English 'One' and 'Ones' as Complex Determiners (2015)

“English One and Ones as Complex Determiners” Richard S. Kayne New York University June, 2015 1. Introduction. 2. One is not a noun. 3. One is a determiner. 4. One and a/an 5. One is a complex determiner containing a classifier 6. English ones, Spanish unos and French uns 7. Defusing the objection. Part II 8. Derivations 9. A further restriction on ones 10. The licensing role of adjectives 11. Non-restrictive adjectives. 12. Numeral one 13. Ordinals 14. One(s) and demonstratives. 15. Singular vs. plural. 16. Possessors. 17. Human one 18. The 19. Just singular one. 20. The -s of ones. 21. Restrictions on what follows one(s). 22. Other determiners that look like pronouns. 23. Conclusion. 1. Introduction. Perlmutter (1970) took prenominal one to be the same element in both of the following: (1) John has written only one paper this year. (2) Mary has just written one hell of a paper. despite the fact that the numeral interpretation perceived in (1) seems to be absent in (2). Other examples of a similarly non-numeral pre-N one are found in: (3) There’s one John Smithfield here to see you. (4) One day, he’ll realize that we were right. In support of Perlmutter’s unified approach to these two types of one is the fact that all are equally incompatible with plural nouns: (5) *He’s written only one papers this year. (6) *She’s just written one hell of papers. (7) *There’s one John Smithfields here to see you. (8) *One days, he’ll realize that we were right. Kayne 2 In this paper I will attempt to extend a unified approach to one to encompass, in addition, the one of: (9) I have a red car and you have a blue one. despite the fact that this one is compatible with a plural: (10) I have red cars and you have blue ones. That this attempt has initial plausibility comes from a point made by Llombart-Huesca (2002, 60), to the effect that the one of (9) and (10) shares with the one of (1)-(4) an incompatibility with mass nouns. Parallel to (1)-(4) we do not have: (11) *He lost only one blood in the accident. (12) *You’ve just had one hell of fun. (13) *There’s one Domino Sugar all over the table. (14) *One money, (and) you’ll succeed Nor, parallel to (9) or (10) do we have: (15) *I like red wine and you like white one. (16) *She’s had good luck but he’s had bad one. The fact that plural ones is possible in (10) will turn out, as I will try to show, not to be incompatible with taking the one of (9), as well as the one of (10), to be essentially the same as the one of (1)-(4). 2. One is not a noun. The one of (1)-(4) looks like a determiner of some sort. But the one of (9) and (10), especially insofar as it is post-adjectival, looks at first glance like a noun, and in that sense looks quite unlike determiner one. If the one of (9) and (10) were really a noun, though, it would have to be recognized as an extremely odd one, since, unlike ordinary nouns, it cannot be a bare plural:1 (17) *I have cars and you have ones, too. Similarly, there is to a large extent no completely bare a one:2 (18) *I have a car and you have a one, too. which would be surprising if one were a noun. In addition, as noted by LlombartHuesca (2002, 61), one cannot be immediately preceded by a numeral in sentences like:3 ———————————— 1 Cf. Stirling and Huddleston (2002, 1515) and Payne et al. (2013, 798, 812). Payne et al (2013, 812) suggest that bare *ones is “preempted by ...some”. This does not seem descriptively correct (quite apart from the absence of a clear notion of preemption). Consider, for example: i) Bicycles have wheels and unicycles have wheels/*ones, too. Here, ones is bad, as usual, but some is inappropriate: ii) ?...and unicycles have some, too with (ii) rather having the status of: iii) ?...and unicycles have some wheels, too. 2 In standard English as opposed to the dialects discussed in McDonald and Beal (1987, 48), Beal et al. (2012, 57). Stirling and Huddleston (2002, 1513n) give You’re a one! as idiomatic. (It would be of interest if these exceptions had no counterpart with plural ones.) The general impossibility of bare *a one was noted by Perlmutter (1970, 236). 3 As noted by Perlmutter (1970, 236) and Lakoff (1970, 630). Halliday and Hasan (1976, 97) say that one hears two ones especially in children’s speech; this needs to be looked into. As does the fact Kayne 3 (19) *You have three cars but I only have two ones. in which respect one is again behaving in an un-noun-like fashion. The conclusion must be, in agreement with Llombart-Huesca (2002, 62),4 that the one of (9) and (10) is not a noun. 3. One is a determiner. If one is never a noun, then a unified approach to all instances of one will lead to the conclusion that one is in all cases a determiner, just as it is in (1)-(4). In which case the phrase a blue one in (9) must contain two determiners. Furthermore in: (20) We have only one blue one. one blue one must contain two determiners that are identical in form. This conclusion, to the effect that English allows two (sometimes identical) indefinite determiners in what looks like one DP is less surprising than it might appear, given the existence in some English (e.g. mine) of:5 (21) It’ll take us a half a day to finish that job. in which a half a day, with two identical determiners, is perfectly natural. Similarly, Wood (2002) had noted, for some English (in this case, not mine):6 (22) a such a wonderful book 4. One and a/an Of course, there is a discrepancy between (21) and (22), which contain two instances of the indefinite article a/an, and (20), which contains two instances of one. But this discrepancy is arguably a relatively minor one, in particular if interpreted against the background of Perlmutter (1970) and Barbiers (2005; 2007), both of whom argue in favor of a close relation between prenominal one and the indefinite article. Perlmutter (p. 234) more specifically took English to have, as a source for the indefinite article, “a rule which obligatorily converts unstressed proclitic one to an”.7 that a Google search yields a number of examples with completely bare ones, which may point to the existence of an as yet unstudied variety of English. 4 Llombart-Huesca’s arguments were not taken into account by Payne et al. (2013). Cf. also the multiple definite articles of Greek, as discussed by Alexiadou and Wilder (1998) among others. 6 Schibsbye (1970, 285) had noted a half a dozen eggs. Probably also belonging here are a helluva good show and a gem of a film. On other Germanic languages, see, for example, Barbiers (2005, 170) and Wood (2013) and references cited there. 7 Left open by this emphasis on phonology is the fact that English sometimes allows a stressed indefinite article, as in: i) I can’t give you the book, but I can give you a book. in which a rhymes with say. This stressed a does not license NP-ellipsis: ii) *...but I can give you a. suggesting that Borer’s (2005, 111n) primarily phonological account of the impossibility of (ii) with unstressed a is not general enough, 5 Kayne 4 Perlmutter’s formulation/rule was not immediately able, as he himself noted, to account for generic-like a/an, given the absence of a comparable (stressed) generic prenominal one that would be its source:8 (23) A spider has eight legs and many eyes. (24) One spider has eight legs and many eyes. The generic-like reading of (23) does not seem to carry over to (24). The rule that Perlmutter suggested was meant to treat pairs like: (25) That was a hell of a paper. (26) That was one hell of a paper. as involving, respectively, an unstressed and a (somewhat) stressed variant of the same element a/one, with the same interpretation. As just noted, the kind of pairing that holds for (25) and (26) does not hold for (23) and (24). In part similarly, the intended pairing breaks down for: (27) too long a book which has no counterpart with one: (28) *too long one book A third such problem for Perlmutter’s conversion rule lies in: (29) a few books (30) *one few books where, again, the indefinite article has no one counterpart to serve as a plausible source. A fourth problem for the pairing of a and one can be seen in: (31) They’re selling one-drawer desks in the back of the store. (32) *They’re selling a-drawer desks in the back of the store. in which, this time, prenominal one is possible, but cannot be replaced by a/an. Despite these several discrepancies between one and a/an, I will, in partial agreement with both Perlmutter and Barbiers, take there to be a significant relation between a/an and one, to be spelled out in the next section. 5. One is a complex determiner containing a classifier Let me execute the idea that a/an is a reduced form of one in a different way from Perlmutter (and Barbiers). Let me start from generic-like (23) and (24) and in particular from (23) vs. (24) being reminiscent of a fact from Chinese. According to Cheng and Sybesma (1999, 533-534; 2012, 640), a singular classifier in Chinese cannot occur within a generic DP (whether or not yi (‘a/an/one’) is present).9 This leads me to think that one cannot occur in (24) with a generic-like reading for the same reason that singular classifiers are excluded from Chinese generic DPs, which in turn leads to the following proposal: (33) An English DP with one contains a singular classifier. ———————————— 8 He suggests generic a might perhaps derive from any one, but note: i) Any/*A spider whatsoever would be able to eat that insect. ii) Hardly any/*a spider would eat that insect iii) Not just any/*a spider could have done that. 9 Cf. Simpson et al. (2011, 188) on Vietnamese; also Simpson and Biswas (2015, 7) on Bangla. Kayne 5 (Conversely, an English DP with a/an can (perhaps must) lack a classifier.) That one is associated with a singular classifier, while a/an is not, is close to Perlmutter’s idea that a/an is a ‘reduced form’ of one, though by expressing the notion of ‘reduction’ in terms of the more specific notion of the presence vs. absence of a classifier, we can formulate an account of (23) vs. (24) that Perlmutter’s less specific proposal was unable to do. More precisely put, the phrase one spider in (24) must, by (33), contain a singular classifier. But, judging from Chinese, singular classifiers are plausibly incompatible with generic readings. Therefore, (24) cannot be a generic type of sentence in the way that (23) can be. In Cardinaletti and Starke’s (1999) terms, we might want to go further and relate the fact that one is associated with extra syntactic material (the singular classifier) to