Chapter 4 The ordering of modifiers with de
4.1.1 Two-layer analysis
Bolinger (1967) presents a contrast between prenominal and postnominal adjectives. Prenominal adjectives show inherent and characteristic features (i-level
21 There has been a debate on the restrictive versus appositive distinction of pre-DNC or post-DNC relative clauses (Lin 2003, Del Gobbo 2005, Del Gobbo 2010, Lin &
reading) while postnominal adjectives show temporary features (s-level reading), as shown in (3a) and (3b) respectively. In addition, the prenominal position can also take s-level readings, but the adjective closer to the noun should be interpreted as i-level while the farther one as s-level, as in (4). An inherently visible star can be temporarily nonvisible while an inherently nonvisible star is never visible.
(3) a. the visible stars (i-level) b. the stars visible (s-level) (4) a. the nonvisible visible stars
b. #the visible nonvisible stars
Based on the contrast, a two-layer analysis is proposed (Del Gobbo 2005, Larson
& Takahashi 2007, Cinque 2010). It is proposed that there are two layers of modification: one is DP modifier while the other is NP-modifier. As is illustrated in (5), NP-modifier (α) is inside the NP and can only occur prenominally while DP-modifier (β) is outside the NP and can occur either before the NP-DP-modifier22 or postnominally. The i-level reading is mapped to NP-modification while the s-level reading is mapped to DP-modification.
(5) [DP D β [NP Γ e [ α N] β] (α=NP-modifier; β=DP-modifier)
22 When a single adjective occurs prenominally, it is therefore ambiguous between the inherent or temporary readings.
Several studies follow the two-layer approach to analyze the Mandarin data (Del Gobbo 2005, Larson & Takahashi 2007, Hsieh 2005, Zhang 2015). It is assumed that two-layer modification follows from the fact that modifiers in Mandarin can precede a demonstrative. Therefore, the β in the postnominal position corresponds to the pre-demonstrative position, as shown in (6).
(6) [DP β D β [NP Γ e [ α N]]
Del Gobbo (2005) and Larson & Takahashi (2007) base their analysis on the ordering restriction of relative clauses in Mandarin (Del Gobbo 2005, Larson &
Takahashi 2007), Japanese, Korean and Turkish (Larson & Takahashi 2007). I-level relative clauses order freely among themselves and so do s-level relative clauses.
However, when the two types of relative clauses co-occur, an ordering restriction is imposed. They conclude that i-level relative clauses belong to NP-modification while s-level relative clauses belong to DP-modification. Therefore, an ordering restriction emerges when the i-level and s-level relative clauses co-occur. Based on their data and the formula in (6), three predictions can be made. First, the modifiers between D and N can be either level or i-level modifiers. Second, when i-level modifiers and s-level modifiers co-occur, i-s-level modifiers must occur closer to the noun. Third, only s-level modifiers can precede the demonstrative. The third prediction, however, cannot adequately cover the data. In fact, i-level modifiers can precede a demonstrative, which will be discussed later in Lin’s (2008) semantic approach.
Zhang (2015) also follows the two-layer approach and proposes that the nominal phrases with the modifiers before or after the DNC involve nominal internal movement. First, she shows that any de structure including PP complements, NP
modifiers, PP modifiers, locative modifiers, measure phrase modifiers, DegPs and reduplicate adjectives can be extracted beyond the demonstrative.
Moreover, she identifies several restrictions on the extracted modifiers, although some of the restrictions are not well-attested as will be discussed later. First, Zhang (2015:379) maintains that non-predicative APs cannot be extracted, includiing suowei
‘so called’, youji ‘organic’, chu-ji ‘basic-leveled’, er-ji ‘second-leveled’, duoxiang
‘multiple’, man-xing ‘slow-type’, gongtong ‘shared’, gebie ‘individual’, and ren-wei
‘human-made’. However, the adjectives listed above can be extracted, as shown in (7).
