1 Introduction

Minimizers are positive expressions that denote a minimal quantity or extent and that may reinforce negation. As illustrated in (1) and (2), English minimizing expressions like a soul and a word occurring in the scope of negation receive an interpretation that is roughly equivalent to that of a negative polarity item. Such an observation has motivated the definition of English minimizers as “partially stereotyped equivalents of any” (Bolinger 1972:121).Footnote 1

  1. (1)
    figure b
  1. (2)
    figure c

In addition, like polarity items (PIs) of the any-kind, English minimizers can occur not only in negative contexts, but also in non-negative ones, such as questions (both yes/no-questions and wh-questions) (Ladusaw 1979; Giannakidou 1998, 2011; Guerzoni 2004), conditionals (Giannakidou 1998, 2011), directive propositional attitudes (Giannakidou 2011), the scope of only (Giannakidou 2011) and complement clauses of emotive factive verbs (Giannakidou 2011). These different contexts are illustrated in (3)–(7) with examples from the literature.

  1. (3)
    figure d
  1. (4)

    If you give a damn, you’ll listen.

  1. (5)

    She’s still funny and cute and smart and I wish she gave a damn that we aren’t friends anymore. I miss Candice. www.xanga.com/betweenIDs

  1. (6)

    Only Mary gives a damn.

  1. (7)

    I’m glad he said a word. (Giannakidou 2011:1685)

In non-negative contexts such as (3)–(7), minimizers activate alternatives (i.e., they force the speaker to consider different options along a scale, with minimizers being at the low end) and give rise to an even-reading (cf. Pott 1857; Wagenaar 1930; Schmerling 1971; Horn 1972, 1989; Fauconnier 1975a, 1975b; Linebarger 1980; Heim 1984; Giannakidou 2007, 2011).Footnote 2 In the particular case of questions, (3), these are not neutral when they contain a minimizer: they are biased in the sense that a negative answer is expected from the speaker (Ladusaw 1979). As noted in Heim (1984) and Wilkinson (1996), this is the same effect that is obtained in a clause when even relates to the low endpoint of a pragmatic scale (Guerzoni 2003:99), as shown in (8), where subindexed f indicates the relevant endpoint of the scale that associates with the Focus particle even. Such an effect in questions lends plausibility to the analysis of minimizers as containing a tacit even particle (cf. Lahiri 1998; Lee and Horn 1994).

  1. (8)
    figure e

In negative contexts, negation denotes the presence of no quantity at all and, therefore, minimizers are seen as negative reinforcement (Horn 1989:400).Footnote 3 The idiomatic reading of the minimizer, however, is associated with sentential negation, defined as “[n]egation taking scope at least above (the existential quantifier binding the event argument of) the main predicate” (Penka 2007:11). This is the case not only when negation is expressed by means of the affix -n’t—which unambiguously expresses wide scope sentential negation—but also when negation is conveyed by the adverbial not, which can either express sentential or constituent negation, the latter being defined as negation taking scope over a particular constituent rather than the whole proposition (Zeijlstra 2004:47). In other words, a sentence like (9) is ambiguous between the reading in (10a) and the reading in (10b). In (10a) negation takes sentential scope, whereas in (10b), it only scopes over the constituent a word, hence inducing the interpretation of a word as a plain existential DP.

  1. (9)

    Mary said not a word.

  1. (10)
    figure f

In short, English minimizers clearly behave as PIs (see examples (3)–(7) above) in non-negative contexts and in negative contexts where negation merges high in the clause (see (1) and (2) above). In negative contexts, however, they become negative quantifiers (NQs) when negation is merged with the minimizer (as in (9)) and take sentential scope (hence yielding the (10a) reading). In this paper, I try to offer an analysis of English minimizers that can accommodate such facts.