the fact that one is (arguably) morphophonologically ‘bigger’ than a/an. We can do this as follows. One is to be understood as bimorphemic and in particular as ‘w- + an’, where w- (as I will write it) is the classifier and an the indefinite article.10 The difference in vowel quality between one and an might be due to independent properties of English phonology, perhaps involving (in part) stress. The necessary prononciation of the n of one even before a consonant, as opposed to the necessary dropping of the n of an before a consonant, might again just be phonology. Or it might also be related to syntax, especially if the order ‘classifier - indefinite article’ (‘w- + an; cf. Ghosh (2001, chap.3) on some Tibeto-Burman having ‘CLF Num N’ order) is produced by leftward movement from a structure in which the indefinite article precedes the classifier.11 From this perspective, the additional contrasts (beyond the generic one) mentioned earlier between one and a/an look as follows. The contrast in: (34) We have a/*one few days left. can be attributed to a clash between the classifier w- that is part of one and the silent noun NUMBER (capitalization will indicate silence) that accompanies few,12 in a way that is parallel to: (35) We have (only) a/*one small number of days left. as well as to: (36) Mary has written a/*one number of papers this year. In all of (34)-(36), number/NUMBER is not allowed to cooccur with the classifier contained in one. In the variants of (34)-(36) with a, there is no comparable classifer, just the indefinite article, and so no clash. (The clash in question may in turn be related to the classifier-like status of number/NUMBER itself in these sentences - cf. Liao (2015).) As for: (37) too long (of) a/*one book it looks like the classifier that is part of one blocks the preposing of the degree phrase (I return to (37) below). Finally, the restriction seen in: (38) They’re selling one-drawer/*a-drawer desks in the back of the store. ———————————— 10 An alternative that I will not pursue might be to take one to be monomorphemic and to cooccur with a silent classifier. 11 Cf. Leu (2015, 116) on German ein being moved across. 12 Cf. Kayne (2002a; 2005a). Kayne 6 may be linked to: (39) They’re real Brooklyn-lovers. (40) They’re real (*the) Bronx-lovers. via a prohibition against bare articles appearing within compound-like structures, with the classifier contained in one protecting it, in a way that remains to be spelled out, from this prohibition.13 6. English ones, Spanish unos and French uns If one is a complex determiner (containing two subparts, namely a classifier and an indefinite article),14 then ones in examples like: (41) They have blue ones. must be an even more complex determiner with (at least) three subparts, namely a classifer, an indefinite article and plural -s. An immediate objection might be that ones cannot be followed by an overt noun, as seen in: (42) *They have blue ones cars. unlike more familiar determiners. This objection to the determiner status of ones is weaker than it looks, for two kinds of reasons. The first has to do with the fact that Spanish allows sentences like: (43) Yo tengo unos libros. (‘I have some/a few books’) in which the noun libros is preceded by a determiner unos that resembles English ones.15 Both unos and ones contain a plural -s. In addition uno (or un or una) is the Spanish counterpart of English numeral one and often of the English indefinite article a/an, which we saw earlier to be closely related to one. Without saying that unos and ones are identical in composition (whether unos (sometimes) contains a classifier is unclear), the similarity between unos and ones, combined with the fact that unos is followed by an overt noun in (43), shows that the language faculty does not systematically frown on determiners of the ones type. Of course we would also like to understand why unos and ones differ in certain ways, e.g. in (42) vs. (43) with respect to whether they can be followed by an overt N. A possibility that comes to mind is that (42) is excluded in English for the same reason (whatever it is) that an adjective cannot be followed by plural -s if it is also followed by an overt N: (44) They have other(*s) cars. If so, then we have at the same time an account of the contrast between (41) vs. (42), which now reduces to the contrast between (44) and (45). (45) They have other*(s). ———————————— 13 Why one acts differently here from demonstratives remains to be understood. Relevant to the formulation of the prohibition in question is: i) two (beautiful) (*the) seventh inning home runs vs. ii) ?two (beautiful) top of the seventh inning home runs. 14 On the complexity of (most) determiners, see Leu (2015). 15 As noted by Jespersen (1961, sect. 10.12). Kayne 7 If we set aside demonstratives,16 plural -s in English has the property that it cannot be followed by an overt noun (within the relevant DP17), as seen in both (42) and (44). In some cases, as in both (41) and (45), deleting the overt N makes plural -s possible. Spanish plural -s can, on the other hand, readily be followed by an overt noun, as in: (46) buenos libros (‘good books’) (47) muchos libros (‘many books’) as opposed to English: (48) good(*s) books (49) many/*manies books so the well-formedness of (43) is not suprising. A second difference between unos and ones lies in the fact that unos cannot be immediately preceded by an adjective in the way that ones can be in (41): (50) *Yo tengo buenos unos (libros). (‘I have good some/a few (books)’) This is not specific to plural unos; it also holds for singular un(o), una, even in those cases where English allows a/an to be preceded by an adjective (and a degree word): (51) You have too big a house. (52) *Tú tienes demasiado gran(de) una casa. This property of the Spanish indefinite article appears to hold quite generally across Romance languages. Whatever turns out to underlie it, it seems likely that it will not affect the relevance of unos to ones, i.e. the fact that the existence of unos lends plausibility to the determiner status of ones. French uns differs from Spanish unos in that French uns cannot be immediately followed by an overt noun (and in that way resembles English ones). French has: (53) Les uns sont partis, les autres sont restés. (‘the ones are left, the others are stayed’) but not: (54) *Les uns enfants sont partis, les autres (enfants) sont restés. Adding (non-appositional) enfants (‘children’) to les uns in (53) is not possible. Uns is also possible in French in: (55) Quelques-uns sont tombés. (‘some ones are fallen’) Again, adding a noun like livres (‘books’) is not possible: (56) *Quelques-uns livres sont tombés. What French adds to the discussion can be seen in: (57) Jean a mis quelques livres sur la table. (‘J has put some books on the table’) (58) *Jean a mis quelques-uns sur la table. Quelques-uns is possible as a preverbal subject in (55), but not as a postverbal object in (58), and in this respect differs sharply from ’quelques + overt noun’, as seen in (57), suggesting that uns is not a noun, just as ones is not (and unos is not), if the discussion so far is on the right track. ———————————— 16 If the final consonant of these and those is the plural -s, then demonstratives fall outside the text statement (cf. Kayne (2010a)). Alternatively, Bernstein (2015) has argued that the final consonant in these and those is a genitive -s. 17 In students that age, there is probably a silent preposition intervening between students and that age. Kayne 8 What uns is, is a (complex) determiner (un + s), with un an indefinite article/numeral and -s a plural morpheme. This is indirectly supported by the fact that (55) vs. (58) is a contrast found quite generally with indefinite determiners in French (and Italian), e.g.: (59) Trois sont tombés. (‘three are fallen’) (60) *Jean a mis trois sur la table. (‘J has put three on the table’) with both (58) and (60) requiring the addition of clitic en (‘of them/thereof’):18 (61) Jean en a mis quelques-uns sur la table. (62) Jean en a mis trois sur la table. The parallelism between trois and quelques-uns seen in (55)-(62) supports taking quelques-uns to be a (highly) complex determiner of which determiner uns is a subpart. In no way, apart from the very presence of -s, does uns in French act like a (plural) noun. Indirectly, then, French uns increases the likelihood that it is correct to take English ones not to be a noun, but rather a determiner. 7. Defusing the objection. Part II The possible objection to the determiner status of ones based on the impossibility of (42), repeated here: (63) *They have blue ones cars. in addition to being weakened by the considerations of the previous section concerning Spanish unos and French uns, is further weakened by the observation that the contrast between (63) and (64): (64) They have blue ones. is not specific to English ones. What I have in mind involves French interrogative quel (plural quels, for masculine gender), a close counterpart of English which, as in: (65) Which linguists have you invited? (66) Quels linguistes as-tu invités? If the lexical noun modified by which or quels is silent, we have: (67) Which have you invited? (68) Lesquels as-tu invités? English seems straightforward, but in French, in (68), instead of quels, we get lesquels, which is the definite article les followed by quels. The link to ones rests on the contrast between (68) and: (69) *Lesquels linguistes as-tu invités? French interrogative lesquels is like ones in disallowing an immediately following overt N (within the same DP), i.e. (69) is parallel to (63) (and (68) to (64)). To the extent that (les)quels is, as seems clear, a complex determiner and not a noun (any more than English which is), we have indirect evidence that ones, too, is a complex determiner and not a noun. 8. Derivations The question arises as to how to best understand this common behavior of ones and lesquels. Let me begin with (69) vs. (68), which seem to differ only in that (69) has a lexical noun (linguistes) while (68) has a silent noun. That difference does not by itself ———————————— 18 On en corresponding most closely to English thereof, see Kayne (2004). On the subject-object asymmetry at issue in the text, see Pollock (1998). Kayne 9 account for the difference in acceptability between (69) and (68). Consider, then, the proposal in Kayne (2006) to the effect that a silent noun does not end up in the same position as its pronounced counterpart. In that spirit, let