expressions, and those that fill the action position finite verb expressions (1942:72). 'Finite'
here may be a slip on Bloomfield's part: it is not clear that he wished to exclude nonfinite
constructions as in For you to leave now would be unwise from his 'actor-action' type,
though he does say (1942:77) that "infinitives and participles [...] do not serve as centers in
the action position".
2
See Percival (1976) for discussion of Bloomfield's adoption of constituent structure as a
central syntactic notion, under the influence of Wilhelm Wundt (1832-1920).
SYNTAX IN MID-20TH CENTURY AMERICAN LINGUISTICS 409
— not, like the Latin passive, a morphological category — and [...] the
phrase is built or the like is only one of several types of phrases consisting of
a verb plus a past participle (has built, got built). Formally, English is
(loved), just as much as is (lazy), must be classed with builds (a house) or has
(built a house)." In a language such as English, whose passives are marked
by auxiliary verbs, there is a single subject-predicate relation (confusingly
called 'actor-action') in which a subject stands in the same sort of syntactic
relation to an auxiliary verb as to any other verb; by contrast, in languages
such as Latin and Tagalog (on which Bloomfield had worked earlier), the
different morphologically marked 3 'voices' correspond to different predica-
tion relations between the subject and the verb.
3
Unlike Otto Jespersen
(1860-1943) and early generative grammarians, he insisted that "in syntax,
the ultimate constituents of any phrase are words", thus that morphology and
syntax be separate.
Constituent structure played a central role in Bloomfield's thinking about
syntax, as can be seen from his definitions of what are now often called inter-
nal syntax and external syntax: "It is evident that any syntactic construction
has two aspects: on the one hand, the phrases are made up in a certain way;
and on the other, they can fill certain positions in larger phrases" (1942:76).
Bloomfield's term for 'external syntax' was 'function', and it is in that sense
of 'function' that we must interpret his statement that "If a phrase has the
same function as one or more of its immediate constituents, it is an endocen-
tric phrase and has an endocentric construction", which he illustrates by
showing that fresh milk can be substituted for milk in a variety of expres-
sions, from which he infers that fresh milk is an endocentric phrase with milk
as head and fresh as 'attribute'. As Bloomfield defined 'endocentric' and
'head' here, those notions cover much less than did traditional notions of
dependency; for example, preposition phrases do not fit this definition of
endocentric, and the preposition would not be the head of the phrase, as he
stated explicitly in several places. Transitive verb phrases do not fit that
definition of endocentric either, though I have found no passage yet in which
he explicitly classes transitive verb phrases as exocentric. Hockett (1958:191)
followed Bloomfield's preaching in classing transitive verb phrases as exo-
3
Bloomfield does in fact speak of a 'subject-predicate construction', but only in connection
with Chinese (1933:199). I do not know what his rationale was for applying those terms in
Chinese but not in English, nor what putative differences between Chinese and English led
him to say inexactly' when he says of the Chinese construction exemplified by guan men
"close the door" and zài Zhongguo "be in China": "We may call this, somewhat inexactly,
the action-goal construction".
410
JAMES D. McCAWLEYt
centric, while Fries, Harris, and Gleason held looser conceptions of 'endo-
centric' that would take them in.
Bloomfield's discussion of parts of speech is quite insubstantial, and a
surprisingly large part of the four chapters on syntactic topics in his Lan-
guage (chaps. 11-12, 15-16) is devoted to matters at most tangential to syn-
tax, such as contraction and rapid speech. His statements about parts of
speech there and in his chapter in Bloch and Trager provide little to back up
his conclusions; for example, it is not clear that he had any real grounds for
classing pronouns as 'substantives'. The distributional characterization that
he gives of 'pronoun' as a subclass of substantives, namely that they "are
never preceded by modifiers" (1942: 78) contains a fudge, namely 'preceded'
— the class that he wanted to define contained who, what, somebody, some-
thing, all of which can be followed by modifiers, and so if he had said "never
have modifiers" he would have excluded them. (Actually, his distributional
definition would exclude personal pronouns, which can be modified by
words such as only or even : only I, even you.) Why Bloomfield wanted to
recognize a class containing interrogative and indefinite pronouns as well as
personal pronouns is not clear; I conjecture that his concern with the
traditional category 'pronoun' related only to its traditional definitions, not to
its existence as a significant unit of syntax.
