument
structure. Another example is agreement. Lehmann (1982) has
argued that only constituents that relate to the same referent
may overtly agree with each other.
Grammatical Relations
Grammatical relations (or syntactic functions, as they are
sometimes also called (see Grammatical Relations)) hold
between clause constituents. Standard constituents in this sense
are verb, object, and subject but there are others like modifier,
specifier, determiner, etc. (there are many others, most of them
specific to particular grammatical theories, of which there are
more than 100). Every grammatical relation is stipulated by at
least one head-dependent relation (see Section Constituency). A
crucial question is how much of grammatical relations are
derivable from semantics. As it turns out, a lot. The
Chomskian term for this phenomenon is theta criterion: each
argument of the (finite) verb is assigned exactly one theta role
(semantic role, e.g., agent, patient, goal, etc.) and vice versa.
Because of this one-to-one correspondency, it seems that the
two –semantic role and syntactic argument –are merely two
facets of the same thing, viz. meaning and (‘syntactic’)form,
respectively. But if so, then grammatical relations are
fundamentally semantic phenomena, because form-meaning is
the general principle of semantics. Furthermore, the
supposedly syntactic relations like V, S, O, etc. can be also
described with semantic roles in frame semantics or predicates
and arguments in a suitable predicate logic. A closely related
notion used in generative grammar and elsewhere is
subcategorization frame, identifying finite verb’s potential
positions in phrase structure trees by its class (intransitive,
transitive, ditransitive, ergative, etc. (cf Valency and Argument
Structure in Syntax)). In generative grammar, subcategorization
frame (detached from the verb that appears in it) was believed
to be meaningless. However, Kako (1997) experimentally
showed this not to be the case (with stimuli resembling (11)).
Since subcategorization frames have meanings, it seems that
a core part of syntax is semantically driven.
Word Order
All possible (S, V, O) word orders have been attested in the
world languages, although some are very rare (the World Atlas of
Language Structures,http://wals.info/chapter/81;(see Word
Order)). In languages with (relatively) free word order,
semantics and especially pragmatics (e.g., information structure)
are responsible for at least some ordering principles. In
languages with fixed word order, semantics has probably
a minor role in it. The reason for this is that natural language
semantics is order-neutral –nowhere except in formal notations
is there a rule to the effect that a predicate should follow its
argument(s) or vice versa. If a verb has three arguments (like
give), it probably makes sense (for processing considerations at
least) not to put the indirect object first but this is a pragmatic
rather than semantic constraint. Granted, certain fixed orders
are meaningful as they specify constituents with specific
meanings, be they referents (such as the x mentioned above)
or relations (e.g., subcategorization frames and prepositional
phrases –cf Luuk, 2013b). However, there is no semantic
reason for choosing this particular order –we could have
determiners following nouns and postpositions instead of
prepositions. So, of all the syntactic phenomena we have
discussed, at least fixed word order seems largely independent
from semantics (and other levels of linguistic representation),
thus qualifying syntax as an autonomous level after all (cf
Section Syntactic and Semantic Validity). Although we have so
far concentrated on cases of semantics determining syntax
(partly because they are more difficult to see), this
qualification allows us, at least in languages with fixed word
order, to consider also the converse cases. There are many such
examples, e.g.,
14. j loves m
15. mlovesj
Conclusion
As was noted in the beginning (cf Section Introduction: Syntax/
Semantics), an objective account of the syntax–semantics
interface of natural language seems unattainable. Although I
have tried to shun all major schools and to keep my
assumptions to a minimum, I do not know how successful or
convincing my case has been in this short introduction to
what must be one of the most obscure, complex, and technical
areas of natural language. This obscurity and complexity is due
to several reasons, which for this particular field unhappily
coincide: the great number of opposing linguistic schools; the
still greater number of specific grammatical theories; the direct
relevance of grammar (and hence of grammatical theory) for
a description of syntax–semantics interface; the field’s
uncomfortable position between two established (yet
arbitrarily defined) linguistic levels, hence dependence from
both; the relatively central position with regard to all levels
(and thus a near-maximum dependency from all of them (and
hence also of their descriptions, of which there are many));
and lastly (but most importantly) the scarcity of clear,
universal definitions and the (maybe even necessary) lack of
a theory-neutral research program. One can only hope that
perhaps in a not-so-distant future someone will succeed in
advancing this area beyond mere wishful thinking and
haphazard theorizing. On a more optimistic note, the great
number of linguistic schools and grammatical theories in itself
is rather positive –the more theories we have, the greater
(albeit likely still very low) the probability of least one of
them being (sufficiently) correct.
904 Syntax–Semantics Interface
International Encyclopedia of the Social & Behavioral Sciences, Second Edition, 2015, 900–905
Author's personal copy
See also: Grammatical Agreement; Grammaticalization;
Language Contact; Logic and Linguistics; Morphology in
Linguistics; Valency and Argument Structure in Syntax; Word
Order.
Bibliography
Berwick, R.C., Friederici, A.D., Chomsky, N., Bolhuis, J.J., 2013. Evolution, brain, and the
nature of language. Trends in Cognitive Sciences 17 (2), 89–98.
Bickerton, D., 1998. Catastrophic evolution: the case for a single step from protolan-
guage to full human language. In: Hurford, J.R., Studdert-Kennedy, M., Knight, C.