us take lesquels in (68) to be, not ‘les quels NOUN’, but rather:19 (70) [ les NOUN ] quels Assume further that (70) must be derived from: (71) quels [ les NOUN ] by leftward movement of ‘les NOUN’, so that a fuller variant of (70) (and (68)) is: (72) [ les NOUN ] quels < [ les NOUN ] > In other words, the definite article les can come to precede interrogative quels in (68) only via movement (internal merge). Assume more specifically that this movement operation reflects the noun in question needing to reach a position in which it will not be pronounced.20 If so, then, if the noun is not silent, the movement operation in question will not take place. In which case, les will not come to precede interrogative quels.21 That will exclude (69), as desired.22 Along the lines just sketched, consider the following approach to (63) vs. (64), repeated here: (73) *They have blue ones cars. (74) They have blue ones. Ones is a complex determiner. The adjective blue here modifies cars or, in (74), its silent counterpart CARS. In (73), blue is not adjacent to cars, contrary to expectations. Put another way, (73) shows an ‘Adj Det N’ order, rather than the expected (for English) ‘Det Adj N’ order. This is, I think, part of the reason for (73)’s unacceptability. The challenge is then to simultaneously understand the contrasting acceptability of (74). In the spirit of Hendrick’s (1990) analysis of too big a car, and in line with (72), let me take (74) to have a derivation reflected in: (75) [ blue CARS ] ones < [ blue CARS ] > in which ‘[ blue CARS ]’ moves past ‘ones’, starting from the expected ‘Det Adj N’ configuration.23 As in the case of (72), the movement in question will be linked to the (ultimate) silence of the noun, so that it could not apply to ‘ones [ blue cars ]’ to yield: (76) *They have blue cars ones. any more than it could apply to yield (73). As for the impossibility of: (77) *ones blue cars ———————————— 19 This differs in part from Kayne (2008a, sect. 7). There might also be a link here to Kayne’s (2002b) idea that antecedents need to be reached via movement. 21 Why lesquels acts differently in non-restrictive relatives remains to be understood. Cf. Grevisse and Goosse (2011, §619). 22 A remaining question is why French does not then allow: (i) *Quels les linguistes as-tu invités? 23 Cf. Greenberg (1966) and Cinque (2005), though neither attempted to integrate articles. A separate question is whether their ‘Dem Num Adj N’ reflects external merge alone, or whether internal merge is also involved. See also Shlonsky (2004). Barbiers (2005, 172) has the idea that DP moves, triggered by focus, to spec of one in Northern Brabantish, in a partially similar way. 20 Kayne 10 it may be that in English (vs. Spanish unos) there is a problem with the two instances of plural -s. As for: (78) *one (blue) cars it may ‘simply’ be a question of agreement.24 In any event, there is now a key similarity between (74) and the equally acceptable: (79) They have too blue a car. insofar as both have an adjective preceding a determiner. Concerning the contrast between (73) and (79), a solution is suggested by Turkish bir (‘a’/’one’). According to Kornfilt (1997, 106), bir can either precede or follow an adjective, i.e. one can have in Turkish either ‘bir Adj N’ or ‘Adj bir N’; however, when bir precedes the adjective, it corresponds to English numeral one; when bir follows the adjective, it corresponds to the English indefinite article. This leads to the following possibility. English one (classifier + indefinite article) must end up in a higher position within DP than the position of the indefinite article alone. The indefinite article can be preceded by an adjective, as in (79), or be followed by one, as in the usual case: (80) They have a blue car. The position of one, though, is sufficiently high that it may not be preceded by an adjective, as seen in (73). There is one exception, of course, namely (74). Thinking of Cinque (2005), the generalization is: (81) An adjective can come to precede one only if moved along with a noun. In English, this noun must be silent, so (74) contrasts with (76). For Hendrick (1990), blue in (79) comes to precede a as the result of a wh-like movement operation in which blue is pied-piped by too (or another degree word). In the absence of an appropriate degree word, the adjective is not allowed to precede a: (82) *They have blue a car. Thinking of (21) and (22) above, though, it might be that an adjective can precede an indefinite article in English even in the absence of a degree word, as long as the indefinite article is silent, in which case (80) would be:25 (83) ...a blue A car with A the silent article. If so, then (79) might be: (84) ...A too blue a car and similarly for the colloquial: (85) They have too big of a car. which might be: (86) ...A too big of a car as suggested by the large number of Google hits for phrases like a too big of a car (not possible for me). ———————————— 24 With a possible link to: i) A group of three/?two students is waiting in your office. ii) *A group of one student is waiting in your office. and/or to: iii) all three of Mary’s three children iv) *every single one of Mary’s (one) child 25 Cf. Tat (2011) for a similar proposal on Turkic languages. Kayne 11 9. A further restriction on ones The derivation of (74) briefly sketched in the preceding section takes the adjective in blue ones to originate, as a modifier of cars/CARS, below the determiner ones. Not all adjectives are compatible with ones the way that blue is in (74), however. For example, few is an adjective, to judge by the series few, fewer, fewest. Yet we have (for my English): (87) Few linguists went to that talk and few (*ones) to this talk. Similarly: (88) Mary has written few papers this year, but John has written even fewer (*ones). (89) This year, of all the graduate students, it’s John who’s written the fewest papers/*ones. Since few, fewer, and fewest happily occur with all sorts of (plural) nouns, this is another indication that ones is not a noun, but rather a (complex) determiner. In all likelihood, thinking of Kayne’s (2002a; 2005a) proposal to the effect that few is necessarily a modifier of number/NUMBER, the facts of (87)-(89) reduce to those of:26 (90) Only a small number of linguists went to that talk and only a small number (*of ones) to this talk. (91) Mary has written only a small number of papers this year, but John has written an even smaller number (*of ones). (92) This year, of all the graduate students, it’s John who’s written the smallest number of papers/*ones. In (90)-(92), number of cannot be followed by unmodified ones. In (87)-(89), the same holds for NUMBER OF, as in: (93) *...and few NUMBER OF ones to this talk with silent NUMBER and probably silent OF. The restrictions seen in (87)-(92) fall sharply away if an adjective or adjectival phrase is added, e.g.: (94) John has written many papers, but few good ones. (95) Only a small number of good ones were written this year. This fact about few (and fewer and fewest, and also a few) is almost certainly the same fact that we see with numerals:27 (96) John has written three *(good) ones this year. 10. The licensing role of adjectives The way in which the adjective in (94)-(96) ‘saves’ those sentences is arguably not specific to English ones. French has: ———————————— 26 Many acts like few here, as opposed to numerous: i) We’ve bought ?numerous/*many ones. The reason is that numerous is not a modifier of NUMBER in the way that few and many are. For more details, see Kayne (2002a). 27 Payne et al. (2013, 814) give, without appreciating the non-unicity of English, two examples of definite the five ones... which are for me only marginally acceptable, probably in a way related to the discussion in section 18 below. Kayne 12 (97) Vous avez acheté de *(bons) vins. (‘you have bought of (good) wines’) French allows a plural DP to have the form ‘de Adj N’,28 but does not (apart from polarity contexts) allow DPs of the form ‘de N’. If we think of this fact as indicating that in (97) the adjective is licensing the preposition de,29 there is a clear point of similarity to the licensing of ones in (94)-(96). Of importance is the fact that in French a postnominal adjective or relative does not suffice for such licensing: (98) *Vous avez acheté de vins excellents. (‘you have bought of wines excellent’) (99) *Vous avez acheté de vins qui sont bons. (‘...of wines which are good’) The link with ones is strengthened by the fact that (94)-(96) become to my ear unacceptable if the pre-ones adjective is replaced by a (reduced) relative: (100) *John has written many papers, but few ones that are any good. (101) *John has written many papers, but few ones worth reading. (102) *?Only a small number of ones that are good have been written this year. (103) *?Only a small number of ones as good as ours have been written this year. (104) *John has written three ones that you’ll like this year. (105) *John has written only three ones worth reading this year. A licensing property comparable to that seen in (97), again involving prenominal (but not postnominal) adjectives or reduced relatives has been discussed by Leu (2015, 16) for colloquial Slovenian, based on work by Marušič and Žaucer (2006). Colloquial Slovenian has an unstressed non-demonstrative definite article that requires such adjectival licensing, as seen in: (106) ta *(nov) pes (‘the new dog’) Leu (2015, 13) also discusses similar facts found in some Scandinavian languages (and in Swiss German). Although these Slovenian and Scandinavian facts are not identical to the French ones, they share a common property. In each case, what is licensed by the adjective is arguably some kind of determiner. In Slovenian and in the relevant Scandinavian languages it is a definite article; in French it is a preposition that may be part of an indefinite determiner.30 In none of these cases is it a noun that is being licensed. Consequently, the resemblance to the licensing of English ones by a pre-ones adjective seen in (87)-(105) indirectly reinforces the determiner status of ones.31 ———————————— 28 Cf. Pollock (1998, note 24). In French this de can also be licensed by a following determiner, as in: i) Vous avez acheté des (bons) vins. (‘you have bought of-the (good) wines’) in which case an adjective is no longer necessary. 30 For recent discussion of this kind of French de, see Ihsane (2008). 