The article is organised as follows. In Sect. 2, I discuss how minimizers fit into the definition of PI while also showing similarities with NQs, and apply Klima’s (1964) tests to negative sentences with minimizers to show that their idiomatic reading obtains with sentential negation but not with constituent negation. In Sect. 3, the literature on English NQs as decomposable lexical items is reviewed, while in Sect. 4, I deal with the internal structure of English minimizers showing that their syntactic behaviour supports the approach to NQs presented in Sect. 3. Finally, Sect. 5 concludes the paper.

2 The syntactic behaviour of English minimizers

2.1 The distribution of English minimizers: parallelisms with PIs and NQs

The examples in (11) and (12) show that the distribution of minimizers in negative contexts is similar not only to that of PIs like any and its derived forms, but also to that of NQs like no, nobody, nothing, etc. Notice that the difference between (11b) and (12b) is the position of negation (above the verb in the former vs. below the verb in the latter). In (12b), as happens with NQs in English, the negative operator (not) immediately preceding the minimizer can end up in a scope position and convey sentential negation.

  1. (11)
    figure g
  1. (12)
    figure h

Minimizers fit into Giannakidou’s (2006:464) definition of PI given in (13), as they are sensitive to some semantic property of the context, namely, to non-veridicality. Recall that, as has been shown in the previous section, English minimizers occur in non-assertive contexts with non-veridical operators such as interrogatives, conditionals, directive propositional attitudes, in the scope of only and with emotive factive verbs (see (3)–(7) above). They also occur in negative contexts with antiveridical operators (see (1), (9), (11b), (12b)), which constitute a subset of non-veridical operators.Footnote 4

  1. (13)
    figure i

In what follows, I use Vallduví’s (1994) diagnostic tests to further show that English minimizers behave like PIs both in negative and non-negative contexts, but also like NQs in one particular negative context. As will be seen, the crucial factors for NQ-like behaviour seem to be the presence of an adjacent Negative Phrase (NegP), Spelled-Out as not, and the presence of an even Focus Phrase (FocP) as part of the structure of the minimizer.

Although Vallduví’s (1994:270–280) four diagnostic tests aim at determining whether a lexical item is an n-word (Laka 1990) or not, in this paper I discuss how they can be applied to English PIs and NQs and then examine which results these tests yield for English minimizers. Before doing so, nonetheless, the term n-word needs to be defined in relation to the tests.

N-words, which are typically found in Negative Concord (NC) languages such as Italian, Catalan, Spanish, Modern Greek, Romanian and Hungarian, among others, are operator-dependent elements that display quantificational variability. N-words must be licensed by a semantically appropriate operator. In negative contexts, post-verbal n-words clearly behave like PIs, as they need to be c-commanded by a negative marker (or by another n-word in pre-verbal position), whereas pre-verbal n-words are closer to NQs in certain NC languages such as Spanish or Italian—known as Non-Strict NC languages (Giannakidou 1998, 1999, 2000)—where the sentential negative marker cannot co-occur with a pre-verbal n-word if a single negation or NC reading is intended. In Strict NC languages such as Modern Greek or Romanian, by contrast, the sentential negative marker co-occurs with n-words in all contexts. In both kinds of NC languages, n-words can occur in isolation as answers to questions, as shown in (14) and (15) for Spanish and Romanian, respectively.

  1. (14)
    figure j
  1. (15)
    figure k

According to Giannakidou’s (2006:328) definition of n-word, formalised in (16), the ability to occur as answers to questions is a defining property of n-words.

  1. (16)
    figure l

Actually, the characteristic behaviour of n-words stated in (16b) is what sets them apart from other PIs such as the Spanish post-nominal indefinite persona alguna lit. person some ‘anybody’, or the Romanian expression cine știe ce lit. who knows what ‘much’. This is illustrated in (17A) and (18A) for Spanish and Romanian, respectively.

