2. Méi yǒu rén ‘There isn't anybody’ ≈ ‘There is nobody’

The incorrect analysis of méi yǒu rén as a DP ‘nobody’ is clearly an effect of the translation into English of the Chinese negated existential construction with a secondary predicate on rén ‘person’, the internal argument of yǒu ‘exist’ (see (9)). (For secondary predicates in existential constructions, see Huang Reference Huang1984, Reference Huang, Reuland and ter Meulen1987.)

There are myriads of examples involving NPs different from rén ‘person’ available in every good grammar manual such as Lü (Reference Lü2000: 382−383), further highlighting the clausal status of méi yǒu rén ‘there isn't anybody’:

Another argument against the DP-hood of méi yǒu rén ‘there isn't anybody’ is its unacceptability as the complement of a preposition (see Huang Reference Huang, Li and Simpson2003: 4; (11c)):

Huang (Reference Huang, Li and Simpson2003: 4, note 7) likewise states the unacceptability of méi yǒu rén ‘there isn't anybody’ in object position (see (6) above) and therefore evidently evokes the analysis argued for here with méi yǒu rén as a negated existential construction ‘there isn't anybody’. However, Huang (Reference Huang, Li and Simpson2003: 19) discards it in the end (emphasis mine):

Instead, méi yǒu rén ‘there isn't anybody’ is assigned NP status on a par with English nobody. Its unacceptability in object position is then accounted for by the absence of V-to-Infl(-to C) movement in Chinese, which leaves the verb between negation and the NP in object position. As a result, the latter two are not adjacent and their conflation into one NP is not possible, either, as proposed by Christensen (Reference Christensen, Dahl and Holmberg1986) for negative NPs in Norwegian V2 sentences, based on Klima's (Reference Klima, Fodor and Katz1964) analysis of English nobody as the conflation of not and anybody. The same account is applied to the unacceptability of méi yǒu rén ‘there isn't anybody’ as the complement of a preposition, given that negation precedes a preverbal PP adjoined to vP and will therefore never be adjacent to the complement NP within the PP.Footnote ⁸

This analysis is not on the right track, because inter alia it obscures the parallel with the affirmative existential construction, for which Huang (Reference Huang, Li and Simpson2003) provides an example, given below:

Huang (Reference Huang, Li and Simpson2003: 18) cites this example to show the contrast with the corresponding English sentence Someone bought every book, which is ambiguous and allows for two readings: ∃ > ∀ and ∀ > ∃.

In fact, yǒu ‘exist’ in (12) is not optional, unless yī ge rén ‘one cl person’ is to be understood as ‘a certain person’ instead of ‘some person’. Because, as is well-known, Chinese does not allow for indefinite non-specific DPs in subject position;Footnote ⁹ the latter are, however, acceptable as the internal argument of the existential verb yǒu:

The specific reading of a Number Phrase ‘Num CL NP’ is favoured by an episodic predicate (see Fan Reference Fan1985, Li L. Reference Li1990, Tang Reference Tang2005, among many others), where accordingly ‘yī CL NP’ is acceptable in the subject position:

Tang (Reference Tang2005) contrasts (14a) featuring the episodic predicate ‘appeared from behind the girl’ with the individual-level predicate ‘be intelligent’ in (14b), where yǒu ‘exist’ obligatorily precedes yī ge xiǎoháizi ‘a child’. Example (16) is a sentence that reports an observation, and hence has a specific subject and an episodic predicate (see, a.o., Y.-H. A. Li Reference Li1996, Reference Li1998; Huang et al. Reference Huang, Li and Li2009: Ch. 8, for discussion of this constraint on subjects.)

The ban on indefinite non-specific DPs in subject position also explains why Chinese has no DP equivalent for someone, either; instead, this is again to be rendered by the existential construction (17), with an eventual secondary predicate on (yī ge) rén, as in (13a) above:

On a par with méi yǒu rén ‘there isn't anybody’, yǒu (yī ge) rén ‘there is a person’ is unacceptable in object position or as complement to a preposition, a fact well-known by every L2 learner of Chinese who in the beginning produces the unacceptable sentences below based on the wrong assumption that yǒu (yī ge) rén is a DP:

The clausal nature of both yǒu rén ‘there is somebody’ and méi yǒu rén ‘there isn't anybody’ is confirmed by the question-answer pair below, where the yes/no question in (19a) is formed in the syntactic pattern ‘V-not-V’ (see Huang Reference Huang1982), juxtaposing the affirmative and the negative counterparts of the verb:

Example (19b) is the positive answer and (19c) the negative answer; in both, the existential verb yǒu on its own without rén ‘person’ is sufficient, thus further demonstrating the verb status of yǒu and the clausal nature of (méi) yǒu rén.Footnote ¹⁰ The same holds for (20) with a secondary predicate:

Finally, once again, rén ‘person’ is only one of many NPs that can be the internal argument of the verb yǒu ‘exist’ in a yes/no question:

All these well-known data are incompatible with an analysis of méi yǒu rén and yǒu (yī ge) rén as DPs (nobody and someone, respectively), but obtain automatically under the clausal analysis.