31 An example of licensing by a pre-ones reduced relative in English is: i) There are a few old letters on the chair and a few recently arrived ones on the table. On adjectives and reduced relatives more generally, see Cinque (2010). The text cases are to be distinguished from cases in which the licensing modifier needn’t be prenominal; see Longobardi (1994, note 12) on determinerless nouns in preverbal subject position. 29 Kayne 13 It is not easy to see what exactly distinguishes those determiners that require licensing by a adjective and those that do not.32 In English, numerals and few need no adjective, even when the associated noun is silent: (107) Mary has written four papers this year, but John has written only three. (108) Many papers are started, few are finished. Similarly for some, with a deleted/silent plural noun: (109) Some will be finished. and for plural demonstratives: (110) These will be finished. Singular demonstratives are partly different: (111) This will be finished. While (111) is acceptable, the silent noun it contains would seem to be THING, and cannot be dependent on an antecedent in the way that the silent noun of (110) can be. Of interest here is one other determiner in English that looks as if it requires, in a certain kind of context, adjectival licensing of the sort under discussion. Consider: (112) People are often in need of help. (113) Very poor people always need help. The noun people in (113) can be deleted/silent, but in that case a definite article is required: (114) The very poor always need help. (115) *Very poor always need help. Having silent PEOPLE together with a definite article is not possible in (112), however: (116) *The are often in need of help. In effect, the the of (114) needs licensing by an adjective (phrase). As in (98)-(105), a relative clause is not sufficient: (117) *The who have little money are often in need of help. Nor is a reduced relative of the sort that is otherwise postnominal: (118) *The lacking money are often in need of help. The adjectival licensing requirement of this the, then, is similar to that of ones. In both cases, a certain determiner requires an adjective in the context of a silent noun. This parallelism further reinforces the determiner status of ones. Of all the adjective-requiring determiners discussed, though, ones is the only case in which the licensing adjective precedes the determiner in question. Thinking of the discussion around (75) above, in which it was proposed that the adjective preceding ones originates below ones, it may be that the licensing of ones by the adjective takes place prior to that movement,33 in which case all the determiners in question will turn out to be licensed by a following adjective. ———————————— 32 The indefinite article requires licensing by an adjective in: i) You must have spent a *(good, beautiful) three weeks in Italy. ii) You should invite a*(n other) four people. 33 And similarly for: i) too long (of) a book In the cases he discusses, Leu (2015, 92) has the determiner and adjective forming a constituent. Extended to ones, this would mean that (at the point of licensing) ones + adjective is a constituent. Kayne 14 This statement will also cover (for my English):34 (119) We bought a blue one yesterday. (120) *We bought a one yesterday, too. Since the indefinite article does not normally require an adjective, it seems likely that the contrast between (119) and (120) rests on singular one also needing an adjectival licenser here. As in the other determiner cases discussed, a relative clause is not sufficient, in my English: (121) *We bought a one yesterday that was really beautiful. Nor is a reduced relative of the postnominal sort:35 (122) *We bought a one worth reading. In some English (not exactly mine), though, it is possible to have: (123) We bought not a one. In this English, either not itself must be the licenser, or else there must be a silent SINGLE making (123) quasi-equivalent to:36 (124) We bought not a single one. which is fully acceptable to me. (Again, the fact that one in (119)-(122) requires an adjective reinforces the determiner status of one.) 11. Non-restrictive adjectives. Some notion of contrast or focus seems to be relevant to the licensing role that adjectives play with one (cf. Halliday and Hasan (1976, 95, 97) and Llombart-Huesca (2002, 73): (125) People who read interesting books generally profit considerably from the reading of those interesting books/*ones. (126) People who read an interesting book generally profit considerably from the reading of that interesting book/*one. It may be that a non-restrictive adjective cannot be itself license one(s), much as Llombart-Huesca suggests that noun ellipsis is not possible with a non-restrictive adjective in Spanish. If so, a possible account might rest on the discussion around (75) above, in which it was proposed that the adjective preceding ones originates below ones, in which case it might be that non-restrictive adjectives cannot originate below one(s), from which it would follow that they cannot license one(s). ———————————— 34 Cf. note 2. As opposed to: i) We bought a recently published one. with a reduced relative that is prenominal. 36 For another case of a silent adjective with no antecedent, see Kayne (2005a, sect. 7) on GOOD. Possibly the English that accepts not a one has it as: (i) ...not SINGLE a one with SINGLE preceding, rather than following, the indefinite article, with this position for SINGLE licensed by the presence of not (via movement of the phrase ‘not SINGLE’ from postnominal position directly to pre-a position) - cf. Troseth (2009) on not very good of a book. 35 Kayne 15 12. Numeral one Numeral one needs no adjective, even in the presence of a silent noun: (127) John has written three papers. Two are on phonology and one is on syntax. (128) There are three books on the table. Only one is worth reading. In this respect, numeral one behaves like other numerals, as illustrated in (107) above. This may at first seem unsurprising, but Barbiers (2007) has shown that one is quite different from other numerals in some ways, in particular in not lending itself (in a great many languages) to regular ordinal formation: (129) The first/*oneth chapter is the most interesting. Barbiers’s point about the special behavior of one can be further strengthened by noting that in many Romance languages one is the only numeral that shows agreement in gender. In addition, we can note that in French complex numerals that are multiples of 100 (or 1000), one is the only numeral that cannot appear, as seen, for example, in: (130) deux cents (‘two hundred’), trois cents (‘three hundred’)... (131) cent (132) *un cent (‘one hundred’)37 French also displays an asymmetry between one and other numerals in that in the additive compound numerals 21, 31, 41, 51, 61, 71, an overt coordinating element et (‘and’) is necessary, e.g.: (133) vingt-et-un livres (‘twenty-and-one books’) whereas with 22, 23,...32, 33... no coordinating element appears, e.g.:38 (134) vingt-deux livres (‘twenty-two books’) If numeral one is unlike other numerals in various ways, then the fact that in (127)(128) numeral one acts, with respect to the need or non-need for adjectival licensing, like other numerals and not like the one of (120), or like ones, needs to be accounted for. The proposal that comes to mind is that the exceptionality of numeral one relative to adjectival licensing is only apparent and that numeral one is in fact necessarily accompanied by an adjective. More specifically, thinking of the discussion of (123), let me take sentences with numeral one such as: (135) John has two brothers and one sister. to have the analysis: (136) ...and one SINGLE sister. with a silent adjective corresponding to single.39 In some cases, one is natural with a following overt single: (137) You haven’t written one single paper this year. ———————————— 37 With 1000, French has: i) (*un) mille linguistes (‘a thousand linguists’) Possible, with a complex numeral containing one as a subpart, is: ii) trente-et-un mille linguistes (‘thirty and one thousand linguists’) 38 Though there may be a silent one present, to judge by the obligatory pronunciation of the final consonant of vingt in 22, 23... 39 There is a point of similarity here with Borer’s (2005, 196) proposal that Hebrew ’exád (‘one’) is an adjective interpreted as ‘single’. Kayne 16 It should be noted that the title of this section is somewhat misleading. If this paper is on the right track, one is consistently to be understood as ‘classifier + indefinite article’. There is no single morpheme that would correspond exactly to ‘numeral 1’. More than that, one itself is the same (‘classifier + indefinite article’ combination) in all its instances. Some instances of one are numeral-like in that they contrast readily with other numerals, as in (135). Other instances of one are not, e.g. in two blue ones. Thus the term ‘numeral one’ must be taken to pick out those instances of one that occur in a syntactic context whose overall interpretation lends itself to contrast with other numerals. If (136) is correct, then that context will include an adjective like single/SINGLE. In some cases, overt only is very natural: (138) John has two brothers but only one sister. Silent ONLY might be present in other cases. If only is adjectival, it itself might be able to serve as licenser for one, perhaps in some cases instead of SINGLE. Whatever the correct details, it seems extremely likely that the language faculty treats ‘numeral 1’ as complex, not simplex (i.e. not as a primitive). 13. Ordinals The idea that numeral one is to be understood as in (136) is in partial agreement with Barbiers’s (2005; 2007) claim that one is very different from two and numerals higher than two. He took numeral one to be a stressed, focussed version of the indefinite article.40 The present proposal doesn’t rely directly on the notion of ‘focus’, using instead the presence of SINGLE. As mentioned in the previous section, Barbiers emphasized the relative systematicity of the cross-linguistic absence of a regularly formed ordinal based on one: (139) Mary was the first/*oneth linguist to have proposed that. From the present perspective, this must reflect the inability of ordinal -th to combine either with the complex determiner one or with SINGLE or single: (140) *the (a/one) single-th linguist Presumably, the numerals from two on up (apart from complex numerals having 1 as a subpart) do not (necessarily) involve SINGLE. (Why ordinal -th differs from the -ce of once, which can combine with one,41 needs to be elucidated.) 