Abstract
Traditionally the noun phrase has been structurally represented as an NP, the maximal projection of N, with the determiner in its specifier position (Jackendoff 1977, among others):
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
Notes
Although the head noun in (3b) is not accompanied by the definite article, it is interpreted as definite (see section 2.1.1).
The phonological alternation between a head noun in the construct state and its free state counterpart is clearly due to the absence of stress on the former. Thus, for instance, an initial vowel /a/ is reduced to a schwa when non-adjacent to stress; a suffix /t/, elided under stress, fails to delete. For phonological discussion of the construct state, see McCarthy (1979).
When a head in the construct state takes a pronominal clitic as its complement, a coreferring noun phrase can surface in a gel (’of’) phrase, resulting in the so-called clitic doubling construction:(i)beyt-oi šel ha-’išihouse-his of the-man’ the man’s house’ Clitic doubling constructions are dealt with in section 2.4.
Regarding deverbal nouns, Rosén (1957) observes that when the complement of the construct state is the only realized complement within that noun phrase, it generally refers to the internal argument; the gel (’of’) phrase, according to him, tends to designate the possessor. Rappaport and Doron (1990) derive this difference from the fact that the relevant deverbal nouns have more readily event reading when they appear in the construct state. As this matter is not relevant for the arguments in this chapter, I will disregard it here. See next chapter for discussion of event versus result nominals.
Besides construct state nominals which are reasonably productive, Hebrew also has compounds. Compounds consist of exactly two nouns which result in idiosyncratic meaning:(i) beyt sefer house book ‘school’ (ii) roš ’ir head city ’mayor’ In some respects they behave like construct states; for instance, the head noun in both cases loses stress and undergoes the same phonological alternations. For discussion of compounds, see Borer (1989); see also Agmon-Fruchtman (1982).
I put aside deverbal nouns expressing result, which have much freer relations with their complements, as well as known peculiarities of event nominals (like the optionality of the subject), because the issue in this section is how nominal 0-grids are mapped onto syntactic structures. For discussion of result versus event nominals, see chapter 3.
The tripartite structure (18a) should a priori be excluded under a strictly binary X’-theory. For reasons of completeness, I show that it must also be dismissed on empirical grounds.
It has been argued that anaphors in noun phrases are used logophorically and are not governed by grammatical conditions (Reinhart and Reuland 1993 ). However, anaphors appearing with event nouns ((19a), (20)), unlike logophoric anaphors, do not give rise to variations in judgements and seem to require a c-commanding antecedent. Anaphors appearing with concrete nouns indeed seem to be logophoric (see appendix to chapter 3). This may suggest that eventhood is relevant to the definition of the domain where Principle A of the binding theory applies (as suggested by Ben-Shalom and Weijler 1991 ).
If the noun phrase contains more structure than NP, one could argue that the prominence of the Agent argument is obtained via covert raising. However, as will become clear in what follows, the Agent receives its Case within the gel (`of’) phrase, and has no reason to raise further (movement being a Last Resort operation; see chapter 1).
A further step would be to generalize and argue that even in languages where D constitutes an independent word, it bears features it must check with the noun, which probably trigger (covert) noun raising to D ( Chomsky 1993 ). Even if definiteness in these languages is not a feature of the noun, but of a free standing article, N may still need to check other features (say *features ) in D.
I abstract away from the possibility that there are additional functional categories between NP and DP that bear other inflectional nominal specifications such as number (Ritter 1991, Valois 1991, Cinque 1992, 1993, Bernstein 1993, among others), as it has no real effect on the analysis defended here. Although the existence of say NumberP would provide a landing site for the raised noun, reasonably Number does not head the noun phrase, nor does it have the determiner in its specifier position. See subsequent section for the claim that noun phrases involving a çonstruct state contain an AgrP that expresses an agreement relation between the head of the construct state and its genitive DP.
A parallelism between the derivation of VSO surface order and that of NSO surface order was first drawn by Ritter (1987) in the context of construct states (discussed here in section 2.3); see also Fassi Fehri (1989). I use IP instead of a more articulated sentential structure when details are irrelevant (as in (25)). In addition, I assume here for the sake of simplicity that the subject is in its base position. It may however be in an intermediate position between SpecVP and SpecAgraP, given a more articulated IP-structure.
This issue is somewhat more complicated in Romance, as both prenominal and postnominal adjectives can occur in the same language. For discussion of the matter, see Cinque (1992, 1993 ), Bernstein (1993), among others.
In fact, as will become clear in what follows, the genitive argument of the construct state surfaces in a position higher than SpecNP.
Government may be defined as follows: a governs XP if a and XP c-command each other; and if a governs XP, then a governs the specifier and the head X of XP (Chomsky 1986a).
In (42b) John is not affected by the head noun and cannot be realized in the specifier position, according to the Affectedness Constraint, formulated by Anderson (1979) as follows (but see also Tenny 1987, Giorgi and Longobardi 1991, Guasti 1992, among others):(i) If a head noun does not express an action which “affects” (i.e.modifies) the state of the object, the latter cannot be possessivized.For our purposes, it only matters that this realization of genitive Case like the realization via of is contingent upon thematic relationships, and therefore not a structural Case.