4
Charles C. Fries (1887-1967), in his 1952 book The Structure of English,
presented a taxonomy of English words in which he distinguished 19 classes
of words: 4 "parts of speech" in a narrow sense, plus 15 groups of "function
words". He rightly criticized traditional definitions of the various parts of
speech for lacking any consistent classificatory scheme and states that "For
the purposes of adequate classification, the definitions of the various classes
must consider the same kind of criteria" (p.67). (I surmise that he would have
endorsed my comparison of the standard schoolbook definitions with
Borges's Chinese taxonomy of animals, whose humor comes from the fact
that no two of its categories belong to the same classificatory scheme.) Fries
defines each of his word classes in terms of a list of frames, with a word
belonging to the class if it is acceptable in at least one of those frames, as
where he defines his "Class 1" in terms of the list of frames:
5
4
I have argued (McCawley 1992) that indefinite pronouns such as something are NPs in
which the first morpheme is syntactically a determiner and the second morpheme is a N, and
that it is the N part that is the host of modifiers, as in someone tall, nothing that concerns
you.
5
I follow Fries's typographical practices in quoting his frames and examples, in which he
capitalizes the first word but uses no punctuation.
SYNTAX IN MID-20TH CENTURY AMERICAN LINGUISTICS 411
(1) Class 1
Frame A: The was good
The s were good
Frame B: The remembered the
Frame C: The went there
As his definitions progress, his frames become progressively more
schematic, e.g. having defined "class 1", he gives frames for "class 2" in
which " 1 " appears rather than specific words of that class:
(2) Class 2
Frame A: (The) 1 good
Is good
Frame B : (The) 1 (the) 1
Frame C: (The) 1 there
Eventually, he reaches totally schematic representations of frames, as with
the frames that define his Group B of function words:
(3) Group B
A 1 2 3 (e.g., The concert be good.)
A l _ 2 3 4
A l _ 2
Fries notes repeatedly that his classes 1, 2, 3, 4 correspond roughly to the
traditional classes 'noun', 'verb', 'adjective', 'adverb', respectively, but
warns the reader that the correspondence is not exact. For only one of these
four classes does he really make clear how it differs from its traditional
counterpart, noting that many of the words traditionally classed as adverbs do
not belong to his Class 4 but rather to one of his Groups of function words,
e.g., degree words such as very, quite, really comprise his Group D.
6
An
obvious respect in which Class 2 is not the same as 'verb' is that, as he has
defined it, it takes in only verbs that combine with a following adjective or
NP or locative there,
1
i.e., it excludes verbs that combine with a nonfinite VP
6
Fries observes (p.205n.3) that the canonical list of things that an adverb can modify ("a
verb, an adjective, or another adverb") is incomplete, in that the term is commonly applied
also to modifiers of prepositions and "subordinating conjunctions", as in:
In many ways Bob is very much like his brother.
He even swims just like his brother does.
My scare-quotes here reflect my view that (as argued by Jespersen [1924:89]), "subord-
inating conjunctions" are prepositions with sentential objects; in both of these examples, the
modified constituent is a PP.
7
Fries took existential there to be the sole member of his Group H of function words, while
locative there belonged to his Class 4 ("adverbs" in a narrow sense). He treated the word
order with existential there as an instance of inversion, an error that he could have avoided if
he had considered examples in which there are auxiliary verbs in addition to the verb
licensing the there :
412
JAMES D. McCAWLEYt
(many of which meet the definition of Group B), or that combine with a
sentential complement. Fries did not provide any rationale for deciding when
different frames correspond to the same or to different parts of speech, and
thus was not in a position to seriously address the question of whether his
Group B really should be a separate word class from Class 2 rather than a
subclass of it
In traditional schemes of parts of speech, there are two wastebasket cate-
gories: 'adverb' and 'interjection'. It is to Fries's lasting credit that he made a
serious effort to replace these two pseudo-categories with a number of smal-
ler but syntactically more homogeneous class. His separation of 'interjec-
tions' into distinct categories is particularly impressive in view of the wide-
spread failure of both earlier and later scholars to recognize that 'interjec-
tions' can even have an external syntax. He identified the following groups:
Group K, consisting of words that introduce "response utterance units": well,
Dostları ilə paylaş: |