Syntactic Heads and Word Formation (Oxford Studies in Comparative Syntax)
By Marit Julien
* Publisher: Oxford University Press, USA
* Number Of Pages: 416
* Publication Date: 2002-09-26
* ISBN-10 / ASIN: 0195149505
* ISBN-13 / EAN: 9780195149500
Product Description:
Marit Julien investigates the relation between morphology and syntax, or more specifically, the relation between the form of inflected verbs and the position of those verbs. She surveys 530 languages and shows that, with the exception of agreement markers, the positioning of verbal inflectional markers relative to verb stems is compatible with a syntactic approach to morphology.
Introduction: An Overview of the Work
The topic of this book is the formation of morphologically complex words. That
is, it deals with the mechanisms of grammar that may cause two or more of the
minimal meaningful elements of language—the morphemes—to be combined
into one single word. The claim that will be developed in much more detail in
the following chapters is that word formation is mainly a matter of syntax. That
is, the basic building blocks of syntax are individual morphemes, not words,
and it is the syntax that determines the order of morphemes within each complex
word, in very much the same way as it determines the order of words in phrases
and sentences.
But notably, I also argue that the notion of 'word' itself cannot be defined in
syntactic terms. Rather, it appears that the only criteria that can be used to detect
the words in any string of speech are distributional ones. Thus, for speakers as
well as for linguists, a word is a morpheme sequence that shows cohesion internally
and has independent distribution externally.
If words are characterized by their distributional properties, it follows that it
is not necessarily correct to associate the word with one particular type of syntactic
constituent, such as X°. A necessary condition for two morphemes to form a
word is that they are linearly adjacent and that this adjacency is a recurrent
pattern in the language in question. My claim is that this is also a sufficient
condition: when two morphemes regularly appear immediately adjacent to each
other, the two morphemes will tend to be seen as one grammatical word.
Now if a morpheme X precedes a morpheme Y and there is a word XY, there
are at least three different syntactic relations that may obtain between the syntactic
nodes X and Y. The two nodes may be contained in a complex syntactic
head, or X may be the final element of a constituent in Spec-YP, or Y may be
the head of the complement of XP, provided that there is no phonologically
realized material in the specifier position of that complement. But if morphemes
may combine into words in any of these three configurations, it follows that
grammar does not have at its disposal any operations that specifically form
words. On this view, words are perceived rather than formed. That is, the word
in the nonphonological sense is an epiphenomenon that really has no status in
grammar.
In this work, the predictions concerning morpheme ordering that follow from
the approach just sketched are tested against the patterns of verbal inflection
found in a broad sample of languages drawn from all over the world. The conclusion
is that the predictions are borne out to a considerable degree. Even if this
does not prove that the syntactic approach is correct, it certainly suggests that
the syntactic approach is the more adequate one, since it requires fewer auxiliary
assumptions than competing lexical theories.
The study is greatly inspired by the Distributed Morphology approach of
Halle and Marantz (1993, 1994). The theory that I develop, however, makes use
of a smaller set of word-forming tools than Distributed Morphology does. First,
I do not recognize the operation referred to by Halle and Marantz as morphological
merger, which is the operation whereby structurally adjacent heads can be
joined under a zero level node even in the absence of head movement; second, I
do not allow the actual order of morphemes to be attributed to the properties of
individual affixes. Instead, in my theory the surface order of morphemes is a
direct consequence of syntax.
The work is organized as follows. Chapter 1 is a general introduction to the
field of inquiry that I will be concerned with and to the theoretical framework
that I am going to assume. I begin with a brief recapitulation of the history of
lexicalist and nonlexicalist approaches to word formation within the generative
tradition. After that I give an outline of the model of grammar on which the
study is based. This model is characterized by the assumptions that syntactic
structures are built from abstract morphemes and that the phonological shape of
these morphemes is only determined at Spell-Out—that is, at a stage where the
base-generated morpheme order, and even the number of morphemes, may have
been altered by subsequent syntactic operations.
I then go on to deal with various definitions of the concept 'word'. The
conclusion of this discussion is that 'word' in a nonphonological sense is a label
for morpheme strings that share certain distributional properties. Apart from
this, the concept 'word' is of very little relevance to grammar.
In the final section of chapter 1 I present some results from a survey I have
conducted of verbal morphology and word order in a sample of 530 languages,
representing 280 genera from all over the world. It appears, for example, that if
the tense markers of natural languages are classified according to their realization
relative to the verb stem—as prefix, suffix, free proposed marker, or free
postposed marker—one sees that these types are not evenly distributed across
languages. Not only are certain types of tense markers generally preferred over
others but also there are observable correlations between certain tense marker
types and certain word orders within the clause. It is these interdependencies that
are explored in more depth in the remainder of the work.
The topic of chapter 2 is head-to-head movement, a process which is widely
assumed to be a word-forming one. However, there are still quite a few questions
pertaining to head movement that have not been conclusively answered.
The first problem that I take up in chapter 2 is the trigger for head movement. In
the checking theory of Chomsky (1993, 1995), where words come fully inflected
from the lexicon, the only trigger for movement is the need to check the features
of the moving element and of the attractor. In theories that deal more explicitly
with word formation, be it in syntax or in a syntax-like component of the lexicon,
the concept of morphological subcategorization has played a central part.