14. One(s) and demonstratives. There is a clear contrast (in my English - cf. note 2) between (120), repeated here: (141) *We bought a one yesterday, too. and: (142) We bought that one yesterday, too. This may indicate that demonstrative that (or this) is capable by itself of licensing one independently of the presence of a canonical adjective, perhaps because ———————————— 40 As mentioned in an earlier footnote, this view of one faces a challenge dealing with stressed a, as in: i) We don’t need some chocolates, we need a chocolate. with a pronounced to rhyme with say. 41 On once, see Kayne (2014). Kayne 17 demonstratives have something significant in common with adjectives.42 Things are more complex, however, since in some English (for example, mine) there is a contrast between the singular case of (142) and its plural counterpart: (143) *We bought those ones yesterday, too. For those speakers who accept (143), there seems to be no extra complexity. But for speakers like me, who reject (143), the question is why the plural case should be different from the singular. (Again we can note that taking ones to be a noun would incorrectly lead to the expectation that (143) should be acceptable to all.) Not surprisingly, by now, adding an adjective to (143) makes it acceptable to all (as far as I know): (144) We bought those blue ones yesterday. Llombart-Huesca (2002, 77), by taking the appearance of one(s) to be a last resort strategy (her one-support) that comes into play only when NP-ellipsis is not available, could perhaps relate my rejection of (143) to my accepting: (145) We bought those yesterday, too. The problem, however, is that even if the language faculty sometimes has recourse to last resort strategies (which isn’t clear), that sort of approach to (143) would have difficulty accounting for the fact that many speakers accept both (143) and (145). (I don’t know with certainty if there are speakers who accept (143) and reject (145).) Complementarity between one(s) and NP-ellipsis fails to hold in various other cases, too.43 In the context of a shirt store, I accept both of the following: (146) The blues/blue ones are selling well this week. In the context of a day-care center for children, I accept both of these: (147) The three-year olds/three-year old ones are easier to manage than the others. And without any special context, both of the following: (148) The others/other ones are even less expensive. and similarly for: (149) There are others/other ones on the table. as well as (with no -s in the NP-ellipsis variant): (150) John gave several talks. Only the first (one) was understandable. (151) Each (one) was good in a different way. (152) The tallest (one) of the three is really very tall. (153) The taller (one) of the two is really very tall. Failure of complementarity also holds with that (and this), though with a difference in interpretation: (154) Give me that. (155) Give me that one. These two singular demonstrative examples are both possible, but not quite on a par. This can be seen clearly in examples where the antecedent is human: (156) That linguist prefers phonology, while this *(one) prefers syntax. With that one or this one, the antecedent can readily be a human noun like linguist, or not. With that or this alone, the understood antecedent cannot be human. In fact, it arguably cannot be any ordinary noun, a relevant example being: ———————————— 42 43 Cf. Leu (2007; 2015, chap. 2). As noted by Halliday and Hasan (1976, 97). Kayne 18 (157) That decision was made by committee, while this *(one) was made by John alone. With this one, (157) is extremely close in interpretation to: (158) That decision was made by committee, while this decision was made by John alone. With this alone, (157) is possible, but not with the same interpretation.44 A way to understand this difference is to take (154) to be as in: (159) ...that THING and (155) to be as in: (160) ...that one NOUN in which the silent NOUN necessarily has an antecedent (and similarly for (156)). THING in (154)/(159), on the other hand, does not need or take an antecedent in anything like the same sense. Rather the interpretation is close to the also possible: (161) Give me that thing. The ability of THING to appear in this way in (154)/(159) is probably related to its special behavior in:45 (162) something else (163) *some book else If so, the absence of a true plural for (154), which is very clear is some cases: (164) I’ve been wondering that/*those myself. (165) How can you possibly think that/*those? will link to: (166) *somethings else supporting the presence of THING in (154). As Edwin Williams (p.c.) has pointed out to me, thing is also special in that it cannot readily be the antecedent of one(s) (more precisely, from the perspective of (160), of the silent NOUN that accompanies one(s)). Examples of this are:46 (167) They’ve been wondering if it’s time to leave and we’ve been wondering the same thing. In fact, you’ve been wondering the same thing/*one, too, haven’t you? (168) How come you did such a smart thing last night, but such a stupid thing/??one this morning? (169) John said something and Bill said something/*someone else. In the variants of (167)-(169) with one, there must be ‘...one NOUN’, much as in (160). The question is why this silent NOUN cannot readily take thing in (167)-(169) as antecedent. The answer might perhaps be that thing is in these examples classifier———————————— 44 As noted by Halliday and Hasan (1976, 96). Cf. Leu (2005). 46 The first example has a DP apparently Case-licensed by wonder, in contrast to Pesetsky (1991, 6). Note the contrast with adjectives, e.g.: i) *We’re sure it’s time to leave and they’re sure the same thing. It may be that the property of thing at issue is (to some extent) limited to abstract, as opposed to object, thing, a distinction that some languages make explicit; cf. Zepeda (1983, 53, 55) on double counterparts of nothing and what. 45 Kayne 19 like47 and that having a classifier-like NOUN associated with one would clash with the fact that one itself contains a classifier, as suggested earlier. 15. Singular vs. plural. Let us return now to (143), i.e. to the fact that English speakers differ from one another on (*)those ones. We have seen earlier that ones is readily licensed by an immediately preceding adjective, as in:48 (170) They bought three *(blue) ones yesterday. Whereas if ones is immediately preceded by a determiner (broadly construed), we generally have unacceptability, as in:49 (171) *They bought few/a small number of/three/several/some ones yesterday. For speakers like me who reject *those ones, demonstrative those (as well as these) is acting like the other determiners of (171). Speakers who accept those ones and these ones may be taking those/these to be adjectival; alternatively they are taking the silent THERE/HERE that accompanies those/these to be adjectival.50 The contrast, for one set of speakers, between singular this one, that one and plural *these ones, *those ones leads to the question whether the adjectival licensing relevant to plural ones is at all relevant to (certain instances of) singular one. In fact, (171) has a parallel with some singular determiners: (172) John was attending some class (or other) yesterday and Mary was attending some class/*some one (or other) yesterday, too. (173) Mary made one mistake and John made one mistake/*one one, too. As with plural ones, adding an adjective changes the acceptability status: (174) Mary made one bad mistake and John made one bad one, too. With singular some there is clearly improvement, even if the result is not entirely natural: (175) (?)John was attending some weird class yesterday and Mary was attending some weird one (yesterday), too. In addition, I find that every is to some degree like singular some, in some cases: (176) As for spiders, every ?(single) one has eight legs. whereas with each we have: (177) Each spider/one will be found to have eight legs. Why singular one is compatible with a preceding determiner to a greater extent than plural ones is left an open question. 16. Possessors. That ‘adjectival’, as far as the licensing of ones is concerned, might go beyond ordinary adjectives to include THERE/HERE, as suggested in the discussion of (171), is ———————————— 47 Though ‘thing’ seems to be compatible with various classifiers in Cantonese - see Matthews and Yip (1994, 106). 48 On ‘immediately’, note: i) They spent a beautiful three weeks/*ones in France last year. 49 Conceivably there’s a point of contact here with: i) Someone(*s) else called. 50 Cf. especially Leu (2015, 32) on Norwegian. Kayne 20 indirectly supported by some curious facts concerning possessors, which in my English are not by themselves licensers of ones: (178) *Mary’s papers are usually stronger than John’s ones. Yet I find the following more acceptable: (179) ?Men’s shoes are usually less expensive than women’s ones. as if women’s here could count as adjectival in a way that John’s could not.51 The non-licensing property of possessors seen in (178) carries over for me to singular one: (180) *Mary’s paper is stronger than John’s one. As expected by now, adding an adjective makes them acceptable: (181) We appreciate John’s recent one(s). both with ones and with one, reflecting the fact that the licensing conditions for ones and one are to a degree similar. 