Before concluding this section, let us briefly address some general issues. Given that yǒu ‘exist’ can be negated and modified by adverbs like all other verbs, Milsark's (Reference Milsark1974) approach is adopted, where the existential verb is not an operator itself, but introduces an operator (see Y.-H. A. Li Reference Li1996 for a mixed approach). Reformulating Milsark (Reference Milsark1974) by using the now generalized distinction of pivot vs. coda (see McNally Reference McNally, Heusinger, Maienborn and Portner2011: 8136), the pivot nominal following the existential verb is a property restricting the existential operator, whereas the coda (i.e., the secondary predicate)̧ indicates the scope of the existential operator.Footnote ¹¹

1. (22) There are [_pivot students] [_coda waiting in the classroom].

The important question already raised by Milsark (Reference Milsark1974: 19) as to whether pivot and coda form a constituent (an NP immediately dominating an S in Milsark Reference Milsark1974) or whether the coda is a separate constituent attached to VP or to S has so far not been satisfactorily answered for Chinese.

Huang (Reference Huang, Reuland and ter Meulen1987: 236, Reference Huang1988: 57) tentatively suggests an analysis where the DP following yǒu ‘exist’ and the secondary predicate constitute the complement clause of yǒu ‘exist’:Footnote ¹²

The second possible analysis for Chinese takes up McNally's (Reference McNally1992) assumptions that the pivot is the only argument of the existential predicate and that the coda is a VP-internal adjunct modifier that stands in a control relation to the pivot.

1. (24)

   

In (24), the VP consisting of the verb yǒu ‘exist’ and its internal argument is merged with the secondary predicate (also see Irimia Reference Irimia2005), which has the size of TP, given the acceptability of auxiliaries, aspect suffixes, etc. here. Its always covert subject, PRO, is coindexed with the internal argument of yǒu ‘exist’. (For the relevance of “weak” c-command in Chinese, see Huang et al. Reference Huang, Li and Li2009: 335.)

Finally, based on Huang's (Reference Huang1984) early intuition that secondary predicates should be treated on a par with purposive clauses, a third structure is possible:

1. (25)

   

This structure is based on Wei and Li's (Reference Wei and Audrey Li2018) analysis of postverbal purposive clauses as control complements:

Wei and Li (Reference Wei and Audrey Li2018: 309−322) provide ample evidence that structurally, the purposive clause is a complement to the verb and projects a VP, on par with the infinitival complements of control verbs such as kāishǐ ‘begin’ and jìxù ‘continue’ (see Huang Reference Huang2017).

By contrast, as indicated in (25) above, the size of the secondary predicate in existential constructions is that of a TP (with an always covert subject), as evidenced by the presence of aspect suffixes, auxiliaries, and negation, as well as adverbs and adjunct XPs preceding the negation. Since negation indicates the left edge of the extended verbal projection in Chinese, the secondary predicate must be larger than the extended vP:

As for the choice among the three structures, for the purpose of this article, I adopt the third one in (25), with a clear bipartitioning into matrix clause and secondary predicate. As we will see below, this allows us to account for the scopal behaviour of focus adverbs preceding yǒu ‘exist’ in the existential construction. The lack of this bipartitioning is the major drawback of Huang's (Reference Huang1988) structure (23) and to a certain extent also that of (24).

Further research is needed to definitively decide between the configurations (25) and (24), because the few studies on secondary predicates subsequent to Huang (Reference Huang, Reuland and ter Meulen1987) (Tsai Reference Tsai1994, Lin and Tsai Reference Lin, Tsai, Li, Simpson and Tsai2015, a.o.) never address the important issue of the hierarchical position of secondary predicates in the clausal spine with respect to the object DP. The only consensus existing is that the secondary predicate must be located in VP or vP. Merging with a higher projection in TP is excluded by the overall syntax of Chinese, where, due to the systematic head-initiality of the extended verbal projection (including TP), postverbal material must be merged in the vP/VP. “That the XP [i.e., the secondary predicate, WP] when it appears, is under VP, but not immediately under S is assumed in all discussions” (Huang Reference Huang, Reuland and ter Meulen1987: 232).

