A Lexical Functional Account
4. Previous syntactic analyses
4.1 Unified Analyses
Again, details aside, an account that assigns C/M the same direction of branching is considered a unified account here. Whether it is unified left- or right-branching, each has its advantages and disadvantages.
Unified left-branching
The unified left-branching account, where the numeral and C/M form an exclusive constituent, as shown schematically in (31), enjoys a long history, one that has the early support by Greenberg in his seminal work (1990b).
There are many indications that in the tripartite construction consisting of quantifier (Q), classifier (Cl), and head noun (N), Q is in direct construction with Cl and this complex construction, which will be called the classifier phrase, is in turn in construction with N. (Greenberg 1990b:227, emphasis added) (31) Unified left-branching structure
3
Num C/M N
san ben/xiang shu
three C/M-box book
Over the decades, this account has been widely accepted, e.g. Li & Thompson (1981:105), Paris (1981:105-117), Huang (1982), Tang (1990), Croft (1994:151), Lin (1997:419), and Hsieh (2008). In his seminal work on Chinese phrase structure, Huang (1982) posits (32) as a unified analysis for C/M. Fukui & Takano (2000) also propose a similar structure for Japanese C/M, shown in (33).
One-Soon Her
(32) Huang’s (1982) unified left-branching account for Chinese NP
ClP N′
2 !
DetP Cl′ N
! 2
na QP Cl′
‘that’ ! !
yi ben shu
‘one’ CL ‘book’
(33) Fukui & Takano’s (2000) unified left-branching account for Japanese NP
3
#P N′
2 san # ‘three’ satsu hon
CL ‘book’
The most recent endorsement for the unified left-branching account is from Hsieh (2008), in her book dedicated to the study of the Chinese NP. In (34), an example is given to illustrate Hsieh’s (2008) unified account.
(34) 那 三 本/箱 書 na san ben/xiang shu
that three C/M-box book
NP
#P N′
DemP/QP #′
! 3
na NumeralP/QP/AP #′ N
! ! !
san # shu
! ben/xiang
Structure of Classifiers and Measure Words
1229
All unified accounts enjoy the obvious advantage that all the common properties documented in detail in §3 that C/M share only need to be stated once. Yet, a unified left-branching account, in particular, has the advantage over the right-branching counter-part in that it captures the parallel mathematical structure, i.e. [n × base], between the constituent of a high number round figure Num and the constituent of [Num C/M].
However, as Zhang (2011) aptly points out, this account generalizes that all pre-C/M modifiers and Num scope over C/M only, not N, as shown in (35).
(35) Unified left-branching structure
a. 重重的 三 大 本 厚 書
zhongzhongde san da ben hou shu heavy three big C thick book
‘three heavy big thick books’
b. 重重的 三 大 箱 厚 書
zhongzhongde san da xiang hou shu heavy three big M-box thick book
‘three heavy big boxes of thick books’
NP
Mod Num Mod C/M Mod N
zhongzhongde san da ben/xiang hou shu
heavy three big C/M-box thick book
As seen in §2.2, pre-C adjectives and Num must scope over C as well as N. Yet, in (35), the modifiers heavy and big do not c-command N and thus cannot scope over N.
Likewise, Num also does not c-command NP and thus also does not scope over N.
Unified right-branching
On the flip side, many of the more recent formalist studies on this subject favor a unified right-branching analysis, as shown schematically in (36), e.g. Tang (1990:413, 2005), Cheng & Sybesma (1998, 1999), Borer (2005), Watanabe (2006), Zhang (2009), among others.
One-Soon Her
(36) Unified right-branching structure
Num C/M N
san ben/xiang shu
three C/M-box book
Cheng & Sybesma (1998, 1999), for example, propose that C is base-generated as the head of ClP, as in (37a), while M, as shown in (37b), only moves to C from its original lower position. Movement put aside, both C/M form a constituent with N first, excluding Num.