17. Human one There is a clear contrast in my English between (172) and: (182) Let’s invite someone. Here one can be immediately preceded by some (and similarly for every, any and no), in direct contrast with (172). One might ‘just’ say that this one is different, which it of course is. But a primary claim of this paper is that all instances of one are in fact the same in their internal makeup (the more specific claim is that all are classifier + indefinite article), and that it is the syntactic contexts in which one occurs that differ. Since someone, everyone, noone and anyone are basically restricted to humans, it is natural to link the one in them to the one of: (183) When one is hungry, food becomes essential. which is also restricted to humans. This link is strengthened by the fact that neither (182) and (183) allows ones to replace one:52 (184) *Let’s invite someones. (185) *When ones are hungry, food becomes essential. A sentence like (183) is in turn close to: (186) When a person is hungry, food becomes essential. ———————————— 51 Another kind of example with something adjectival, but not strictly speaking an adjective is: i) John makes lots of remarks, including lots of over-the-top ones. Note also: ii) *I’ve read neither John’s papers nor Mary’s ones vs. iii) ?I’ve read neither yesterday’s newspapers nor today’s ones 52 There may or may not be a link to: i) The plates were piled one on top of the other. ii) *The plates were piled ones on top of the other(s). Kayne 21 Thinking of the proposal in Kayne (2005a, appendix) to the effect that John is ‘John PERSON’, as well as of the discussion of (114) above, it becomes natural to think of (182) and (183) as containing:53 (187) some one PERSON; one PERSON with one a determiner, as it now always is. A discrepancy between (182) and (183) concerns else: (188) Let’s invite someone else. (189) *When one else is hungry,... This contrast between (188) and (189) suggests that else in (188) depends on the presence of the determiner some.54 Thinking of the similarity between someone else and some other person: (190) Let’s invite some other person. and more generally between else and other, it may be that one in (188) is licensed by quasi-adjectival else, rather than by some. (As for (182), or even (183), it might be that PERSON itself plays a licensing role for one, rather than some doing so (in (182)); alternatively a silent adjective might be present.) The impossibility of (184) of course recalls: (191) Let’s invite somebody/*somebodies. which may reflect a common prohibition against plural. If so, and if the one of someone has, as I have been suggesting in this section, the same complex determiner status as other instances of one, we seem to be faced with a curious choice. Either body here must be analyzed parallel to one, i.e. as a complex determiner, or else somebody and someone are not as perfectly parallel to each other as they at first look. The latter possibility appears to be supported by the following difference, brought to my attention by Edwin Williams (p.c.): (192) He’s a real somebody/??someone. (193) He’s a real nobody/??noone. With plurals the distinction is for me even sharper: (194) They’re real somebodies/*someones. (195) They’re real nobodies/*noones. With somebody, nobody there’s a natural interpretation as ‘a really important/unimportant person’ that is less readily available with someone, noone, and similarly (but more strongly) in the plural. What this suggests is that rather than matching the one of someone, the body of somebody better matches PERSON itself, i.e. it may be that while someone is:55 ———————————— 53 The extent to which this kind of analysis should be extended to French on, or to Italian si, or to German man is left open here; for relevant discussion, see Cinque (1988) and Malamud (2013). 54 An arguably similar sensitivity to the presence of a determiner is seen in (nonstandard): i) Let’s go somewheres (else). ii) *Wheres (else) should we go? For relevant discussion, see Kayne (2007a, sect. 3). 55 Cf. the non-equivalence of somewhere and someplace discussed in Kayne (2007a). Also: i) He’s living in the middle of nowhere/*noplace. Kayne 22 (196) some one PERSON somebody is: (197) some (ONE) body with body receiving a (partially) idiomatic interpretation akin to that of person. The unacceptability of *someones, *noones in (194), (195) recalls that of: (198) *John ate some apples and Bill ate some ones, too. (199) *John has no friends and Bill has no ones, either. thereby indirectly supporting the common determiner status of the one of someone, noone and the one of some red one, no good one.56 18. The The generalization that other determiners are not licensers for plural ones has one sharp exception that holds for all speakers (as far as I know):57 (200) Bring us the ones that you consider worth reading. This kind of sentence is perfectly acceptable without any overt prenominal adjective, contrary to various cases discussed earlier, e.g.: (201) *Bring us three ones that you consider worth reading. On the other hand, with neither an adjective nor a relative clause, (200) becomes unacceptable:58 (202) *Bring us the ones. Why a relative clause seems to be able to act as a licenser for ones in (200) but not in (201) is not clear. Relevant is the status of: (203) ?Bring us ones that you consider worth reading. with no determiner preceding ones. In my English, (203) is deviant to some degree, as seen also (for me) in: (204) Don’t bring us heavy ones. (205) ?Don’t bring us ones that are heavy. Yet (203) and (205) are acceptable to many, with such speakers presumably allowing the relative to act as licenser, given: (206) *Bring us ones. For my English, the contrast between (200) and (203) suggests that the is playing a key role, too.59 That the definite article can contribute to the well-formedness of DPs ———————————— 56 This way of looking at things makes sense of nobodies vs. *noones, while leaving open the contrast between They’re nobodies and: i) *They have nobodies else. 57 Note that the ‘head’ of the relative here contains not just ones but also at least one silent NOUN. 58 As opposed to some special cases like: i) John and Mary have both signed up, but so far they’re the only ones. These may involve a deleted relative - cf. Stirling and Huddleston (2002, 1513n) and for a sustained proposal Collins (2014). 59 Possibly via a silent prenominal THERE that might be relevant to the contrast between English and Dutch concerning ‘the...one...’ noted by Barbiers (2005, 163). Jespersen (1961, sect. 10.12) notes that some Jutland dialects allow a counterpart of definite that abominable one, as opposed to standard Danish. Kayne 23 with ones is indirectly supported by the comparable role played by demonstrate those/these in: (207) We’d like to buy *(??those) three ones, please. While *three ones by itself is bad, adding a demonstrative improves things to some extent,60 perhaps especially for those who accept those ones/these ones. 19. Just singular one. A remaining question is why the following is possible without any adjective or modifier of any sort: (208) We bought one, too. In agreement with Payne et al. (2013, 798), I take this instance of one to be closer to the one of: (209) We bought one book. than to the one of: (210) We bought an expensive one. In essence, the question is when exactly singular one needs an adjective. As an initial approximation, we may have: (211) Only when preceded (within the relevant DP) by another determiner does singular one need to be preceded by an adjective. 20. The -s of ones. Saying that ones is plural normally goes with the (usually implicit) assumption that one and plural -s form a constituent. But that assumption is not straightforward.61 Nor is it in the case of others, as in: (212) Give me the others. which I take to have an analysis as in: (213) ...the other NOUN s where -s is associated with the silent noun rather than directly with other.62 An imaginable alternative would have other itself sometimes being a noun in addition to usually being an adjective. There are, however, reasons for thinking that at least this kind of category multiplication/neutrality is not made available by the language faculty. If, in addition to being an adjective, other could also sometimes be a noun, one would wrongly expect the following to be straightforwardly possible: (214) ?Give me that other. (215) *You’ve eaten every (single) other. Furthermore, there is a striking fact having to do with the interaction between other and other adjectives. One has: (216) The other American invasions took place years ago. (217) The other American ones took place years ago. ———————————— 60 As opposed to adding some in: i) Mary has published some twenty papers/*ones in the last five years. 61 Cf. Koopman and Szabolcsi (2000) and Julien (2002). 62 Cf. Kayne (2003). Kayne 24 If other could also be a noun, one would wrongly also expect to have:63 (218) *The American others took place years ago. Whereas if other is consistently an adjective, (218) reduces to: (219) *The American other invasions took place years ago. The appearance of plural -s following a silent noun is allowed in a variety of adjectival cases, not just with others: (220) They have two four-year olds. (221) If I had a choice among those crayons, I’d take all the reds. (222) In that linguistics department, the first-years are under a lot of pressure. In other cases, this is not possible, for reasons yet to be determined: (223) *Speaking of invasions, the Americans took place years ago. (224) Those three books are more interesting than these four(*s). It seems almost certain that the -s of ones has the same property as the -s of others (and the -s of four-year olds, reds and first-years), namely this -s is associated with a silent noun, rather than simply with one itself. For example the following:64 (225) I prefer red cars, but you prefer blue ones. is to be analyzed, parallel to (213), as containing as a subpart:65 ———————————— 63 The impossibility of this kind of example was noted by Stirling and Huddleston (2002, 1524), who did not, however, draw the conclusion that other is always an adjective. Their reason was that the others is possible, combined with the belief that adjectives never take plural -s in English. British and American English seem to differ in that only British English has, with a simple numeral: i) Mary has three millions in the bank. in which the -s is likely associated with silent POUND. In (my) American English, this -s does not appear: ii) Mary has three million in the bank. though DOLLAR(S) is presumably present. Her and Tsai (2015, 592) note the existence of doublets like: iii) There are three grand pianos/grands in the storeroom. which they interpret as showing that grand in (iii) is a noun. Alternatively, it is an adjective occurring with either piano or PIANO. The monetary grands (possible for some speakers) that they discuss in their section 4.2 is compatible with monetary grand being an adjective, in the same way. The question whether the specific analysis of monetary grand proposed in Kayne (2012) is on the right track is beyond the scope of this paper. 