3. Zhǐ yǒu DP ‘There is only DP’

Recall from (3b), repeated as (28a) below, that ‘zhǐ yǒu DP’ is to be analyzed, not as a DP, but as the existential construction ‘there is only DP’, as evidenced by its unacceptability in object position (28b):

Accordingly, zhǐ yǒu is not a single word zhǐyǒu ‘only’, but the adverb zhǐ ‘only’ preceding the existential verb yǒu (pace Tsai Reference Tsai2004, Erlewine Reference Erlewine2015, a.o.),Footnote ¹³ as evidenced by the compatibility of zhǐ ‘only’ with other verbs ((28c), (29a–b)) and the compatibility of yǒu ‘exist’ with the nearly synonymous adverb jǐnjǐn ‘only’ (29c).

Being an adverb, zhǐ ‘only’ must merge with a verbal projection and precede its highest head (see Paul Reference Paul and Sybesma2017a and references therein), which explains the unacceptability of both zhǐ DP and zhǐ PP:

The only way to render the meaning intended in (30a) is with the adverb zhǐ ‘only’ preceding and modifying the vP, as in (30b) (also see (28c) above).

The well-formedness of the sequence ‘zhǐ PP VP’ does not invalidate the observations above, because like any other adverb (e.g., chángcháng ‘often’), zhǐ ‘only’ combines with the vP containing the adjunct PP (31a). When in the topic position above TP, ‘adverb + PP’ is clearly unacceptable (31b):

As illustrated in (31c), the same unacceptability is observed for zhǐ yǒu ‘only exist’. Accordingly, neither the adverb zhǐ ‘only’ nor its combination with the verb yǒu ‘exist’, zhǐ yǒu, can be analyzed as “constituent only” in the sense of Beaver and Clark (Reference Beaver and Clark2008: 235), as suggested by an anonymous reviewer.

This is confirmed by Lü (Reference Lü2000: 678−679), who postulates a covert verb for the rare acceptable cases of ‘zhǐ DP’:

Crucially, zhǐ DP ‘only DP’ is confined to a position where an existential construction with zhǐ ‘only’ preceding an unaccusative verb (yǒu ‘exist’ or shì ‘be’) is acceptable, that is, either the sentence-initial position, as in (32b), or following a locative postpositional phrase, as in (32a).

Similar to English and other languages, definite DPs and Number Phrases (as in (28a) and (29c) above) are perfectly acceptable when zhǐ ‘only’ modifies the existential verb, whereas bare nouns are infelicitous, because it leads to an uninformative statement (see Beaver and Clark Reference Beaver and Clark2003: 336):Footnote ¹⁴

Given the (crosslinguistic) pervasiveness of existential constructions with ‘only’ and a definite DP as internal argument of an existential verb, a way must be found to rule them in, alongside other well-known examples (35a–c) that disobey the otherwise observed Definiteness Effect (DE), which excludes definite DPs from existential constructions (34):

Exceptions to the DE have been observed since Milsark (Reference Milsark1974) (see (35a–c)) and different approaches (syntactic, semantic, pragmatic) have been pursued ever since (see, a.o., McNally Reference McNally1998, for the necessity of distinguishing between definite NPs and proper names on the one hand, and quantificational NPs on the other; see Fischer et al. Reference Fischer, Kupitsch and Rinke2016 for an overview).

In Chinese as well, the constraints ruling the DE are still very poorly understood. Not much progress has been made since Huang (Reference Huang, Reuland and ter Meulen1987), who considers his own in-depth investigation of the DE in Chinese as “inconclusive” on p. 250, given that too many different factors are involved. Two cases should suffice to illustrate his point (also see Y.-H. A. Li Reference Li1996).

Definite DPs are acceptable when they are members of a list (36b). Definite DPs are likewise allowed as the internal argument of unaccusative verbs such as lái ‘come’ and zǒu ‘leave’ in non-root contexts, although native speakers’ judgements differ here (36c):

Against this backdrop, a violation of the DE does not constitute a counterargument against the clausal analysis proposed here, the more so as the ‘only exist’ sentences form a clearly definable class of “exceptions”, precisely excluding indefinite NPs, which are otherwise acceptable as internal arguments par excellence in the canonical existential construction.

Going back to the fundamental difference between the adverb zhǐ ‘only’ and the existential construction zhǐ yǒu DP ‘there is only DP’, the following observation by Erlewine (Reference Erlewine2015: 24) provides further evidence in its favour:Footnote ¹⁵

In (37), zhǐ ‘only’ can only focus Lisi, not any of the DPs in the complement clause, irrespective of whether they bear phonological stress or not.