  1. (17)
    figure m
  1. (18)
    figure n

N-words such as Italian nessuno ‘anybody’/‘nobody’, Catalan cap ‘any’/‘no’ or Spanish nada ‘anything’/‘nothing’, successfully go through Vallduví’s tests in (19a–c), whereas other PIs like Catalan gaire ‘many’/‘much’ and the Spanish post-nominal indefinite N alguno lit. N some ‘any N(oun)’ do not. This has been shown in (14), (15), (17) and (18) above for the test in (19a), and is illustrated in (20) and (21) for tests (19b, c) with data from Catalan and Spanish.Footnote 5

  1. (19)
    figure o
  1. (20)
    figure p
  1. (21)
    figure q

Whereas the tests in (19a–c) show that n-words may behave as NQs, the test in (19d) aims at establishing whether they are inherently negative or not. Vallduví (1994) takes this test from Laka (1990, 1993), who holds the view that n-words are PIs and, hence, non-negative.Footnote 6 In (22), the data from Catalan show that, unlike what was the case for (19a–c), both n-words and PIs successfully go through the test in (19d).Footnote 7

  1. (22)
    figure r

As is well known, Standard English does not have NC as part of its grammar, as negation cannot be expressed more than once in the same clause without inducing a Double Negation reading. Hence, if English (at least the Standard variety) does not have the possibility of deriving NC structures, it cannot be assumed to have n-words, either. English has a set of non-assertive indefinites with any- (referred to as PIs in this paper), and a set of negative indefinites with no- (referred to as NQs in this paper) (Quirk et al. 1973:376). Since PIs and NQs occur in the contexts where n-words occur in NC languages, Vallduví’s (1994) tests can be applied to English to gain some insights into the following question:

  1. (23)

    Are minimizers best characterized as PIs, as NQs, as both, or as neither?

In order to answer the question in (23), Vallduví’s (1994) tests have to be applied not only to minimizers, but also to any-PIs and to no-NQs so that the behaviour of the three kinds of indefinite expressions can be compared. Let us start with the test in (19a), which assesses the ability of a given word to occur in isolation.

  1. (24)
    figure s

As shown in (24B), NQs can occur in isolation as answers to a question, whereas this is not possible for PIs, (24B′). Likewise, NQs can be modified by almost or absolutely, whereas PIs cannot. This is illustrated in (25).

  1. (25)
    figure t

Notice, though, that if the PI occurs in a negative context, as in (26), almost/absolutely modification is possible.

  1. (26)

    She almost/absolutely didn’t eat anything for dinner.

With respect to the test in (19c), namely grammaticality in pre-verbal position, only NQs successfully go through it. As shown in (27b), PIs are not grammatical if they are not bound by a semantically relevant operator (Giannakidou 2006:464). Hence, they are excluded from a pre-verbal position.

  1. (27)
    figure u

As far as the test in (19d) is concerned, NQs fail it: they can occur in yes/no-questions and conditional contexts, but the meaning they convey is necessarily a negative one, which shows that they are somehow negative words. This is exemplified below, where (28a) is interpreted as questioning whether it is true that the addressee did not have any Erasmus student in his/her class rather than whether s/he had any Erasmus student in the class, and (28b) is pragmatically very odd, as it is not usual for people to call the police if it is not the case that someone is trying to break into a house. The opposite situation totally justifies such a call, though.

  1. (28)
    figure v

If the examples in (29) are construed with PIs, by contrast, they are pragmatically felicitous and interpreted as conveying a non-negative meaning.

  1. (29)
    figure w

Let us see now what happens when Vallduví’s (1994) tests are applied to English minimizers. As can be seen in (30a)–(34a), it turns out that if the negative adverb not is adjacent to the minimizer, this aligns with NQs when going through the tests. Conversely, when an adjacent not is absent, as in (30b)–(34b) minimizers align with PIs (see (24)–(29)).Footnote 8

  1. (30)
    figure x
  1. (31)
    figure y
  1. (32)
    figure z
  1. (33)
    figure aa
  1. (34)
    figure ab

Minimizers also align with PIs under affix -n’t negation for the tests in (19b, d), as shown in (35)–(37), which correspond to the examples in (31a), (33a), and (34a) now with affix negation. Notice that if (33b) and (34b) contained the negative affix, the reading could not be non-negative, as required by the test in (19d).