(7) a. Women jia zuijin dou chi youji de na zhong shucai.
our home recently all eat organic DE that CL vegetable ‘Our family started to eat organic vegetables recently.’
b. renwei de na chang zuotian fasheng de zainan human.made DE that CL yesterday happen DE disaster rang dajia hen qifen.
make everyone very angry
‘The disaster, which is human made, makes everyone angry.’
Second, extracted modifiers cannot show intersective reading while in-situ modifiers can have both intersective and non-intersective readings. Zhang’s (2015:5) examples are shown in (8) and (9). However, the interpretation difference in (9) might not be solid since both of the orders in (9) make sense to me and the informants I consulted.
(8) a. liang wei naixin de laoshi (Zhang 2015:382) two CL patient DE teacher
‘two teachers, who are patient persons’
‘two teachers, who teach patiently’
b. naixin de liang wei laoshi patient DE two CL teacher ‘two teachers, who are patient persons’
# ‘two teachers, who teach patiently’
(9) [liang wei naixin de laoshi / # naixin de liang wei laoshi]
two CL patient DE teacher patient DE two CL teacher hui-dao jia-li piqi baozao.
home-in return-to temper grumpy
‘The two patient teachers returned to their homes, and they became grumpy.’
Third, extracted modifiers lose their idiomatic readings while in-situ modifiers can have both idiomatic and literal meanings, as shown in (10).
(10) a. na liu ben huangse de xiaoshuo23 (Zhang 2015:383) that six CL yellow DE novel
‘the six yellow novels’
‘the six pornographic novels’
23 As is pointed out by Professor Chen-sheng Liu, the two meanings of huangse might belong to two lemmas. The two meanings arise from lexical ambiguity rather than
b. huangse de na liu ben xiaoshuo yellow DE that six CL novel
‘the six yellow novels’
# ‘the six pornographic novels’
Fourth, the intensifying demonstrative zheyang ‘such’ cannot be followed by extracted modifiers, as shown in (11).
(11) a. zheyang (de) san ben Luxun xie de shu (Zhang 2015:383) such DE three CL Luxun write DE book
‘such three books that Luxun wrote’
b. * zheyang (de) Luxun xie de san ben shu such DE Luxun write DE three CL book
Fifth, an indefinite noun phrase modified by an extracted modifier cannot be non-specific, i.e. it is always specific (Huang et al. 2009, Zhang 2015), as shown in the following tests (12)-(13). The specificity restriction according to Zhang is due to movement to the left edge of the D-domain.
(12) irrealis context (Zhang 2015:385)
a. Ruguo ni xiang kan yi bu hen duan de dianying,…
if you want see one CL very short DE movie ‘If you want to see a very short movie,…’
b. * Ruguo ni xiang kan hen duan de yi bu dianying,…
(13) appearance verbs (Zhang 2015:385)
a. Lai-le san ge dai yanjing de xuesheng.
come-PRF three CL wear glasses DE student
‘Here come three students who wear glasses.’
b. * Lai-le dai yanjing de san ge xuesheng.
come-PRF wear glasses DE three CL student
Last, de-stranding cannot occur in a noun phrase with extracted modifiers, as shown in (14). In (14), de is stranded in the clause-final position with the elements after de elided.
(14) a. Lili mai-le (na) liu ben hen hou de shu, Lili buy-PRF that six CL very thick DE book Nini ze mail-le (na) liu ben hen bo de.
Nini but buy-PRF that six CL very thin DE
‘Lili bought six very thick books, but Nini bought six very thin ones.’
b. * Lili mai-le hen hou de (na) liu ben shu, Lili buy- PRF very thick DE that six CL book Nini ze mail-le hen bo de (na) liu ben shu.
Nini but buy- PRF very thin DE that six CL book
Based on Cinque’s (2010) theory of functional projections (FP) and Zhang’s (2015:389) own observation, she assumes that there are two FP zones where “the lower one hosts direct modification adjectives” while “the higher one hosts RCs and
indirect modification adjectives, which are reduced RCs” and that each adjective is at Spec of an FP.