The same holds for zhǐ yǒu DP ‘there is only DP’ with a secondary predicate:Footnote ¹⁶

In (38a–b), zhǐ ‘only’ exclusively focuses on Lisi as the internal argument of the matrix verb yǒu ‘exist’. The adverb zhǐ ‘only’ cannot associate with a DP in the secondary predicate, be it in postverbal (38a) or preverbal position (38b) (again irrespective of whether they bear phonological stress or not). The exclusiveness effect observed here points to a clear bipartitioning of the sentence into focus and presupposition and provides an additional argument for the structure proposed in this article, with the existential verb yǒu and its internal argument in the matrix clause and hence in a domain distinct from the secondary predicate TP.Footnote ¹⁷

The facts in (38a–b) contrast with (39) from Erlewine (Reference Erlewine2015: 24), which is not an existential construction, but a standard SVO sentence: zhǐ ‘only’ occurs below the subject Lisi and precedes the matrix verb shuō ‘say’, which in turn selects a clausal complement.

As noted by Erlewine (Reference Erlewine2015: 24) zhǐ ‘only’ can “associate with focus” with either the subject or the object DP in the clausal complement, where intonational prominence on the respective DP is required. A third possibility (not mentioned by Erlewine Reference Erlewine2015) is association of only with the entire clausal complement, as in (39iii) (Liu Chang p.c.).

To summarize, when preceding the verb below the subject in a simple SVO sentence, zhǐ ‘only’ involves “association with focus” with any (intonationally prominent) DP in its c-command domain, that is, to its right. By contrast, when zhǐ ‘only’ precedes the verb yǒu ‘exist’ in the existential construction, this results in an exclusive focus on its internal argument, not on the DP(s) within the secondary predicate, thus indicating a bipartitioning into focus and presupposition.Footnote ¹⁸

4. Hěn shǎo rén ‘few people’ and hěnshǎo yǒu rén ‘there are rarely people’

The case of hěn shǎo rén ‘very few person’ = ‘few people’ is the most complex of the three alleged DP/QP candidates, foremost because quite a number of speakers downright reject it or only very marginally accept it.Footnote ¹⁹ However, the same speakers use the existential construction hěnshǎo yǒu rén ‘rarely exist person’ = ‘There are rarely people’, where the adverb hěnshǎo ‘rarely’ precedes the verb yǒu ‘exist’; since sometimes yǒu ‘exist’ remains covert, this gives rise to an apparent QP: hěn shǎo rén ‘few people’ (see the discussion below).

Furthermore, even for those speakers who accept hěn shǎo rén ‘few people’ (without any covert yǒu ‘exist’), it is not simply the antonym of the QP hěn duō rén ‘many people’, as tacitly assumed in the literature. In particular, it is not possible to simply attribute differences between the two to the semantic contrast between monotone decreasing vs. increasing quantifiers. Instead, the differences observed are foremost due to syntax, in particular the fact that hěn shǎo rén ‘few people’ does not have the same distribution as hěn duō rén ‘many people’.

As demonstrated below in this section, native speakers can be divided into three groups. I start out with providing the data baseline, representative of group 1, which will then serve as backdrop for the description of groups 2 and 3. I call this data baseline, because groups 2 and 3 likewise use the constructions judged acceptable by group 1. Group 3 is the most “encompassing” group, for it in turn accepts the constructions judged well-formed by group 2. Importantly, young speakers (i.e., university students) are present in all groups, although to a lesser degree in group 1. We thus do not observe an “ongoing change” here, because for a given individual speaker, there is no change at all, given that s/he has a fixed set of syntactic and semantic properties associated with her/his grammar of hěnshǎo yǒu rén ‘there are rarely people’ and hěn shǎo rén ‘few people’, respectively. “Ongoing change” is an unfortunate metaphor used by the linguist when confronted with the simultaneous existence of groups of native speakers having different internalized grammars for a given linguistic phenomenon. (See Hale Reference Hale2007 for extensive discussion of syntactic change vs. diffusion of that change.)

4.1. The data baseline: Group 1

As just mentioned, when preceding the existential verb yǒu ‘exist’ (40a) or other verbs (40b–c), hěnshǎo instantiates the adverb ‘rarely, on few occasions’:

By comparison, the adverb shǎo means ‘a bit, little, less’:

When wanting to ascribe a predicate to a small number of people, speakers from Group 1 use the existential construction with a secondary predicate, where yǒu ‘exist’ is modified by the adverb hěnshǎo ‘rarely’. They never use hěn shǎo rén ‘few people’, and every sentence-initial hěn shǎo rén is spontaneously corrected by adding yǒu ‘exist’: hěnshǎo yǒu rén ‘rarely exist person’.

This existential construction is also used with NPs different from rén ‘person’ (42b).