64 With the indefinite article in place of one, we get the impossible: i) *...but you prefer blue a’s Presumably, this is the same fact as: ii) *John has a car and you have a, too. in turn akin to: iii) John likes the *(car), too. 65 One here is associated with plural ‘CAR s’, contrary to: (i) We have one car(*s). Thinking of Heim (1987), van Riemsdijk (2005) and Leu (2008), this might suggest: (ii) ... blue one KIND CAR -s Kayne 25 (226) ... blue one CAR -s 21. Restrictions on what follows one(s). Baker (1978, 415) considered sentences like the following:66 (227) *In this university, the students of physics are generally stronger than the ones of chemistry.) For Baker, this kind of example merits a full *. To my ear, this sentence is deviant, but not dramatically so. Baker took this kind of restriction to constitute an argument in favor of an innate component to the language faculty. Payne et al. (2014) try to show that Baker’s argument does not go through, by observing, via a corpus study, various (relatively) acceptable examples of the same general sort as (227). I myself would tend to accept: (228) ?The assassination of the prime minister had taken place two years before the one of the president. Baker had taken (227) to indicate that one(s) could not be followed by what would correspond to a complement of the antecedent, i.e. that one(s) could not replace N, as opposed to N-bar. For those speakers who fully accept examples like (228), this cannot be right as a general characterization of one(s). From the perspective of the present paper, one(s) does not ‘replace’ a lexical noun at all; rather, in, say, (228), one is followed by a silent counterpart of assassination. Moreover, since I have argued elsewhere that nouns do not take complements,67 restrictions like the one seen in (227), for those who reject it,68 must be formulated otherwise. Baker’s argument for an innate component to the language faculty is strengthened by the finding of more sharply unacceptable examples than (227) itself, such as in the following: (229) A large number of syntacticians were talking with a large number/*one of phonologists. (230) Mary has a whole lot of money; John has a whole lot/*one of money, too. (231) The kinds of horses we need are easier to find than the kinds/*ones of cows we need. (232) Mary likes those kind of horses and John likes those kind/*one of horses, too. (233) What do you think of those absurd goings-on and of these even more absurd goings-on/*ones-on? (234) A quiet taking out of the trash will be less onerous than a loud taking/*one out of the garbage. in which one goes with silent KIND (or some other additional NOUN) and -s with silent CAR. Pursuing this possibility would be beyond the scope of this paper. 66 Cf. Stirling and Huddleston (2002, 1515) for judgments like Baker’s. 67 Cf. Kayne (2008b). 68 Payne et al. (2014) gave me the impression, perhaps wrongly, that they believe that there is one ‘English’. For an sharply opposing view, see Kayne (1996; 2013). On the richness of syntactic variation within what we call English, see Algeo (2006), Kortmann et al. (2005), Zanuttini and Horn (2014) and many other such works. Kayne 26 In these cases of sharp unacceptability (as in many others discussed in this paper) a lot must depend on built-in principles, as Baker (1978) had it, apart from the details of his proposal. From the present perspective, what is more specifically at issue in (229)-(234) must have to do with the silent NOUN following one(s). If Kayne (2006) is on the right track, that translates into the question of what the silent noun can leave behind when it moves. Which in turn probably becomes the question of what can (non-contrastively) scramble out before such (remnant) movement takes place. For example, in the unacceptable variant of (234), with *...a loud one out of the garbage, we might have: (235) ...one TAKING out... and it may well be that a particle like out is not amenable to such scrambling.69 Similarly, in the unacceptable variant of (229), with *...a large one of phonologists, we would have: (236) ...one NUMBER of phonologists and it may be that of phonologists is not amenable to the required non-contrastive scrambling. 22. Other determiners that look like pronouns. In taking the position that one and ones are in all instances complex determiners, I have been taking a position akin to the one taken by Postal (1966) for personal pronouns like he and she, which he analyzes as being types of definite articles. (Postal took ones itself to be a [+Pro] noun.) Assuming Postal’s determiner analysis to be correct for at least third person pronouns, the position I have arrived at here claims that both third person pronouns (he, she, it, they) and one(s) are determiners, the difference being that third person pronouns are (as in Postal (1966)) definite determiners (associated with a silent or deleted noun), whereas one and ones are for me indefinite determiners (associated with a silent or deleted noun). The (complex) determiner status that third person pronouns and one(s) have in common may be reflected in what looks like a shared restriction concerning compoundlike phrases. Having a third person pronoun within a compound is not possible:70 (237) *Nixon’s supporters didn’t realize how many him-haters there were. The restriction illustrated in (237) is probably best understood as a restriction involving the silent noun associated with the third person pronoun. Compounds do not admit silent nouns within them (which may in turn be interpreted as a prohibition against movement of the noun (or NP) from within a compound up to the position of its antecedent, thinking of the discussion of (229)-(236)). Consider now one. Numeral one is allowed to appear in a compound: (238) One-drawer file cabinets are not very useful. There may well be a silent SINGLE associated with one in such examples, as in (136) above, but that SINGLE has no antecedent (and the noun drawer is overt), so ———————————— 69 With a point of similarity to: i) *their explanation away of the problem which must also involve non-contrastive scrambling, if Kayne (2008b) is on the right track. 70 Cf. Postal (1969) and Harris (2006). Kayne 27 movement need not be involved. On the other hand, the one of a red one does involve an antecedent, and so arguably the silent noun following one in a red one has moved. This is what I think underlies the restriction on this subtype of one that resembles (237): (239) One-drawer file cabinets are less useful than two-drawer/*two-one file cabinets. (240) First-time house buyers are less experienced than second-time house/*second-time one buyers. (241) John is a black-bear-lover and Mary is a brown-bear/*brown-one lover. In the versions with one of each of these examples, one is associated with a silent noun that has arguably had to move to reach its antecedent, if Kayne (2002) is on the right track. Such movement out of a compound is prohibited. It is of interest that compounds do not prohibit subparts of them from being involved in a certain kind of antecedent relation if there is no silent element at issue: (242) John is a black-bear lover and Mary is an animal lover, too. (243) John is a black-bear lover, whereas Mary is an animal hater. Here, destressed animal takes black bear as a kind of antecedent, or subject of predication.71 Postal (1966) also mentions dialects that allow: (244) you’uns; we’uns Standard English has neither this nor: (245) *you ones; *we ones As Postal notes, standard English does allow: (246) you honest ones; we smart ones in a way that is by now familiar, instantiating the adjectival licensing that we have seen in a variety of cases, as well as the non-licensing by a certain kind of determiner, in (245). (Possibly, the availability in some dialects of (244) is related to the nonpronunciation of the initial /w/ of ones, which might reflect the absence of a classifier.) 23 Conclusion. One and ones are complex determiners whose relation to their antecedent, when they have one, is mediated by a silent noun. They are never themselves nouns taking an antecedent directly. The question arises as to why the language faculty would turn its back on an analysis of one and ones as anaphoric nouns in certain cases. A possible answer might be in part the one given in Kayne (2002) for personal pronouns, namely that the antecedent-pronoun relation is necessarily mediated by movement of a ‘double’ of the pronoun. This might hold, now, not only for he, she..., but also for one(s), even though ———————————— 71 In the following: i) John is a self-promoting scoundrel. the antecedent of self must not be John, but rather the silent subject of promote. The relation between that subject and scoundrel needs looking into. In: ii) We’re having a three-wine dinner tonight. there may well be a silent KIND, as in Kayne (2003, note 26), but that KIND has no compound-external antecedent, just as SINGLE does not in the text discussion. Kayne 28 the character of the double would be different in the two cases. With personal pronouns, the double is a DP, with one(s) a NP. A broader question now arises. Why would the language faculty use movement (internal merge) to express antecedent-‘pronoun’ relations in general? The most interesting answer I can think of is given in:72 (247) All non-local syntactic relations necessarily involve internal merge (movement). That is, there is no possibility, in a derivational syntax, of ‘coindexing’ or directly relating two phrases in any way distinct from internal merge (or external merge, in a highly local fashion). Acknowledgements. Earlier versions of this paper were presented at the University of Cambridge, England in June, 2009; at the Workshop on Bare Nouns, Paris VII in November, 2009; at the Giornata di Dialettologia, University of Padua, in June, 2010; at the Chinese University of Hong Kong, at National Chiao Tung University, Taiwan, and at University of Rome 3 in April, 2013. I am grateful to all those audiences for useful comments and questions. References: Algeo, J. (2006) British or American English? A Handbook of Word and Grammar Patterns, Cambridge University Press. Alexiadou, A. and C. Wilder (1998) “Adjectival Modification and Multiple Determiners,” in A. Alexiadou and C. Wilder (eds.) Possessors, Predicates and Movement in the Determiner Phrase, John Benjamins, Amsterdam, 305-332. Baker, C.L. (1978) Introduction to Generative-Transformational Syntax, Prentice-Hall, Englewood Cliffs, New Jersey. Barbiers, S. (2005) “Variation in the Morphosyntax of One,” Journal of Comparative Germanic Linguistics, 8, 159-183. Barbiers, S. (2007) “Indefinite Numerals One and Many and the Cause of Ordinal Suppletion,” Lingua, 117, 859-880. ———————————— 72 Cf. Kayne (2010b). As formulated, the text statement, which has implications for the derivation of relative clauses and tough-movement, also prohibits instances of Agree that do not also involve internal merge - cf. Koopman (2003; 2005), Kayne and Pollock (2012; 2014). On NPI-licensing as involving movement, see Chomsky (1973, 242) (for the particular case of not...many) and especially Collins and Postal (2014). On topicalization and (a certain kind of) left-dislocation, note the reconstruction effects that hold (for me) in: i) His youngest daughter every man is especially fond of. ii) As for his youngest daughter, every man is bound to think she’s a genius. Similarly: iii) His youngest daughter is easy for a man to admire. Kayne 29 Beal, J.C., L. Burbano-Elizondo and C. Llamas (2012) Urban North-Eastern English: Tyneside to Teesside, Edinburgh University Press. Bernstein, J. (2015) “The Expression of Case Across English Demonstrative Forms,” talk presented at New York University. Borer, H. (2005) In Name Only, Oxford University Press, Oxford. Cardinaletti, A. and M. Starke (1999) “The Typology of Structural Deficiency: A Case Study of the Three Classes of Pronouns,” in H. van Riemsdijk (ed.) Clitics in the Languages of Europe, Mouton de Gruyter, Berlin, 145-233. Cheng, L.L.