Numeral classifiers in Chinese: The syntax-semantics interface by XuPing Li (review)

This book, which is a revised and extended version of the author’s doctoral thesis, represents a state-of-the-art work, a milestone, on the syntax and semantics of Chinese-style classifiers. The aim of the book (see next paragraph) is clearly stated and broken down into incremental steps with well-formulated research questions and a rigorous set-up of empirical evidence. The author incorporates valuable insights from publications in Chinese not easily accessible to a worldwide audience. The integrated syntactic/semantic analysis is a departure from the prevalent cognitive/ontological paradigm characteristic of most works in the past fifty years. There are, however, certain inconsistencies in the semantic analysis that I highlight below.

The author argues against a grammatical basis for dividing Chinese nouns or Chinese classifiers into subcategories at the word level (e.g. count versus mass nouns, or sortal versus mensural classifiers). Instead, he proposes a division of classifiers at the phrasal level, into count classifier phrases versus measure classifier phrases. This underlying idea is reminiscent of the (in)alienable property that, similarly, is not a feature of words but of phrases. (In)alienability is the (im)possibility of separating the possessee from the possessor (Chappell & McGregor 1996).¹

The author reviews empirical evidence for distinguishing between count and mass nouns, and between sortal and mensural classifiers. There is a ‘signature property’ (Chierchia 2010) that divides nouns into those that can be directly modified by numerals (count nouns) and those that cannot (mass nouns). Since all Chinese nouns carry the signature of mass nouns, the distinction of count versus mass nouns cannot be made at a grammatical level, only at an ontological level. Bare nouns, which the author investigates in a separate chapter, are underspecified between a kind-level reading and a (count/mass) object-level reading (Krifka 1995). The author deploys two tests that demonstrate this ambiguity. First, bare nouns can take kind-level predicates like juezhong ‘extinct’ or become kind-level predicates after the copula. Second, bare nouns that are objects of opaque verbs like zhao ‘seek’ are ambiguous between a kind-level reading and a definite object-level reading.

Moreover, the author discounts any evidence that would allow drawing a grammatical distinction between sortal and mensural classifiers that in turn could be used to differentiate between count and mass nouns. He argues against three pieces of evidence that Cheng and Sybesma (1998) advanced in support of a distinction between sortal and mensural classifiers. First, it is not always the case that sortal classifiers are denominal morphemes and mensural classifiers nominal morphemes. Some sortal classifiers, for example, are grammaticalized from verbs (gua ‘hang’) and adjectives (wan ‘curved’), while some mensural classifiers cannot be used as nouns that take classifiers. Second, not all sortal classifiers disallow modification by the adjectives da ‘big’/xiao ‘small’. (Mensural classifiers, by contrast, can take size adjectives.) The author quotes Lu (1987), who identifies twenty-four sortal classifiers with possible adjectival modification. Third, there are many counterexamples to Cheng and Sybesma’s claim that sortal classifiers cannot be followed by the nominalizer de, while mensural classifiers can. A summary of these counterexamples is presented in Table 1.

Table 1.

Sortal classifiers and counterexamples.

In support of the author’s argumentation, we can further cite the fact that certain sortal classifiers can categorize both count and mass nouns. The classifier tiao, for example, categorizes lengthy count objects and also mass objects such as precious metals.

1. (1).

   1. a.

   2. b.

The main proposal of the book is a syntactic dichotomy of Chinese count and measure classifier phrases. Many scholars observed that container classifier phrases like three bottles of water are ambiguous between a count reading and a measure reading (e.g. Selkirk 1977).

1. (2).

   1. a. John carried [three bottles of water] home.          (count)

   2. b. I poured [three bottles of water] into the soup.          (measure)

On the count...