  1. (35)

    She almost/absolutely didn’t say a word.

  1. (36)

    Did*(n’t) she lift a finger to help you?

  1. (37)

    If you *(don’t) give a damn, call her tonight.

To summarise, Table 1 displays the results of Vallduví’s (1994) tests for PIs, NQs and minimizers in English.

Table 1 Results of Vallduví’s (1994) tests for PIs, NQs and minimizers in Standard English
Full size table

As can be seen in Table 1, NQs and PIs show the right opposite behaviour when going through the tests in (19). While minimizers unsurprisingly align with PIs (for, as has been shown earlier, they can occur in non-veridical contexts), they display NQ-like behaviour if negation (in the form of not) is directly adjacent to them.

The (obvious) question is why this is so, and I try to answer it by arguing that English minimizers share properties of both PIs and NQs because these two kinds of words are structurally related in a way that is also relevant for minimizers. In line with Klima (1964), Jacobs (1980), and, most recently, Penka (2011), Iatridou and Sichel (2011), Zeijlstra (2011) and Temmerman (2012), I argue that NQs are decomposable syntactic objects consisting of a NegP, not, and a PI. The merging possibilities of not (below or above the verb) play a crucial role in the quantificational-like behaviour of minimizers: being a scope-taking operator, negation merged in a low structural position (i.e., with the PI, below the verb) may induce Q(uantifier) R(aising) of the whole syntactic object to an IP-adjoined position from where negation takes scope over the whole proposition. If, by contrast, negation is merged higher in the structure (i.e., above the verb) and independently from the indefinite part (i.e., the PI/minimizer), minimizers display polar behaviour, and the negative indefinite surfaces discontinuously, namely as sentential negation binding a PI/minimizer.

The merging possibilities of NegP, however, cannot be entirely responsible for the NQ-like behaviour of the structure not + minimizer since, as shown in (38), not every indefinite occurring with an adjacent not can undergo QR. While (38a) is equivalent to (2b) (repeated here as (39a)), (38b) is equivalent to (39c), rather than to (39b).

  1. (38)
    figure ac
  1. (39)
    figure ad

In Sect. 4, I show that English minimizers contain a Focus particle even which, when merged with NegP, allows minimizers to QR to a position where the negation they are associated with can take sentential scope.

2.2 Sentential negation and the idiomatic reading of English minimizers

That the idiomatic reading of English minimizers requires sentential negation to obtain (see examples (9) and (10) above) is confirmed by the results of Klima’s (1964) tests in (40)–(43).

  1. (40)
    figure ae
  1. (41)
    figure af
  1. (42)
    figure ag
  1. (43)
    figure ah

As can be seen in (40a, b)–(43a, b), Klima’s tests do not yield different results for the structure -n’t + verb + minimizer and verb + not + minimizer. The (c) examples, by contrast, show that when negation does not take sentential scope, the reading for the minimizer is not the idiomatic one, but one which corresponds to a plain indefinite.

According to the first of Klima’s tests (1964:262–265), sentential negation takes a positive question tag, whereas if negation is non-sentential (i.e., constituent), the tag is negative. In (40a, b), the tag is positive both with sentential negation being expressed by means of -n’t and by means of not, but it has to be negative in (40c), where negation only has scope over the constituent a word.

The second of Klima’s tests, illustrated in (41) and (42), predicts that a sentence with sentential negation can be coordinated with a negative clause with either, or be followed by a pro-form with neither. By contrast, this is not possible with a sentence where negation is non-sentential. Notice that, as shown in (41a, b) and (42a, b), the same polarity continuation test with (n)either yields the same results both for -n’t and sentential negation not. Conversely, opposite results are obtained when the test is applied to an example with constituent negation, as shown by (41c) and (42c).