The adverb status of hěnshǎo ‘rarely’ and hence the clausal nature of ‘hěnshǎo yǒu DP’ is particularly neat in the examples below provided by an anonymous reviewer, where the internal argument DP can be independently quantified (43a).

Examples (43a–b) likewise demonstrate that hěnshǎo ‘rarely’ only modifies the matrix existential verb.

Given its clausal status, hěnshǎo yǒu rén ‘there are rarely people’ is naturally unacceptable in the postverbal object position (with verbs selecting only DPs) and as complement of prepositions; it thus contrasts with the QP hěn duō rén ‘many people’, which as a nominal projection is acceptable here (for all speakers):

Again, the same holds for NPs other than rén ‘person’:

The meaning intended in (45a) ‘He knows few people/students’ can be rendered as in (46), with the quantitative adjective shǎo ‘be little, few’ as matrix predicate:

The adverb hěn ‘very’ is required for the positive degree and therefore remains untranslated, in contrast to other degree adverbs such as tài ‘too’ (see Paul Reference Paul, Cabredo-Hofherr and Matushansky2010 and references therein).

There is also a translation corresponding structurally more closely to the English ‘He knows few people/students’, with hěn shǎo as modifier of rén ‘person’ and xuésheng ‘student’, respectively, and followed by the subordinator de:Footnote ²¹

The construction in (47) with an uncontroversial DP as object is in principle acceptable for many speakers, although to different degrees (as indicated by “%”). Importantly, while (47) is subject to many constraints (see section 4.2 immediately below), this is not the case for (46), which is the preferred, most “natural” version, even for speakers who fully accept (47).

The discussion of the constraints holding for (47) will lead us beyond the data baseline and will confront us with variation among native speakers, indicating the co-existence of several groups. Importantly, the speakers from groups 2 and 3 accept the constructions judged as well-formed in the baseline data, but differ in whether and in which syntactic contexts they accept hěn shǎo rén ‘few people’ and hěn shǎo de rén ‘few people’.

4.2. Beyond the data baseline: Group 2

Liu (Reference Liu2011: 103) provides the following triplet to illustrate the constraints holding for hěn shǎo ‘very be.few’ as DP modifier:Footnote ²³

The contrast between (48a) and (48b) shows that hěn shǎo ‘very be.few’ as Adjectival Phrase requires the subordinator de when modifying a DP. However, even with de, (48b) is not 100% felicitous, either, but requires the presence of the adverb zhǐ ‘only’ preceding the verb, an observation confirmed by the native speakers consulted.

The judgements in (48a–c) likewise hold for rén ‘person’ in hěn shǎo rén ‘few people’ as well as hěn shǎo de rén ‘few people’ and define the native speakers I refer to as group 2:

For PPs with hěn shǎo de rén ‘few people’ as complement, zhǐ ‘only’ modifying the entire VP, including the preverbal adjunct PP, is again required for full acceptability. Note that group 1 speakers likewise accept these PPs:

Besides the presence of the adverb zhǐ ‘only’, Liu (Reference Liu2011: 104) observes other constraints holding for ‘hěn shǎo de NP’, such as a parallelism requirement:

Furthermore, he construes several minimal pairs of the type illustrated in (52a) and (53a) below with hen duō ‘very be much’ and hěn shǎo ‘very be few’ as modifiers in an object DP, and states the systematic unacceptability of the latter. Recall from the discussion of (46) above that the paraphrases in (52b) and (53b) are perfectly acceptable for all speakers.

Liu (Reference Liu2011) therefore concludes that hěn shǎo de NP ‘very be.few sub NP’ is not on a par with hěn duō (de) NP ‘very be.much sub NP’ = ‘many NP’, where no such constraints are observed and where the subordinator de is optional, not obligatory as it is for hěn shǎo (see (49a–b) above).

Concerning hěn shǎo rén ‘few people’, Liu (Reference Liu2011: 103) reports 189 instances in texts dating from the late 1990s in the corpus of the Center for Chinese Linguistics at Peking University, all of them occurring in sentence-initial position.Footnote ²⁶ There are no examples of hěn shǎo ‘very be few’ directly preceding NPs, other than rén ‘person’, in his corpus. The few examples of hěn shǎo NP (i.e., without the subordinator de) found via a Google search are judged as only marginally acceptable by Liu (Reference Liu2011: 104: 30, 31). In addition, hěn shǎo ‘very be few’ is not a DP modifier here, because adding de leads to an unacceptable result; instead yǒu ‘exist’ must be reconstructed as in (54) below, i.e., these are cases of the existential construction with hěnshǎo as adverb ‘rarely’ and a secondary predicate on the internal argument NP of yǒu ‘exist’: ‘there are rarely NP VP-ing’:

Going back to the 189 instances of the sequence hěn shǎo rén attested in the Peking University corpus, the fact that they exclusively occur in sentence-initial position provides us with an important clue. Given the lack of an expletive subject in the existential construction in Chinese and the fact that hěnshǎo ‘rarely’ is a vP level adverb, the only XPs liable to precede the existential construction hěnshǎo yǒu rén are sentence-level adverbs or adjunct XPs (such as the PP zài déguó ‘in Germany’ in (54)). When yǒu ‘exist’ is covert, the surface sequence hěn shǎo rén can be reanalyzed as a nominal projection in the subject position (SpecTP), and the secondary predicate as matrix predicate. This is the reason why hěn shǎo rén ‘few people’ exclusively occurs in the subject position, that is, in a sentence-initial position in the broad sense, as explained above. Examples (55a) and (55b) below show the two relevant parsings:

This is a plausible reanalysis, because the c-command relations between all constituents are maintained in (55b); that is, (55b) shows the same hierarchical relations as (55a), in accordance with the Conservancy of Structure Constraint (Whitman Reference Whitman, Pintzuk, Tsoulas and Warner2001). Importantly, both constructions (55a) and (55b) remain in use and can be employed by the same speaker (see the discussion immediately below). Concerning the semantic side, quantifying over a situation as in hěn shǎo yǒu rén VP ‘there are rarely people VP-ing’ can – depending on the meaning of the sentence – imply ‘few people VP’, and it is this possible implication which gives rise to the analysis of hěn shǎo rén with a covert yǒu ‘exist’ as a QP ‘few people’ when followed by a secondary predicate.

Note that assuming a covert yǒu ‘exist’ is not an isolated fact limited to the existential construction with the adverb hěnshǎo ‘rarely’; a covert verb was likewise postulated for the adverb zhǐ ‘only’ plus DP in sentence-initial position (see (32a) above, repeated in footnote 28 below). However, while in the case of sentence-initial zhǐ DP ‘only DP’, there are indeed reasons to assume that yǒu ‘exist’ is always present, albeit covertly,Footnote ²⁸ the situation is different for hěn shǎo rén ‘few people’. Because for a subset of group 2 speakers, sentence-initial hěn shǎo rén ‘few people’ no longer involves a covert yǒu ‘exist’, but has been reanalyzed as a QP ‘few people’, as evidenced by the difference these speakers make between hěn shǎo yǒu rén ‘there are rarely people’, on the one hand, and hěn shǎo rén ‘few people’, on the other:

In (56b), yǒu ‘exist’ is ruled out, because hěn shǎo rén ‘few (of them)’ has tāmen ‘they’ in the first part as antecedent; accordingly, only a nominal projection is acceptable here and the existential construction is excluded. Example (56a), however, lacks such a constraining syntactic context and therefore allows for the existential construction ‘there are rarely people’ plus a secondary predicate, given that the overall meaning of the sentence is compatible with such a general statement.Footnote ³⁰

4.3. Group 3

The situation seems to have further evolved since the time of Liu's (Reference Liu2011) article and there is another, third group in addition to the baseline speakers (group 1) and to those described by Liu (Reference Liu2011), that is, my group 2. This third group not only accepts the QP hěn shǎo rén ‘few people’ in subject position (as group 2 does), but also in object position, where group 2 only accepts the DP hěn shǎo de rén, with hěn shǎo ‘be few’ as a DP internal modifier. Hěn shǎo rén ‘few people’ as complement in a PP is likewise fine for group 3:

In addition, again unlike group 2, group 3 speakers also allow QPs with NPs other than rén ‘person’ (58a). They also accept hěn shǎo as DP modifier with de (58b), as group 2 does, modulo the fact that the adverb zhǐ ‘only’ is not required. In other words, group 3 speakers seem to analyze hěn shǎo ‘few’ on a par with hěn duō ‘many’, either as a modifier (with de) or as a quantifier (without de):

Group 3 co-exists with the two other groups and accepts all constructions judged well-formed for groups 1 and 2; importantly, all three of them include young speakers (i.e., university students), although as a minority in group 1. Vice versa, speakers of groups 1 and 2 are very well aware of group 3 speakers, as reflected in comments such as “This construction is unacceptable for me, but it may be fine for others”. This is especially the case for group 2 speakers when confronted with hěn shǎo rén ‘few people’ in object position.