(15) [FPH Aindirect [FPL Adirect N]]
Also, she assumes that de is the realization of a functional head, as shown in (16).
Following Cinque (2010:97), Zhang proposes that de in an indirect modification structure is necessarily pronounced while it is sometimes unpronounced in a direct modification structure.
(16) FP
3 XP 3 Spec F YP
DE Complement
Under the assumptions, she explains the restrictions. To explain the immobility of non-predicative adjectives, she proposes that indirect modifiers, which stay in the FPH, can move and direct modifiers in the FPL do not move.
The most crucial problem for Zhang’s (2015) analysis is the distinction between FPH and FPL. To summarize Zhang’s (2015) ideas, modifiers occurring in the low zone include (i) adjectives whose de can be optionally pronounced, (ii) idiomatic adjectives, and (iii) non-predicative adjectives.
First, from Zhang’s perspective, adjectives which cannot be extracted are in the low zone and those which can be moved are in the high zone. Also, adjectives whose de can be optionally pronounced belong to the low zone in terms of Zhang’s proposal.
such as huang ‘yellow’ in huang (de) gou ‘yellow dog’. In this case, huang cannot serve as predicates and de can be optionally pronounced, so it should belong to the low zone. However, huang can serve as the target of extraction as in (17), contrary to her predictions. In contrast, Paul (2005)’s proposal is clearer, as discussed in the last paragraph of Section 3.1. According to Paul (2005), any adjective without de such as huang gou ‘yellow dog’ (de-less modification) occurs in the low zone while any
adjective with de occurs in the high zone uch as huang de gou ‘yellow dog’.
(17) huang de na zhi gou bijiao keai yellow DE that CL dog more cute
‘The dog which is yellow is cuter.’
Moreover, to explain the limitation on idiomatic expressions as in (10), Zhang (2015) proposes that modifiers receive idiomatic readings in the FPL and literal readings in the FPH. The dichotomy seems arbitrary. A modifier of the literal meaning such as huangse ‘yellow’ in (10) belongs to the high zone when ambiguity between literal and idiomatic meanings arises. On the other hand, when it occurs in a noun phrase without the idiomatic reading (which causes no ambiguity) such as huangse (de) waitao ‘yellow coat’, the same modifier (with its literal meaning) belongs to the
low zone. Most important of all, the modifier huangse can be extracted in either case where they are supposed to be in the low zone (when interpreted literally without de or when interpreted idiomatically) since de is optional, as in (18). In (18a), the sentence is natural even without certain context and the modifier huangse can be extracted even if it is analyzed as occurring in the low zone. In (18b), with a proper context where huangse is the topic of the sentence, the sentence becomes acceptable.
(18) a. Huangse (de) na jian waitao bang wo na guolai.
yellow DE that CL coat help me get here ‘Help me get the yellow coat over here.’
b. Context: A father witnesses his son reading pornographic novels and says the following sentence.
? Huangse (de) na zhong xiaoshuo bu-yao kan tai duo.
yellow DE that CL novel not-should read too much24
‘Don’t read the pornographic novels too often.’
Last, a set of non-predicative adjectives receive literal meanings in the FPL because they cannot serve as predicates. However, in contrary to the proposal that adjectives in the FPL do not move, they can precede a demonstrative as shown previously in (7).
In summary, Zhang’s nominal-internal movement proposal accounts for various properties observed with adjectives and the relevant ordering for the most part, but her two-layer analysis cannot correctly capture the empirical data and some of her proposed restrictions on movement, therefore, do not solidly hold.
24 Although I think (18b) is accepted in a proper context, it is in fact not widely-accepted to all speakers. Among the ten speakers I consulted, five of them could get the idiomatic reading while the other five speakers could not get the idiomatic reading or would judge the sentence as ungrammatical. However, a crucial fact is that, for