The third test, illustrated in (43), shows that a sentence containing sentential negation (either expressed by -n’t or by not) can be continued with a phrase or clause headed by not even, as in (43a, b). This is not possible if negation has scope only over a constituent, as in (43c).Footnote 9

A fourth test consisting in replying to a previous utterance with another utterance that contains an expression of agreement with so or not, as exemplified in (44), also allows us to conclude that the idiomatic reading is associated with sentential negation when minimizers occur in negative contexts.

  1. (44)
    figure ai

As can be seen in the clauses in (44a.B) and (44b.B), a not clause expresses agreement with a clause containing an instance of sentential negation and a minimizer (i.e., the clauses in (44a.A) and (44b.A) in this case), whereas this is not possible with an affirmative clause, as in (44c), or a clause containing a plain indefinite with negation scoping over it, as in (44d). Since, as shown in (44a.B) and (44b.B), speaker B can use a not clause to agree with speaker A’s utterance, which contains a minimizer, we can conclude that (44a.A) and (44b.A) involve an instance of sentential negation. Again, notice that this is the case regardless of whether negation is expressed by means of the affix -n’t or the negative adverb not. I return to this issue in Sect. 4.

In this paper, I analyse English minimizers as PIs with an in-built Focus even particle that can associate with negation. Hence, the varying (and otherwise puzzling) behaviour of English minimizers is integrated into an account of negative indefinites as decomposable expressions that combine a NegP and an existential quantifier, which is presented in Sect. 3.

3 Negative quantifiers as decomposable lexical items

Several scholars have argued in favour of an analysis of English NQs as decomposable into a negative component and an existential quantifier (Klima 1964; Jacobs 1980; Ladusaw 1992; Rullmann 1995; Larson et al. 1997; Sauerland 2000; Penka 2007, 2011; Penka and Zeijlstra 2010; Iatridou and Sichel 2011; Temmerman 2012; among others). That these two parts form a single lexical item has been postulated to be the result of a PF-operation (Klima 1964; Jacobs 1980; Rullmann 1995; Iatridou and Sichel 2011; Zeijlstra 2011; Temmerman 2012).

The existence of Split Scope readings in English (Potts 2000; Penka 2007, 2011; Zeijlstra 2011) is evidence in favour of the assumption that the negative meaning associated with NQs must be ‘detachable’ from the quantifier.Footnote 10 In the presence of operators that take scope such as modal verbs or object intensional verbs, NQs may not behave as is expected from quantifiers. It is well known that quantifiers can take wide or narrow scope. In a sentence like (45), the reading in (45a) corresponds to the quantifier taking wide scope with respect to the modal verb (a deontic may), while in (45b) the quantifier has taken narrow scope with respect to the modal (an epistemic may).

  1. (45)
    figure aj

In some occasions, however, wide scope and narrow scope of the NQ with respect to the modal verb do not yield the correct reading. This is the case in (46), where the most prominent reading is (46a), with negation scoping over the modal verb and the existential quantifier scoping under it (i.e., a Split Scope reading). The wide scope reading is illustrated in (46b). As shown in (46c), the narrow scope reading is not possible at all.

  1. (46)
    figure ak

In (47), where the sentences contain a post-copular NQ subject in a there-construction with an epistemic modal, only the Split Scope reading is possible, as shown in (47a, b). This is unexpected, as it is the case that post-copular subjects in there-constructions with modals necessarily take narrow scope with respect to the modal, as shown in (48).

  1. (47)
    figure al
  1. (48)

    There must be someone in John’s house. must >∃, ∗∃> must (Penka 2007:91)

Given that Split Scope readings cast doubt on the analysis of NQs as lexical units, I adopt the view, in line with Temmerman (2012:86), that English NQs contain a negative operator, not, which enters the derivation as an independent lexical item.Footnote 11 The structure of a NQ in English is assumed to be the one in (49).

  1. (49)

PIs would have the structure in (50), which allows the sentences in (51) to be morphologically related.

