Scholarly article on topic 'Once and Twice'

Once and Twice Academic research paper on "Languages and literature"

Share paper
Academic journal
Studies in Chinese Linguistics
OECD Field of science

Academic research paper on topic "Once and Twice"


Once and Twice

Richard S. Kayne New York University

Studies in Chinese Linguistics

t i if x m &


The study of English once and twice yields evidence that each of them is actually a complex phrase containing two visible morphemes and one silent one. Neither is a simple lexical item. The -ce morpheme is akin to a postposition, despite English being primarily prepositional. The silent element associated with once and twice is a silent counterpart of time, represented as TIME. This instance of TIME is singular, even in the case of twice. There appears to be a link between TIME and the syntax of classifiers.

The presence of silent TIME with once and twice indirectly provides evidence for the presence in the human language faculty of other antecedentless silent nominal elements such as NUMBER. Silent elements of this sort are not visible (even via an antecedent) in the primary data available to the learner. Their properties, for example, their singularity or plurality and their licensing conditions, therefore provide us with a privileged window onto the invariant core of the language faculty itself.

The presence of silent elements such as TIME and NUMBER can, in part, be traced back to a principle of decompositionality, to the effect that the human language faculty imposes a maximum of one interpretable syntactic feature per lexical item.


time, silent, classifier, decompositionality, interpretable

Studies in Chinese Linguistics, Volume 36, Number 1, 2015, 1-20 DOI: 10.1515/scl-2015-0001 ©2015 by T.T. Ng Chinese Language Research Centre, Institute of Chinese Studies, The Chinese University of Hong Kong

1. Introduction

Although the correspondence is obviously close between (1) and (2),

(1) We've been there only one time.

(2) We've been there only once.

questions arise as to the way in which the grammar of speakers of English expresses this correspondence. It seems clear that once contains one as a proper subpart; if so, what is the status of the suffixal -ce? It seems equally clear that once should be associated with 'time', as in the corresponding one time; if so, and if this association is expressed through the presence, with once, of a silent counterpart of time, namely TIME (I will use capitals for silent elements), what are the properties of TIME?

In this paper, I will pursue the idea that this -ce is akin to a postposition, and I will consider evidence that suggests that TIME here must be singular. The latter suggestion is of course straightforward for once, less so, and therefore more interesting, for twice, to which I will return later on.

2. -ce as a postposition

The parallelism between (1) and (2) that immediately supports the idea that once contains one, i.e., that once = 'one + -ce', involves sentences in which one and once have a numeral-like interpretation. But the parallelism also holds for cases in which one and once are not felt to be numeral-like, such as:

(3) We were young once

with destressed once, and

(4) We were young at one time

with destressed at one time.

The pair of examples (3) and (4) show in addition that once can, at least in some cases, correspond to a PP (with P = at, in this case). Correspondence with a PP (with P = on) also holds between example (2) and the following:

(5) We've been there on only one occasion.

This correspondence with PPs is part of the reason that I will take suffixal -ce to be a P (other reasons will follow further on). I will call this -ce a postposition simply because -ce ends up being preceded by one, in the case of once. (By antisymmetry, if -ce is a projecting head, one cannot be in the complement position of -ce) As for the interpretive contribution of -ce, it may be neutral between that of temporal at, as in (4), and temporal on, as in (5); alternatively, or in addition, thinking of its apparent origin as an adverbial genitive,1 -ce may be related to the, for me, archaic of found in of an evening.

1 For discussion, see Jespersen (1961, sect. 18.1).

If we now combine the idea that -ce is a postposition with the idea that silent TIME is present, we have as a fuller representation for once :2

(6) one TIME -ce

In (6), 'one TIME' is an indefinite phrase. The postposition -ce can also be preceded by a definite phrase:

(7) You might help us just this once.

With a definite article, the result is more 'special', but possible in at least some English:

(8) They helped us just the once. The representation for (7) and (8) is:

(9) this/the one TIME -ce

That -ce is postpositional, in addition to being suggested by the parallelism with temporal at, on, and of, is further suggested by certain discrepancies in behavior between once and one time. One discrepancy is found in relative clause contexts:

(10) They told us about the one time they thought they were really in danger.

(11) *They told us about the once they thought they were really in danger.

Despite the possibility of (8), example (11) is appreciably worse, if not completely impossible. The reason, I think, is that the 'head' of a relative clause cannot be a PP, whether the P is a postposition or a preposition. This is illustrated by the contrast between (10) and the following:3

2 In some cases, the postpositional phrase with -ce can combine with a preposition:

(i) We'll do it at once.

(ii) For once, they're telling the truth.

Note that the at of (i) is not exactly the same as the at of:

(iii) At one time, they were in agreement with us. which seems closer to -ce itself:

(iv) Once, they were in agreement with us. The at of (i) seems more like the in of:

(v) We'll do it in a/one second/minute.

though the following contrast will need to be accounted for:

(vi) We'll do it in two seconds/*at twice.

3 Cf. also:

(i) The *(place) under the bed where they're hiding is well-concealed. Related to the text discussion is:

(ii) For every two times you make a contribution,...

(iii) *For every twice you make a contribution,...

with the P of twice incompatible with what is probably a relative clause context. Similarly.