Note, though, that notwithstanding the acceptance by native speakers of hěn shǎo rén ‘few people’ in object position in judgement tasks, the actual distribution of hěn shǎo rén ‘few people’ and hěn shǎo de rén ‘few people’ is much more constrained than that of hěn duō (de) rén ‘many people’. More precisely, the majority of hěn shǎo rén ‘few people’ occurs in sentence-initial position, that is, the position where the reanalysis of hěn shǎo yǒu rén ‘there are rarely people’ as a QP hěn shǎo rén few people’ took place, and there are only a few cases of hěn shǎo rén ‘few people’ in the postverbal object position. By contrast, the majority of hěn shǎo de rén ‘few people’ (with the subordinator de) are found in postverbal object position. This is the result of a corpus search (filtered by checks with native speakers) for hěn shǎo yǒu rén ‘there are rarely people’, hěn shǎo rén and hěn shǎo de rén ‘few people’.Footnote ³¹ The brief overview of the figures for each sequence below does not claim any statistic validity; its main purpose is to highlight the complexity of the data situation for both hěn shǎo rén and hěn shǎo de rén ‘few people’ and to insist on the fact that they are not simply the counterpart of hen duo (de) ren ‘many people’ and can therefore not be directly compared with, for example, the English QPs many people and few people, either.

Let us start with hěn shǎo yǒu rén ‘there are rarely people’, with 419 examples from literary works, 2253 from newspapers and periodicals, and 1892 from the microblogging website Weibo. Nearly all examples feature a secondary predicate (59b–c), and there are only a handful of examples with hěn shǎo yǒu rén on its own (59a):

The high frequency of hěn shǎo yǒu rén ‘there are rarely people’ in Weibo clearly shows that hěn shǎo yǒu rén is likewise used by the younger generation (as the probable majority among bloggers),Footnote ³³ a result confirmed by an informal acceptability judgement test with 15 native speakers (between 22 and 27 years) carried out by Yan Shanshan (p.c.) at Peking University

Turning now to hěn shǎo rén ‘few people’, the corpus provides 147 examples from literary texts, 307 examples from newspapers and periodicals, and 886 from the blog Weibo. Importantly, the majority appears in the subject position (including the subject position in complement clauses, see (63)), where it may be preceded by sentence-level adverbs such as guòqù ‘in the past’ and hòulái ‘afterwards, later’, as well as topicalized phrases (see (60)).Footnote ³⁴ As explained above, these are the very same syntactic environments that allow for the existential construction, given that hěnshǎo ‘rarely’ as VP-level adverb must follow sentence-level adverbs and topicalized XPs. This is illustrated by (63), where the native speakers consulted about the corpus sentence (63a) in fact either preferred or required the presence of yǒu ‘exist’ (see (63b)), because the existential construction was judged more appropriate for conveying the intended general statement:

The case of (63a–b) illustrates the necessity to control for a covert existential verb yǒu in the instances of sentence-initial hěn shǎo rén ‘few people’ and the impossibility of automatically assigning it the same QP status as hěn duō rén ‘many people’.

The few examples of hěn shǎo rén ‘few people’ in postverbal position, hence as a QP, show it as internal argument of either the unaccusative verb lái ‘come’ or the existential verb yǒu preceded by the adverb zhǐ ‘only’:

Evidently, (64a–c) are only acceptable for the speakers from (a subset of) groups 2 and 3 who, in addition to hěn shǎo yǒu rén ‘there are rarely people’, also have the QP hěn shǎo rén ‘few people’ in their grammar. By contrast, speakers from group 1 simply reject (64a–c).

Note that it is the necessity of presenting the facts in a certain order that gives the impression of a linear development, with new groups adding on successively, but this does not reflect the real situation. Instead, the three groups seem to have co-existed for a long time, as demonstrated by the early attestation of hěn shǎo rén ‘few people’ as a QP in postverbal position in (64c) above from a 1953 article of the Rénmínrìbào ‘People's Daily’; similarly, (64b) dates back to 1987 in the same newspaper. What we observe evolving in time is the diffusion among the speakers of the analysis of hěn shǎo rén as QP, with a clear increase in the last decade.

The preceding discussion demonstrates that hěn shǎo rén ‘few people’ does not show the same distribution as hen duo rén ‘many people’ and can therefore not be considered as its counterpart in syntax, notwithstanding their antonymic relationship.

Turning now to the DP hěn shǎo de rén, with the AP hěn shǎo ‘very be.few’ as modifier of rén ‘person’, the corpus provides 16 examples from literary texts, 76 from newspapers and periodicals, and 108 from Weibo. Across these different text sorts, there are hardly any occurrences in subject position. The majority of cases occur in the object position of a verb modified by the adverb zhǐ ‘only’ (including many instances of the existential construction zhǐ yǒu ‘there is only’ (65)), thus confirming Liu's (Reference Liu2011) observation (see section 4.2 above). Among the 108 examples from Weibo (again with many sentences occurring twice), there are only five with hěn shǎo de rén ‘few people’ in subject position (66):

Finally, there is a variant of hěn shǎo de NP, where hěn shǎo does not modify a bare noun, but the Number Phrase ‘jǐ CL NP’ = ‘several NP’, as in (67) from Liu (Reference Liu2011):

Note that the relative order is rigid. Given that jǐ ‘several’ refers to a number between 3 and 9, in combination with hěn shǎo ‘very be.few’, this results in the meaning of ‘very few NP, a (mere) handful of NP’.