  1. (50)
  1. (51)
    figure am

The NQ in (51a) is analysed as a PI (see (50)) that merges with NegP, as in (49). That is, NQs are not stored as such in the Lexicon. Rather, they are complex objects built in the syntax. Neg°, which is part of the Numeration of a negative sentence, can merge with the DP, as in (51a), or in a position above the vP, as in (51b, c).

Following Matushansky (2006), I assume the clausal structure of English to be one where negation is in the Specifier of AuxP, a functional head that is sandwiched between the vP and the TP (Matushansky 2006:92). In Matushansky’s proposal, negation is both a head and a maximal projection (Neg°/NegP) and, hence, it may undergo m-merge with Aux° (Matushansky 2006:81).Footnote 12 This results in the creation of a new head Aux°, which can raise to T°.

In (49) and (51a), NegP and the rest of the DP amalgamate at PF, thus resulting in what is commonly known as a NQ (cf. Temmerman 2012). In (51b, c), NegP is in Spec, AuxP and can optionally m-merge with it. After m-merger, Neg°/NegP is Spelled-Out as -n’t, as in (51b). If m-merger with Aux° does not take place, NegP is Spelled-Out as not, as illustrated in (51c).

4 The internal structure and syntax of English minimizers

In line with Linebarger (1980), Heim (1984), Horn (1989), Giannakidou (2007) and Kuno (2008), I assume that English minimizers are associated with the Focus particle even.Footnote 13 Given that even is an adverb (Quirk et al. 1985; König 1991), I follow Cinque (1999) in assuming that it occupies a Specifier position. Hence, the syntactic structure for the Focus particle associated with English minimizers is the one in (52), with Foc° being phonologically empty and Spec, FocP optionally occupied by overt even.

  1. (52)

In addition, given that one is possible in most minimizers (e.g. a/one red cent, a/one soul, sleep a/one wink, budge a/one inch, a/one bit, etc.), I take the relevant functional projection not to be a DP, but a Num(eral) P(hrase), the head of which can be Spelled-Out as either a or one.Footnote 14

  1. (53)

As in the case of NQs, the structure in (53) is built in the syntax and can be associated with a negative operator, Neg°. As shown in (54), even (when overt) is Spelled-Out lower than negation and higher than NumP.

  1. (54)

    I said not (even) a/one word at the meeting.

Therefore, NegP is assumed to merge with FocP, as represented in (55).

  1. (55)

As has been shown in Sect. 2.2, English minimizers successfully go through Klima’s (1964) tests for sentential negation not only when negation is merged above the vP, as in (56a, b), but also whenever it is merged below, as in (57).

  1. (56)
    figure an
  1. (57)

    I said not (even) a/one word at the meeting.

This is due to the fact that, as has been claimed to be the case for NQs (Zeijlstra 2011), the merger of negation and the minimizer in (57) undergoes QR to an IP-adjoined position and is interpreted as having wide scope at LF. PF Spells-Out the lower copy. The syntax of the structures in (56a, b) is discussed later in the paper in connection with the distribution that even shows with respect to negation and NumP.

As discussed in Rooth (1985), focused constituents may undergo QR. In the case of English minimizers, they contain FocP (with an overt or covert even) and, hence, they can undergo QR when merged with not. Their doing so allows negation to take sentential scope, which, as has been discussed earlier in the paper, results in the idiomatic reading of the minimizer. This is further illustrated in (58), where the overt presence of the Focus particle even in (57) excludes the non-idiomatic reading of the minimizer under constituent negation.

  1. (58)
    figure ao

But why do English minimizers require sentential negation to be interpreted with an idiomatic reading (i.e., at the low endpoint of a scale) rather than as a plain indefinite (e.g. (10b))? At this point, constituent negation needs to be explored more in depth, especially with regards to the not X but Y configuration that has been used in all the examples of constituent negation included in this paper. McCawley (1991:189) refers to such a configuration as contrastive negation, as negation “contrasts two ways of filling a syntactic position, one that (according to what the sentence says) results in a false proposition and one that results in a true proposition,”Footnote 15 and argues that it involves coordinating two different sentences.Footnote 16 That is, a sentence such as (59) would be interpreted as in (60) if a word were a plain indefinite.