(Eds.), Approaches to the Evolution of Language: Social and Cognitive Bases.
Cambridge University Press, C ambridge, pp. 341–358.
Bickerton, D., 2007. Language evolution: a brief guide for linguists. Lingua 117 (3),
510–526.
Chomsky, N., 1964[1957]. Syntactic Structures. Mouton, The Hague.
Chomsky, N., 2010. Some simple evo-devo theses: how true might they be for language?
In: Larson, R.K., Yamakido, H., Deprez, V. (Eds.), Evolution of Human Language:
Biolinguistic Perspectives. Cambridge University Press, Cambridge, pp. 45–61.
Cowan, N., 2001. The magical number 4 in short-term memory: a reconsideration of
mental storage capacity. Behavioral and Brain Sciences 24 (1), 87–114 discussion
114–185.
Davidson, D., 1967. Truth and meaning. Synthese 17, 304–323.
Ferrer-i-Cancho, R., Riordan, O., Bollobas, B., 2005. The consequences of Zipf’s law for
syntax and symbolic reference. Proceedings of the Royal Society of London B Bio-
logical Sciences 272 (1562), 561–565.
Fillmore, C.J., 1976. Frame semantics and the nature of language. In: Harnad, S.,
Steklis, H.D., Lancaster, J. (Eds.), Annals of the New York Academy of Sciences:
Conference on the Origin and Development of Language and Speech, pp. 20–32.
Gärdenfors, P., 1998. Some tenets of cognitive semantics. In: Allwood, J., Gärdenfors, P.
(Eds.), Cognitive Semantics: Meaning and Cognition. John Benjamins, Amsterdam,
Philadelph ia, pp. 19–36.
Haspelmath, M., 2007. Pre-established categories don’texist:consequencesfor
language description and typology. Linguistic Typology 11, 119–132.
Hengeveld, K., Mackenzie, J.L., 2008. Functional Discourse Grammar: A Typologically-
Based Theory of Language Structure. Oxford University Press, Oxford.
Hurford, J.R., 2003a. The neural basis of predicate-argument structure. Behavioral and
Brain Sciences 26 (3), 261–283 discussion 283–316.
Hurford, J.R., 2003b. Ventral/dorsal, predicate/argument: the transformation from
perception to meaning. Behavioral and Brain Sciences 26 (3), 301–311.
Hurford, J.R., 2004. Human uniqueness, learned symbols and recursive thought.
European Review 12 (4), 551–565.
Jackendoff, R., 1999. Possible stages in the evolution of the language capacity. Trends in
Cognitive Sciences 3 (7), 272–279.
Jackendoff, R., 2002. Foundations of Language: Brain, Meaning, Grammar, Evolution.
Oxford University Press, Oxford, New York.
Johansson, S., 2006. Working backwards from modern language to proto-grammar. In:
Cangelosi, A., Smith, A.D.M., Smith, K. (Eds.), The Evolution of Language:
Proceedings of the 6th International Conference on the Evolution of Language
(EVOLANG6). World Scientific, Singapore, pp. 160–167.
Kako, E., 1997. Subcategorization semantics and the naturalness of verb-frame
pairings. University of Pennsylvania Working Papers in Linguistics 4 (2),
155–167.
Kako, E., Wagner, L., 2001. The semantics of syntactic structures. Trends in Cognitive
Sciences 5 (3), 1 02–108.
Langacker, R.W., 1990. Concept, Image, and Symbol: The Cognitive Basis of Grammar:
Cognitive Linguistics Research. Mouton de Gruyter, Berlin, New York.
Lehmann, C., 1982. Universal and typological aspects of agreement. In: Seiler, H.,
Stachowiak, F.J. (Eds.), Apprehension: Das sprachliche Erfassen von Gegenständen.
Narr, Tübin gen, pp. 201–267.
Luuk, E., 2010. Nouns, verbs and flexibles: implications for typologies of word classes.
Language Sciences 32 (3), 349–365.
Luuk, E., 2013a. The structure and evolution of symbol. New Ideas in Psychology 31,
87–97.
Luuk, E., 2013b. New methods for analyzing case and adposition meaning. Language
Sciences 37, 136–146.
Luuk, E., Luuk, H., 2014. The evolution of syntax: signs, concatenation and embedding.
Cognitive Systems Research 37, 136–146.
Montague, R., 1970. Universal grammar. Theoria 36, 373–398.
Pinker, S., Bloom, P., 1990. Natural language and natural selection. Behavioral and Brain
Sciences 13 (4), 707–784.
Seuren, P.A.M., 1996. Semantic Syntax. Blackwell, Oxford.
Soames, S., 1992. Truth, meaning, and understanding. Philosophical Studies 65,
17–35.
Ujhelyi, M., 1998. Long call structure in apes as a possible precursor for language.
In: Hurford, J.R., St uddert-Kennedy, M., Knight, C. (Eds.), Approaches to the
Evolution of Language: Social and Cognitive Bases. Cambridge University Press,
Cambridge, UK.
Relevant Website
http://wals.info/.
Syntax–Semantics Interface 905
International Encyclopedia of the Social & Behavioral Sciences, Second Edition, 2015, 900–905
Author's personal copy