The examples in (43) are due to Julia Horvath (personal communication), who notes that the deverbal noun seems to form a complex with the predicate, analogously to the corresponding verb. This is not of our concern here.
Alternatively, Agr en may bear weak N-features; in this case, strong N-features of D (see section 2.2.2) will suffice to trigger overt noun raising to D through Agrgen•
Traditionally, the suffix -n is analyzed as an indefinite marker. But see Fassi Fehri (1993) for the claim that it is an instantiation of a functional head, POSS(essor).
Being part of Agr en the [tdefinite] feature presumably need to undergo both Spec-head checking with the genitive DP) and head-head checking in D.
Alternatively, it can be claimed that there is some basic incompatibility between Agrgen features and an intrinsic definiteness feature. The hypothesis would be that this combination of features is morphologically impossible (along lines proposed by Fassi Fehri 1989 and Siloni 1990b, 1991a). This means that a noun in the construct state is inserted underspecified with regard to definiteness. Consequently, an additional mechanism has to be assumed to explain the way the noun acquires a [±definite] value. As its definiteness value is identical to that of the genitive member of the construct, with which it is in an agreement relation, it could be suggested that agreement configurations allow feature endowment between Spec and head in addition to feature checking (the dynamic agreement mechanism utilized by Rizzi 1991a for different reasons).
The examples in (50) are grammatical under the reading that takes sel ha-’ir (`of the city’) to be the complement of ha-cava (`the army’). This is completely irrelevant here.
When the external argument is not phonetically realized, the internal argument cannot receive accusative Case. This impossibility is dealt with in detail in the subsequent chapter (section 3.5).
In section 2.4 I discuss clitic doubling configurations, and in the appendix to chapter 3 concrete nouns, which can involve more than one instance of genitive Case. English noun phrases involving both a prenominal and a postnominal genitive DP (i) constitute a further, salient counterexample:(i) the army’s destruction of the city One possibility that immediately comes to mind would be that the prenominal instance of genitive Case is not assigned by the noun itself (see Abney 1987, Cinque 1994). Another line of reasoning could develop the claim that the prenominal DP is not a real argument of the noun (à la Grimshaw 1990). I will not discuss this type of examples any further here.
Agrgen is too rich to allow PRO, and too poor to license pro (see chapter 5 for more evidence in favor of this claim). When Agr en hosts a pronominal clitic, it can license pro (see section 2.4). Moreover, recall that the head of the construct state loses stress and the main stress falls on the genitive member of the construct, which means that the phonetic realization of the latter may also be obligatory for PF reasons (see also section 2.3.5).
A nominal clitic can be doubled by a sel (`of’) phrase which itself realizes a pronominal clitic; it has though an emphasizing nuance that doubling with a full DP does not have:(I) hofa’at-o1šel-oi appearance-his of-him’his appearance’
Borer (1984) has already proposed that clitics are spell-outs of the Case features of their host. She analyzes Semitic and Romance object clitics alike. There is, however, growing evidence that Romance object clitics form a distinct group. Unlike Semitic clitics, they are morphologically close to determiners, mostly proclitic, and create clusters. In some Romance languages, clitic constructions exhibit a sensitivity to syntactic principles typical of movement. For example, it is known that wh-movement can take place out of a subcategorized PP (i), but not out of an adjunct (ii), as it would violate Huang’s Condition on Extraction Domains (see Huang 1982, Chomsky 1986b, Cinque 1990, among others). Cliticization patterns on a par ((iii), (iv)) (thanks to Luigi Rizzi for the Italian data; for more diagnostics of movement, see Couquaux 1981, van Riemsdijk and Williams 1986, Kayne 1989b, Sportiche 1992, among others)):(i) L’ uomo [a cuit Maria si è messa [pp accanto ec ; ]the man to whom Maria herself was put near (ii) * L’ uomo [a cui]; Maria è felice [pp accanto ec ; the man to whom Maria is happy near (iii) Maria gli si è messa [pp accanto ec ; ]? Maria him ; herself was put near(iv) *Maria gli è felice [PP accanto ec 1 ]? Maria him ; is happy near This has led linguists to argue that at least in some Romance languages cliticization involves movement. For a more thorough comparison of Semitic and Romance clitics, see Shlonsky (1994) and Siloni (1994a). Shlonsky (1994) analyzes Semitic clitics as Agr heads that are incorporated with their host syntactically and do not absorb Case. Shlonsky’s analysis does not predict the complementary distribution of clitics and full noun phrases illustrated in (65).
Belletti (1993) independently proposes that Case Checking can take place in a head-head configuration in Romance.
Author information
Authors and Affiliations
Rights and permissions
Copyright information
© 1997 Springer Science+Business Media Dordrecht
About this chapter
Cite this chapter
Siloni, T. (1997). Noun Raising and Genitival Relations. In: Noun Phrases and Nominalizations. Studies in Natural Language and Linguistic Theory, vol 40. Springer, Dordrecht. https://doi.org/10.1007/978-94-015-8863-8_2
Download citation
DOI: https://doi.org/10.1007/978-94-015-8863-8_2
Publisher Name: Springer, Dordrecht
Print ISBN: 978-90-481-4866-0
Online ISBN: 978-94-015-8863-8
eBook Packages: Springer Book Archive