  1. (59)

    I said not a word, but a sentence.

  1. (60)

    I didn’t say a word, but I said a sentence.

Notice, however, that if not a word in (59) were a minimizer (with a covert even FocP as part of its structure), it would undergo QR to an IP-adjoined position, thus yielding the interpretation in (61) for the first focus of the contrastive negation (namely the not X). As shown in (62), this would rule out the second focus of the contrastive negation on pragmatic grounds: if the speaker said nothing at all, s/he cannot have said a sentence.Footnote 17

  1. (61)

    I didn’t say anything at all.

  1. (62)
    figure ar

As shown in (58) above, the presence of an overt even excludes the possibility that a/one word is interpreted as a plain indefinite as in (60) and, hence, contrastive negation is not felicitous.

Let us now turn to discussing how it is possible to claim that the three sentences in (58), all containing a minimizer, are structurally related to one another. As advanced earlier in the paper, I will connect this fact to the relation that also exists between negative sentences containing NQs and discontinuous not/-n’ t …PI configurations, as in (63).

  1. (63)
    figure as

As illustrated by the tree in (55) above, I have assumed NegP to merge with FocP. What I want to show now is that while the ingredients of English minimizers are NumP (a/one + NP), FocP (with even in its Specifier) and negation, the merging possibilities of FocP, on the one hand, and of NegP on the other, crucially result in a complex distribution of the three different parts of a minimizer.Footnote 18

Let us start by substantiating the idea that FocP does not have to be adjacent to NumP when a minimizer is syntactically assembled. Consider, for example, the data in (64), where the minimizer not even a/one soul occurs as part of a prepositional complement.

  1. (64)
    figure at

As can be seen in (64a, b), the negative marker not and the Focus particle even can occupy different positions with respect to the preposition: either FocP is selected by P°, as in (64a), or FocP merges with the PP, as in (64b). The relevant syntactic structures for (64a, b) are given in (65a, b).

  1. (65)
    figure au

In (66), the verb is inflected for Tense, as negation is below the vP. The sentences in (64), however, can be paraphrased with the sentences in (66), where negation is first-merged higher up in the clause, namely in a position above the vP.

  1. (66)
    figure av

I assume, following Matushansky (2006), that sentential negation is Spelled-Out as an affix because Neg°/NegP sits in Spec, AuxP, from where it can m-merge with Aux°. That is, whereas the negative component of an English minimizer can merge with FocP, as in (65a, b), it does not necessarily have to do so. Not can also first-merge in Spec, AuxP, thus preventing QR of the minimizer to an IP-adjoined position.

Flexible merging possibilities of not are also attested with other indefinites, in particular, with NQs of the no-type and PIs of the any-type.Footnote 19 While in the former NegP merges with the existential DP (a(ny) body, a(ny) thing, etc.), which then undergoes QR at LF (Zeijlstra 2011) and Fusion Under Adjacency at PF (Temmerman 2012), in the latter NegP merges in the IP domain. The result is the pair in (67).

  1. (67)
    figure aw

In the particular case of English minimizers, the pairs in (68) and (69) crucially diverge with respect to whether NegP is first-merged with FocP, as in (68a, b), or if, by contrast, it is first-merged in Spec, AuxP, as in (69a, b). When it is merged in Spec, AuxP, NegP can undergo m-merger with Aux° (as in (69a, b)), or not (as in (70a, b)).

  1. (68)
    figure ax
  1. (69)
    figure ay
  1. (70)
    figure az

To summarise, in this section I have shown that while NumP, FocP and NegP are the main ingredients of English minimizers, the merging possibilities of two of these ingredients need to be taken into account when accounting for their syntax. Firstly, FocP, the projection headed by a covert Focus particle (with even optionally occupying its Specifier) that is associated to minimizers, may or may not be adjacent to NumP. Secondly, NegP can merge either with FocP below the vP or in {Spec, AuxP} above the vP. In the latter case, in addition, m-merger of Neg°/NegP with Aux° is optionally possible. As shown in Table 2, these various structural factors combine in complex ways to result in the (otherwise puzzling) syntax of English minimizers.