Thus, it is often assumed that if a lexical root adjoins to an inflectional marker,
it is because the inflectional marker subcategorizes morphologically for an
element of that particular lexical class. Conversely, in the apparently exceptional
cases where root and inflection marker do not combine morphologically,
although one would expect them to do so, many theorists have proposed that it
is because the root does not meet the morphological subcategorization require
ments of the inflectional marker and that these requirements may then trigger the
insertion of auxiliaries in such constructions.
I propose instead that head movement is triggered by a strong head feature in
the attracting head. That is, if a head has a strong head feature it will attract the
next head down. However, the derivation will crash if the complement feature of
the attractor is not checked. Hence, the head movement operation is only successful
if the head of the complement satisfies both the head feature and the
complement feature of the attracting head. That is, for any given X°, its
complement feature will ensure that the complement of X° is of the right
category. Now if X° always has a complement of the category YP, and the head
Y° of that complement is always attracted to X°, the consequence is that X° in
all its occurrences forms a complex head with a Y°. Hence, X° will have the
appearance of an affix. In this way, the concept of morphological subcategorization
is reformulated in syntactic terms.
While the concept of feature checking is thus retained, my proposal differs
from that of Chomsky (1993, 1995) in that I argue that words do not come fully
inflected from the lexicon but, instead, morphemes are inserted separately into
syntactic structures, so that syntactic movement may have morphological consequences.
The next topic of chapter 2 is excorporation. The question of excorporation is
crucial to a theory like the one I am developing here, since if excorporation is
possible—if a head may move through another head—two heads need not be
adjacent at Spell-Out even if they constitute a complex head at an earlier stage of
the derivation. Then the connection between head movement and word formation
becomes rather loose. After taking both theoretical and empirical arguments into
consideration, I conclude that excorporation is not possible and that the morpheme
order inside a word formed by head movement is tightly conditioned by
the underlying sequence of heads.
I then go on to deal with 'morphological merger', an operation which has
been proposed in order to account for certain cases of word formation where head
movement is claimed to be absent. My conclusion is that it would be better to
refrain from assuming this operation. After that, chapter 2 is rounded off with a
discussion of the word properties of complex syntactic heads.
In chapter 3 I turn to a discussion of head-final languages. The analysis I
propose of the syntax of these languages is based on the suggestion in Kayne
(1994) that head-final order combined with suffixing and agglutinating morphology
might be the result of successive movements of complements to specifier
positions or, more precisely, of the complement of every head H in IP to the
Spec of HP. In my analysis, this movement is triggered by the complementselectional
features of the complement-taking heads in IP, which take effect in
the absence of strong head features and strong features triggering argument
raising.
In addition, I show that head-final languages tend to make extensive use of
movement to the CP-domain. This latter type of movement is what lies behind
the dislocation phenomena traditionally referred to as 'scrambling'. My claim is
that movement to the CP-domain, which can have profound consequences for
the ordering of clausal constituents, is triggered by discourse-related features.
Chapter 4 deals with prefixed verbal inflectional markers. I argue that since
neither the head movement analysis nor the movement-to-Spec analysis can be
extended to these markers, it must be the case that when an inflectional marker
is prefixed to a verb root, the inflectional marker and the verb root represent
syntactic heads that have not been moved with respect to each other. That is, the
prefix is simply the spellout of a head that is in a higher position than the root
it combines with. Such an element can be seen as part of the verbal word if it
regularly appears in front of the verb and if it cannot be separated from the verb
by phrasal constituents. We then have a complex word which is not the outcome
of any particular operation but just the consequence of the distributional properties
of the morpheme string that makes up the word.
While chapters 2 through 4 are thus concerned with the question of how individual
inflectional markers come to precede or follow the verb, chapter 5 is an
attempt at giving an overview of the distribution within clauses of the verb root
and of verbal inflectional markers. The positioning of verbal markers in different
word orders is discussed along with the relative ordering of various inflectional
markers—in particular tense and aspect markers. It appears that the patterns that
we find can all be derived syntactically from a uniform base order and that data
which have been pointed to as counterevidence of the syntactic approach to word
formation also allow analyses that are not in conflict with the view that word
formation is syntactic.
However, while the syntactic approach to word formation implies that individual
morphemes are associated with separate syntactic heads, a brief examination
of agreement markers reveals that agreement markers must be analyzed
differently from other verbal inflectional markers. There is much more variation
to be found cross-linguistically in the distribution of agreement markers than
what I have shown to be the case with markers of other categories. Together with
the insight that agreement markers are also different from other inflectional
markers in that they have no independent content, this leads to the conclusion
that agreement markers do not in themselves represent syntactic heads but,
instead, agreement markers spell out features that are added to heads whose basic
content is something else.
In the last main chapter, chapter 6,1 raise the question of whether grammar
has a separate morphology module. If complex words are formed in syntax, one
might think that there is nothing left for such a module to do. However, there
are cases where the relation between the morphemes in the surface order and the
underlying sequence of syntactic heads is not as straightforward as the
discussion in the preceding chapters has suggested. The relevant phenomena,
which might be taken to be the workings of morphology, are discontinuous
marking, fused marking, allomorphy, and syncretism.
Concerning discontinuous marking, I argue that it does not really exist; what
we have instead is two markers that combine to give a particular meaning. As
for fused marking, it is the process whereby two sister nodes coalesce, leaving
only one terminal that will be spelled out by one single lexical item. This can
be seen as an essentially syntactic operation, which is ultimately contingent on
the contents of the lexicon......................... |
楼主: 帆帆
ictionary of Language and Linguistics. However, particularly in the field of phonetics and phonology the book is an undefeatable source of information.