Hěn shǎo de jǐ ge NP ‘a handful of NP’ (including rén ‘person’ as NP) is fully acceptable for all speakers across the three groups and is a bona fide DP on par with hěn shǎo de NP ‘few people’; hence, it is acceptable in object position (see (67), (68)) and as complement of a preposition (see (69) elicited from a native speaker, there being no examples of this type in the corpus), modulo the required presence of the adverb zhǐ ‘only’ for some speakers.

As already observed for hěn shǎo de rén ‘few people’, the occurrence of hěn shǎo de jǐ CL NP ‘a mere handful of NP’ in subject position is relatively rare (70).

5. Back to the beginning: How to account for the observed contrasts

Having established that méi yǒu rén ‘there isn't anybody’, zhǐ yǒu DP ‘there is only DP’, and the subset of hěn shǎo rén that contains a covert existential verb yǒu ‘there are rarely people’ are existential constructions, not nominal projections, we can now explain the contrast between (1a) and (1b), repeated as (71a) and (71b) (my parsing, glosses, and translation):

As explained in section 1 above, cí zhí ‘resign job’ = ‘resign’ is a secondary predicate on rén ‘person’.Footnote ³⁵ Example (71a) is unacceptable, because wh-questions are banned from a secondary predicate when the matrix verb is negated or modified by a quantificational adverb (also see the non-felicitous English translation). In (71b), by contrast, wèishénme ‘why’ is in the matrix clause of the existential construction and acceptable; again, the same holds for English, as illustrated by the translation. (For hěn shǎo rén as QP, i.e., without a covert existential verb yǒu, see (78) below).

The unacceptability of (72) with shénme ‘what’ in object position confirms that (71a) is excluded by a general ban on wh-questions (adjunct and argument alike) in secondary predicates under a negated or quantified matrix existential verb:

Again, the English translation reflects rather well that the presence of a wh-phrase inside the secondary predicate is the source of the unacceptability, modulo the fact that a relative clause is used to translate the secondary predicate in Chinese. In fact, if at all, (72) can only be accepted as an echo question (also see appendix).

By contrast, a wh-phrase in sentence-initial topic position is fine, provided it is construed as Discourse-linked (see Pan Reference Pan2011b):

Concerning example (2) from Ko (Reference Ko2005: 886), repeated as (74) (with my parsing and glosses), it is ruled out by the simple fact that the sequence Zhāngsān shuō [(tā_i/tāmen_i) hěn cōngmíng] cannot serve as a secondary predicate for rén ‘people’ or Lǐsì.

In fact, (74) conflates several sources, each of which is responsible for the unacceptability, as shown by the comparison with the acceptable (75):

The subject of the secondary predicate must not be overt; accordingly, enclosing the pronouns in parentheses, as Ko (Reference Ko2005) does in (74), is completely misleading, as their presence or absence is relevant to the acceptability of the sentence:

Adding Zhāngsān shuō ‘Zhangsan said’ again amounts to an overt subject in the secondary predicate and leads to unacceptability:Footnote ³⁶

As a result, when both Zhāngsan shuō ‘Zhangsan said’ and the pronoun are present in the secondary predicate, the sentence is indeed completely garbled and hard to parse and interpret.

To conclude, the two ill-formed sentences (71a) (= (1)) and (74) (= (2)) can be straightforwardly explained by constraints observed for secondary predicates in general. Crucially, no (A-bar) movement nor intervention effects are involved here.

Let us finally turn to hěn shǎo rén ‘few people’ as a genuine QP in subject position and revisit the contrast in (1a–b), repeated as (78a–b) below:

As mentioned in footnote 5 above and presented in more detail in the appendix, Ko's (Reference Ko2005) proposal neglects the well-known fact that the default position for wèishenme ‘why’ is TP-internal, that is, to the right of the subject, and incorrectly stipulates SpecCP as the only position available. That is the reason why Soh's (Reference Soh2005) analysis is adopted here: covert feature movement of wèishénme ‘why’ to SpecCP in (78a) crosses the QP hěn shǎo rén ‘few people’ and induces an intervention effect, whereas this is not the case for overt movement of wèishénme ‘why’ to SpecCP in (78b).