The Syntax of Numeral Noun Constructions

| 23 →

Chapter Two: The Genitive of Quantification: a history of solutions

2.1. Introduction

In the vast literature on constructions with numerals ≥ 5 in Slavic languages, we encounter various approaches attempting to explain peculiar properties of Genitive of Quantification. The case pattern in numerically quantified phrases and the position of modifiers within these phrases have given rise to a number of proposals which, by means of generative procedures, try to derive properties of phrases with higher cardinal numerals. However, the major problem for establishing the structure for these constructions is the fact that depending on the syntactic context they have different properties. In structural case positions, i.e. in positions to which nominative or accusative are assigned, the quantified noun is marked for genitive, whereas in oblique case positions, i.e. in positions in which lexical cases are distributed, both the numeral and the noun occur in case dictated by the verb or preposition, e.g.

The examples in (2.1) and (2.2) clearly show that in structural contexts the Numeral Noun Construction resembles a typical binominal structure with the adnominal genitive, while upon the assignment/distribution of the lexical case, the numeral and the noun become case-congruent. What is even more striking, with lower numerals (i.e. 1–4) the numeral presents typical adjectival properties, e.g. (2.3a-c), i.e. it agrees in number, gender and case with the modified noun. Also, the whole phrase in the subject position triggers agreement with the verb, e.g. (2.3a). ← 23 | 24 →

Additionally, in other Slavic languages, we witness interesting variations in case distribution which make it even more difficult to introduce one uniform account of Genitive of Quantification.1 As a result, over the span of last decades, there have been various accounts introduced to adequately address the structure of GoQ. Examining these analyses, it can be noticed that one of the major puzzling aspects of these constructions is determining their structure which would accommodate their distinctive properties depending on the syntactic context. The first glance at various approaches immediately reveals that discussions are built around the issue of headedness, so the most prominent element of the phrase, which also affects case relations between constituents of numerically quantified phrases. Consequently, in the following subsections we focus on a brief review of selected analyses of phrases with numerals to present the spectrum of approaches and problems these structures may generate.²

2.2. Noun as the head in phrases with higher numerals

One of the most renown accounts of Genitive of Quantification in the tradition of the transformational grammar is discussed by Babby (1987). His analysis, limited to the Russian data, is based on the mechanism of case assignment to the maximal projection and further on the percolation of a case feature down within the phrase. The nominal phrase, i.e. NP, being in a structural position receives nominative or accusative which spreads to other elements within the NP, i.e. the head and its modifiers. When, the numeral becomes introduced into the structure, the QP ← 24 | 25 → becomes the source of genitive assigned to the head noun. Although in structural contexts, it is nominative or accusative case that are assigned to the maximal projection of N, the QP modifier assigns genitive to the intermediate projection of N which later percolates down to the head noun, which is the head of the numerically quantified phrase.3 Such a step becomes possible if the assignment of the abstract case is determined by NP’s structural relations supported with the case hierarchy as proposed by Babby (1987), i.e. in the case of conflict, genitive takes precedence over nominative and accusative. Also, case conflict resolution takes place in oblique contexts, in which case, lexical cases take precedence over configurational ones, hence homogenous syntax.⁴ Some further support for the process of case assignment, i.e. the fact that it is not controlled by the head but it spreads from the maximal projection, comes from the so-called discontinuous agreement, e.g. (2.4)

Usually modifiers preceding the numeral in Russian occur in the case congruent with the numeral, i.e. nominative or accusative, as they are not in the c-command domain of the QP; they do not belong to the case percolation path. Some prequantifiers, however, can be marked for genitive as in example (2.4), which is accounted for by means of their position in the structure. Babby (1987) proposes that such modifiers are attached to one of the intermediate projections of N but lower than accusative or nominative modifiers, and this way they are in the c-command domain of the quantifier. Hence, they are assigned genitive.⁵ Although Babby’s account comprehensively discusses GoQ, it is limited only to Russian data and to higher numerals only. Considering that in Slavic languages we observe some variations in case patterns in GoQ, including modification by numerals 1–4, the analysis might require some modifications. ← 25 | 26 →