(37) a. 十 枝 筆 shi zhi bi
ten C pen
ClP 3
shi Cl′
‘ten’ 3
Cl NP
zhi bi
CL ‘pen’
b. 三 碗 湯
san wan tang
three M-bowl soup
ClP 3
san Cl′
‘three’ 3
Cl NP
wani 3
‘CL’ N NP/ClP
ti ! tang
‘soup’
Structure of Classifiers and Measure Words
1231
Zhang (2009), on the other hand, contends that C is base-generated as the head of a SortP, with a preceding NP as its complement, and moves up to Q, where M is base-generated. This account is shown in (38). Again, details aside, both C/M form a con-stituent with N first, excluding Num.
(38) DP 3
D QP (for counting) 3
three Q′
3
Q SortP (for sorting & dividing) 3
numeral NP Sort
4 !
counting unit Clsort [DIV]
Tang (2005), in contrast, has both C and M straightforwardly base-generated as the head of ClP, or classifier phrase, with no movement involved.
(39)
D
Num 3
C/M NP
na san ben/xiang shu
those three C/M-box book
Borer (2005) proposes that Chinese C parallels English plurality marker -s and functions as a mass divider that affords the noun a count interpretation in the structure in (40a). M, on the other hand, parallels English measure phrases and has the structure in (40b), possibly as the head of # max.
One-Soon Her
(40) a. (DP
D) 3#max
three23#
3
<e2># CLmax
b. (DP
D) henduo2
much
<e2># N
qian
money In spite of Borer’s (2005) claim that lexically all nouns in all languages are mass by default and the two different structures assigned to C/M, C/M behave the same in forming a constituent with N first, excluding Num. Like the unified left-branching accounts, these unified right-branching accounts have the advantage that all the common properties documented in detail in §3 that C/M share only need to be stated once.
However, like the unified left-branching account, the unified right-branching account also generalizes the scope of a pre-C/M adjective and the quantification scope of Num (Zhang 2011).
(41) Unified right-branching structure
a. 重重的 三 大 本 厚 書
zhongzhongde san da ben hou shu heavy three big C thick book
‘three heavy big thick books’
b. 重重的 三 大 箱 厚 書
zhongzhongde san da xiang hou shu heavy three big M-box thick book
‘three heavy big boxes of thick books’
ge3
‘piece’
<div>3
<e3>DIV Nmax ren
‘person’
cat
#max
#
Structure of Classifiers and Measure Words
1233
NP
6 5
Mod Num Mod C/M Mod N
zhongzhongde san da ben/xiang hou shu
heavy three big C/M-box thick book
The structure in (41) shows that all pre-C/M modifiers and Num c-command C/M as well as N and thus scope over C/M as well as N, if C/M are assumed to be functional.
Yet, if C/M are assumed to be lexical, then pre-C/M modifiers and Num only scope over C/M, but not N. This is a dilemma given that, as seen in §2.2, M numerals and pre-M adjectives scope over pre-M, but not N.
Zhang (2011) thus also rejects the right-branching account, which treats C/M uni-formly. Yet, a possible solution to this dilemma is found in Vos (1999) and van Riemsdijk (1998), where it is assumed that M is lexical and thus opaque blocking modification and quantification to scope over N, while C is semi-lexical, thus more functional than purely lexical items such as M, and is therefore transparent. This is similar to Kubo’s (1996) proposal to distinguish syntactic, semantic, and cognitive elements in grammar, and to view M as semantic and C as cognitive. Her & Hsieh (2010) likewise point out that C is a closed set and M an open set in Chinese, corresponding to a functional category and a lexical category respectively. Note that in the left-branching structure of (35), C, being semi-lexical and thus transparent, still does not allow the non-c-commanding adjectives and Num to scope over N. The right-branching account, therefore, fares better than the left-branching counterpart in relation to C/M differences in scope phenomena. However, in §5.1 we shall see that the left-branching account fares better in capturing C/M’s common behaviors and that there are also other good reasons from word order typology to reject the right-branching account. An ultimate LFG solution is then offered in §5.2, which also takes advantage of the insight that C is less lexical and more functional than M.