-S. and R. Sybesma (1999) “Bare and Not-So-Bare Nouns and the Structure of NP,” Linguistic Inquiry, 30, 509-542. Cheng, L.L.-S. and R. Sybesma (2012) “Classifiers and DP,” Linguistic Inquiry, 43, 634650. Cinque, G. (1988) “On Si Constructions and the Theory of Arb,” Linguistic Inquiry, 19, 521-581. Cinque, G. (2005) “Deriving Greenberg’s Universal 20 and Its Exceptions,” Linguistic Inquiry, 36, 315-332. Cinque, G. (2010) The Syntax of Adjectives. A Comparative Study, The MIT Press, Cambridge, Mass. Collins, C. (2014) “Relative Clause Deletion,” ms., New York University. Ghosh, R. (2001) Some Aspects of Determiner Phrase in Bangla and Asamiya, Doctoral dissertation, Tezpur University, Assam. Greenberg, J.H. (1966) “Some Universals of Grammar with Particular Reference to the Order of Meaningful Elements,” in J.H. Greenberg (ed.) Universals of Language, second edition, The MIT Press, Cambridge, Mass., 73-113. Grevisse, M. and A. Goosse (2011) Le Bon Usage. 75 Ans, De Boeck Duculot, Brussels. Halliday, M.A.K. and R. Hasan (1976) Cohesion in English, Longman, London. Harris, A.C. (2006) “Revisiting Anaphoric Islands,” Language, 82, 114-130. Heim, I. (1987) “Where Does the Definiteness Restriction Apply? Evidence from the Definiteness of Variables,” in E.J. Reuland and A.G.B. ter Meulen (eds.) The Representation of (In)definiteness, The MIT Press, Cambridge, Mass., 21-42. Hendrick, R. (1990) “Operator Binding in NP,” in A. Halperin (ed.) Proceedings of the Ninth West Coast Conference on Formal Linguistics, University of Chicago Press, Chicago. Her, O.-S. and H.-C. Tsai (2015) “On Silent Elements: a Case Study of Grand and its Silent Entourage,” Natural Language and Linguistic Theory, 33, 575-605. Ihsane, T. (2008) The Layered DP. Form and Meaning of French Indefinites, John Benjamins, Amsterdam. Jespersen, O. (1961) A Modern English Grammar on Historical Principles. Part II. Syntax. First Volume, George Allen & Unwin, London and Ejnar Munksgaard, Copenhagen. Julien, M. (2002) Syntactic Heads and Word Formation, Oxford University Press, New York. Kayne, R.S. (1996) “Microparametric Syntax. Some Introductory Remarks” in J.R. Black and V. Motapanyane, eds., Microparametric Syntax and Dialect Variation, Benjamins, Amsterdam, ix-xviii (reprinted in Parameters and Universals). Kayne 30 Kayne, R.S. (2002a) “On Some Prepositions That Look DP-internal: English of and French de”, Catalan Journal of Linguistics, 1, 71-115 (reprinted in Movement and Silence). Kayne, R.S. (2002b) “Pronouns and Their Antecedents” in S. Epstein and D. Seely (eds.), Derivation and Explanation in the Minimalist Program, Blackwell, Malden, Mass., 133-166 (reprinted in (2005b) Movement and Silence). Kayne, R.S. (2003) “Silent Years, Silent Hours”, in L.-O. Delsing et al. (eds.) Grammar in Focus. Festschrift for Christer Platzack. Volume 2, Wallin and Dalholm, Lund, 209-226 (reprinted in Kayne (2005)). Kayne, R.S. (2004) “Here and There” in C. Leclère, E. Laporte, M. Piot & M. Silberztein (eds.) Syntax, Lexis & Lexicon-Grammar. Papers in Honour of Maurice Gross, John Benjamins, Amsterdam, 253-273 (reprinted in (2005b) Movement and Silence). Kayne, R.S. (2005a) “On the Syntax of Quantity in English”, in (2005b) Movement and Silence (also in J. Bayer, T. Bhattacharya and M. T. Hany Babu (eds.), Linguistic Theory and South-Asian Languages. Essays in Honour of K.A. Jayaseelan, John Benjamins, Amsterdam. Kayne, R.S. (2005b) Movement and Silence, Oxford University Press, New York. Kayne, R.S. (2006) “On Parameters and on Principles of Pronunciation”, in H. Broekhuis, N. Corver, R. Huybregts, U. Kleinhenz and J. Koster (eds.) Organizing Grammar. Linguistic Studies in Honor of Henk van Riemsdijk, Mouton de Gruyter, Berlin, 289-299 (reprinted in (2010c) Comparisons and Contrasts). Kayne, R.S. (2007a) “A Short Note on where vs. place”, in R. Maschi, N. Penello and P. Rizzolatti (eds.) Miscellanea di Studi Linguistici offerti a Laura Vanelli da amici e allievi padovani, Forum, Udine, 245-257 (reprinted in Comparisons and Contrasts). Kayne, R..S. (2008a) “Some Preliminary Comparative Remarks on French and Italian Definite Articles”, in R. Freidin, C.P. Otero and M.L. Zubizarreta (eds.) Foundational Issues in Linguistic Theory. Essays in Honor of Jean-Roger Vergnaud, MIT Press, Cambridge, Mass., 291-321 (reprinted in (2010c) Comparisons and Contrasts). Kayne, R.S. (2008b) “Antisymmetry and the Lexicon,” Linguistic Variation Yearbook, 8, 1-31 (also in A.M. di Sciullo and C. Boeckx (eds.) The Biolinguistic Enterprise: New Perspectives on the Evolution and Nature of the Human Language Faculty, Oxford University Press, London, 329-353 (2011)) (reprinted in Comparisons and Contrasts). Kayne, R.S. (2010a) “Why Isn’t This a Complementizer?”, in (2010c) Comparisons and Contrasts (also in P. Svenonius (ed.) Functional Structure from Top to Toe: A Festschrift for Tarald Taraldsen, Oxford University Press, New York, 2014). Kayne, R.S. (2010b) “The DP-Internal Origin of Datives”, talk given at the 4th European Dialect Syntax Workshop in Donostia/San Sebastián. Kayne, R.S. (2010c) Comparisons and Contrasts, Oxford University Press, New York. Kayne, R.S. (2013) “Comparative Syntax”, Lingua, 130, 132-151. Kayne, R.S. (2014) “Once and Twice”, in Inquiries into Linguistic Theory and Language Acquisition. Papers offered to Adriana Belletti, C. Contemori and L. Dal Pozzo (eds.), Siena, CISCL Press (also in Studies in Chinese Linguistics (2015) 36, 1-20). Koopman, H. and A. Szabolcsi (2000) Verbal Complexes, The MIT Press, Cambridge, Mass. Kornfilt, J. (1997) Turkish, Routledge, London. Kayne 31 Kortmann, B., T. Herrmann, L. Pietsch, and S. Wagner (eds.) A Comparative Grammar of British English Dialects. Agreement, Gender, Relative Clauses, Mouton de Gruyter, Berlin. Lakoff, G. (1970) “Global Rules,” Language, 46, 627-639. Leu, T. (2005) “Something invisible in English,” UPenn Working Papers in Linguistics 11.1 (Proceedings of PLC 28), p.143-155. Leu, T. (2007) “These HERE demonstratives,” in T. Scheffler, J. Tauberer, A. Eilam and L. Mayol (eds.) Proceedings of the 30th Annual Penn Linguistics Colloquium, Penn Working Papers in Linguistics vol. 13.1, 141-154. Leu, T. (2008) “What for Internally,” Syntax, 11, 1-25. Leu, T. (2015) The Architecture of Determiners, Oxford University Press, New York. Liao, W.-W. R. (2015) “Once upon an Invisible TIME: On Frequentative Phrases in Chinese,” Studies in Chinese Linguistics, 36, 21-33. Llombart-Huesca, A. (2002) “Anaphoric One and NP-Ellipsis,” Studia Linguistica, 56, 59-89. Longobardi, G. (1994) “Reference and proper names,” Linguistic Inquiry 25, 609-665. Malamud, S.A. (2013) “(In)definiteness-driven Typology of Arbitrary Items,” Lingua, 126, 1-31. Marušič, F. a nd R. Ža uce r (2006) “The ‘de finite a rticle ’TA in colloquia l S love nia n,” Formal Approaches to Slavic Linguistics, 14, 189-204. Matthews, S. and V. Yip (1994) Cantonese. A Comprehensive Grammar, Routledge, London. McDonald, C. and J.C. Beal (1987) “Modal Verbs in Tyneside English,” Journal of the Atlantic Provinces Linguistic Association, 9, 43-55. Payne, J, G.K. Pullum, B.C. Scholz and E. Berlage (2013) “Anaphoric One and Its Implications,” Language, 89, 794-829. Perlmutter, D.M. (1970) “On the Article in English,” in M. Bierwisch and K.E. Heidolph (eds.) Progress in Linguistics, Mouton, The Hague, 233-248. Pesetsky, D. (1991) “Zero Syntax. Volume 2: Infinitives,” ms., M.I.T. Pollock, J.-Y. (1998) “On the Syntax of Subnominal Clitics: Cliticization and Ellipsis,” Syntax, 1, 300-330. Postal, P.M. (1966) “On So-Called ‘Pronouns’ in English,” in F.P. Dineen, edl, Report of the Seventeenth Annual Roundtable Meeting on Linguistics and Language Studies, Georgetown University Press, Washington, D.C., 177-206 (reprinted in Reibel, D.A. and S.A. Schane, eds., Modern Studies in English (1969), Prentice-Hall, Englewood Cliffs, New Jersey). Postal, P. (1969) “Anaphoric Islands,” Chicago Linguistic Society, 5, 205-239. Riemsdijk, H. van (2005) “Silent nouns and the spurious indefinite article in Dutch,” in M. Vulchanova and T. Åfarli, Grammar & Beyond: Essays in honour of Lars Hellan, 163–178, Novus Press, Oslo. Schibsbye, K. (1970) A Modern English Grammar. Second Edition, Oxford University Press, London. Shlonsky, U. (2004) “The Form of Semitic Noun Phrases,” Lingua, 114, 1465-1526. Simpson, A. and P. Biswas (2015) “Bare Nominals, Classifiers and the Representation of Definiteness in Bangla,” handout of talk presented at FASAL-5, Yale University. Simpson, A., H.L. Soh and H. Nomoto (2011) “Bare Classifiers and Definiteness. A Cross-linguistic Investigation,” Studies in Language, 35, 168-193. Kayne 32 Stirling, L. and R. Huddleston (2002) “Deixis and Anaphora,” in R. Huddleston and G.K. Pullum (eds.) The Cambridge Grammar of the English Language, Cambridge University Press, Cambridge, United Kingdom, 1449-1564. Tat, D. (2011) “APs as Reduced Relatives: The Case of Bir in Some Varieties of Turkic,” Proceedings of WAFL7, MIT Working Papers in Linguistics, 62. Troseth, E. (2009) “Degree Inversion and Negative Intensifier Inversion in the English DP,” The Linguistic Review, 26, 37-65. Wood, J.L. (2002) “Much about Such,” Studia Linguistica, 56, 91-115. Wood, J.L. (2013) “Indefinite Determiner Doubling,” abstract of talk presented at the Germanic Linguistics Annual Conference, 19, SUNY Buffalo. Zanuttini, R. and L.R. Horn (2014) Micro-Syntactic Variation in North American English, Oxford University Press, New York. Zepeda, O. (1983) A Papago Grammar, University of Arizona Press, Tucson.