Table 2 The syntax of English minimizers as the product of the merging possibilities of FocP and NegP, and (lack of) m-merger of NegP in Spec, AuxP
Full size table

5 Conclusion

In this paper I have shown that a parallelism exists not only between minimizers and PIs in English, as has been discussed in the literature, but also between minimizers and NQs. This results in a somewhat puzzling syntactic behaviour that is difficult to explain in accounts that do not envisage any kind of structural and derivational relationship between PIs and NQs in English. However, as I discuss in the paper, the distribution of English minimizers in negative contexts fits well into an analysis of English NQs as PIs that are associated with a NegP that has different Spell-Out possibilities: it may amalgamate with an existential quantifier at PF resulting in a NQ of the no-kind, or it can be first-merged in the IP domain, above the vP, hence being Spelled-Out as not, from where it can optionally m-merge with an Aux° Spelling-Out as the affix -n’t.

Integrating minimizers in negative contexts into a view of English NQs as syntactic objects that can be decomposed into several functional projections not only lends further support to the view of negative indefinites as compositionally-built expressions, but it also allows certain apparent idiosyncrasies of English minimizers and minimizing expressions to be given a principled explanation. Different functional heads compose the syntax and the meaning of minimizers and, therefore, the merging possibilities and restrictions of these heads interact in different ways, resulting in a complex syntax of minimizers.

In this paper, English minimizers have been argued to contain three different ingredients that contribute to their syntax and semantics. These are a NumP (e.g. a/one word, a finger, a damn, etc.), a Focus particle (with even optionally sitting in Spec, FocP), and NegP. Both NumP and FocP merge in a position lower than the vP, but they do not have to be necessarily adjacent.

In addition, NegP can merge with FocP, below the vP, or in the Specifier of AuxP, above the vP. In the former case, NegP Spells-Out as not, whereas in the latter case, it can Spell-Out either as not or as the affix -n’t if it m-merges with the head Aux° and then the newly formed head moves to T°. One of the interesting properties of English minimizers is that, regardless of where negation merges, it needs to take sentential scope for the minimizer reading to obtain. I showed that this is the case by applying Klima’s (1964) tests. I have claimed that, as happens with NQs in English, when negation is Spelled-Out lower than the vP, minimizers (which contain a FocP) undergo QR to an IP-adjoined position. If negation is merged above the vP, no movement operation is required for negation to be interpreted as sentential and QR is precluded.

As is the case for English PIs, minimizers can occur in negative contexts, but also in non-negative ones. In this paper I have claimed that this is because negation is a potential component of a minimizer, but not an obligatory one. The Focus particle, by contrast, is always part of the structure of a minimizer and is responsible for its scalar behaviour. Syntactically, it blocks the possibility that, as happens with NQs, NegP and NumP undergo Fusion at PF resulting in one single lexical item, and makes it possible for the minimizer to QR when NegP is merged with it. The optional presence of the overt adverb even signals the position of FocP in the syntactic structure.

It has been shown that when English minimizers occur with a verb inflected for Tense (i.e., without do-support) and they are preceded by the negative operator not, they align with NQs. In all other cases (i.e., when negation co-occurs with do-support, or when the minimizer occurs in a non-negative context), English minimizers align with PIs (which is not surprising, as minimizers meet all the criteria to be considered PIs). This behaviour can be given a principled explanation within an account that allows us to relate PIs and NQs structurally, with the presence of NegP and FocP in a vP-internal position being what determines the NQ-like behaviour. This new way of looking at English minimizers advances our understanding of their syntax, also supporting the idea that English is best analysed as having negative indefinites with polar properties that may optionally associate with negation rather than non-decomposable NQs that are semantically negative.