A more recent analysis of Genitive of Quantification is introduced by Rappaport (2002, 2003), who describes the process of case distribution in phrases with numerals on the basis of Chomsky’s operation Agree. Rappaport (2002), also for Russian and similarly to Babby (1987), proposes that the head of numerically quantified phrases is the noun and not the numeral, but case assignment and distribution of GoQ is accounted for in the minimalist spirit. Both lexical and functional elements enter the derivation either with valued or unvalued case features and case valuation is reached by means of Agree (cf. Chomsky 2000, 2001). In what follows, we derive the heterogeneous case patterns when the numeral has a valued case feature and the noun an unvalued one. The numeral then provides value for case to the noun and the noun provides value for gender to the numeral. As a result, the noun is marked for genitive. What is interesting, however, in this account is that numerals are not claimed to enter a derivation with genitive case later copied onto the noun, but they have a fixed value for case called quantitative which is syncretic with genitive (Rappaport 2002: 334f.). In homogenous syntax, when the numeral and the noun have the same case value, the noun is selected from the lexicon with a valued case feature and the numeral with the unvalued one. This time, the value for case, determined by the external selector, i.e. a verb or a preposition, is copied from the noun to the numeral via Agree. Although at first sight the mechanism based on Agree seems to be straightforward, the operation of feature valuation/copying becomes a little more complex when considering Polish data, specifically considering forms of numerals in different genders. The first problem, indicated by Rappaport (2003), is the form of a higher numeral in direct case positions in feminine, neuter and masculine ±animate e.g. (2.5a), versus masculine virile, e.g. (2.5a) where we have the following patterns:

The form of a masculine numeral is genitive and is syncretic in nominative, accusative and genitive. Considering that the noun is marked for genitive, one may assume that we deal here with a homogenous pattern, which is not the case, as according to Rappaport quantitive case of numerals is spelled out differently for non-virile numerals, i.e. feminine, neuter and masculine animate/inanimate, on the one hand and for masculine virile numerals on the other. Some rules in the morphological component specify that non-virile numerals are expressed as ← 26 | 27 → syncretic with accusative case, with ø ending, and masculine virile numerals are syncretic with genitive. The same morphological rules are responsible for the form of lower masculine virile numerals which present the syntax of higher numerals, i.e. they select for a noun in genitive having themselves a form syncretic in nominative, accusative and genitive contexts, e.g.:

Lower numerals in feminine, neuter and masculine animate/inanimate exhibit agreement in case with a modified noun. These case patterns for lower numerals in direct case positions are explained by postulating that non-virile numerals are spelled out as accusative contrary to masculine virile numerals associated with genitive. Case values proposed by Rappaport are based on the assumption that numerals do not have nominative case, which remains consistent with the Accusative Hypothesis. Summarizing Rappaport’s account, in structural contexts the numeral becomes the source of the case value for the noun, and in oblique contexts the noun provides the case value to the numeral. The case copying/valuation is enhanced with some morphological rules which determine the morphological form of numerals.

2.3. The numeral as the key element in Numeral Noun Constructions

Within the numerous accounts of Genitive of Quantification we encounter analyses adopting structures in which the noun is no longer the major element of the phrase. Such an approach to GoQ has been discussed by Przepiórkowski (1999) and Bailyn (2004) who demonstrate, although in different frameworks, that the projection of the numeral plays a crucial role in numerically quantified phrases.

In Przepiórkowski’s (1999) analysis conducted within the Head-Driven Phrase Structure Grammar, the numeral becomes a designated head of the phrase with the noun as its subject. It is argued, on the basis of examples with ellipsis of the noun from the numerically quantified phrase, that the numeral as the distributional representative constitutes the major element in the phrase, e.g. based on Przepiórkowski’s examples (1999: 178):