(iv) We liked that film the first two times/*the first twice (we saw it).

(12) *They told us about the at one time they thought they were really in danger. Similarly, one has:

(13) Now I've met the two people you were telling me about.

(14) *Now I've met the about two people you were telling me.

Furthermore, if Amritavalli and Jayaseelan (2003) are correct to take adjectives to be K(ase)Ps (and if KPs are akin to PPs), then (10) vs. (12) and (13) vs. (14) are paralleled by:

(15) You're not the genius your sister is.

(16) *You're not the intelligent your sister is.4

The restriction seen in (12), (14), and (16) could be stated as a requirement that the and other determiners should not take a PP/KP as their complement (or as the Spec of their complement, from the perspective of a raising analysis of relatives), though one would hope to be able to go deeper than that. In any event, it seems likely that the restriction in question, however ultimately understood, will carry over to (11), if once is a PP (or perhaps a KP), i.e., if -ce is a P (or perhaps a K).

It should be noted that if we 'undo' the relative clause in (10), we get:

(17) They thought they were really in danger (at) that time. in which an at can be pronounced, in a way that recalls:

(18) They thought they were really in danger on that occasion.

In other words, (10) probably contains a silent P associated with one time. If so, then either the restriction seen in (12), (14), and (16) must not come into play with

4 Although comparatives share important properties with relatives (v. Chomsky (1977)), there are significant differences, e.g.:

(i) You're not as intelligent as/*that your sister is. In:

(ii) the most intelligent that you've ever been

(iii) the fastest that you've ever run

the adjective or adverb, which is not the target of relativization, has been pied-piped by the (non-PP) superlative morpheme, which is. To some extent, superlative -est can pied-pipe a PP, though:

(iv) ?the most on the ball that he's ever been in a way that indirectly recalls Spanish lo in:

(v) Lo a la ligera que hablaría,...

with the PP a la ligera ('at the light = lightly, cavalierly') pied-piped by lo (example from Alvarez (1999, 3752)).

Although woody and wood-like seem close, they differ in a way that needs to be accounted for:

(iv) woodier, woodiest

(v) *wood-liker, *wood-likest

silent Ps, or, more likely, the silent P in (10) has been stranded within the relative clause.5

The PP character of once, with P = -ce, is also relevant to the following contrast, I think:

(19) He's going to be just a one-time champion.

(20) *He's going to be just a once champion.

The idea is that compound-like phrases such as one-time champion disallow Ps, as seen in:

(21) He's a former champion.

(22) *He's an at one time champion.

where (22) also contrasts with the non-compound-like:

(23) He was a champion at one time.

In other words, (20) is excluded parallel to (22), supporting the proposal that once contains a P.

3. Twice

The facts of the preceding paragraph are mimicked, to my ear, by corresponding facts with two time(s) vs. twice:6

(24) He's going to be just a two-time champion.

(25) *He's going to be just a twice champion.

suggesting, not surprisingly, that the -ce of twice has the same postpositional status as the -ce of once. Put another way, (25), like (20) and (22), runs afoul of the restriction barring PPs from appearing within compounds.7

5 The stranding of a silent P may also be at issue in:

(i) a five-thousand dollar car if:

(ii) This car is just $5000. contains a silent AT.

6 And similarly for archaic thrice:

(i) He's going to be just a three-time champion.

(ii) *He's going to be just a thrice champion.

whose -ce is certainly the same morpheme as the -ce of once and twice.

Despite being archaic relative to my English, thrice displays differential behavior in:

(iii) ?They were thrice criticized.

(iv) *They were criticized thrice.

7 Note the contrast with the non-compound example: (i) Twice winner of the Open, Mary...

It is worth noting that the fact that time in (24) must be singular:8

(26) *He's going to be just a two-times champion.

reflects a widespread restriction (in my English) concerning 'compounds', e.g.:

(27) You're an avid newspaper(*s)-reader, I see.

Moreover, the very fact that two-time is possible in (24) leads to the possibility that the silent TIME associated with twice (exactly as TIME is associated with once) is actually singular rather than plural; in other words, twice might have the representation:9

(28) twi- TIME -ce

I will return to this question shortly.

Before doing so, let me note that twice also mimics once with respect to the relative clause facts of (10) and (11):

(29) They told us about the two times they thought they were really in danger.

(30) *They told us about the twice they thought they were really in danger.

As in that earlier discussion, the proposal is that (30) is excluded because the 'head' of a relative clause cannot be a PP, which twice is, as in (28), with P = '-ce'. (Again as in the earlier discussion, if (29) contains a silent P, then either the

8 I have the impression that at least some British English allows -s in some such cases more readily than my English does. The details of this cross-English difference need looking into.

9 Not important for the present discussion (though interesting in its own right) is the question whether the 'twi-' here is one morpheme or two. The same question arises with twin, twenty, twelve, two, and between. What is clear is that the 'tw-' of twice is identical to the 'tw-' of the other forms. (The non-pronunciation of the 'w' of two is in all likelihood predictable from general properties of English phonology.)

Worth noting is that the very close link between twice and two times is not limited to cases in which time is akin to occasion, given:

(i) This car is worth at least two times/twice what that car is worth.

(ii) This car is two times/twice as valuable as that one. Gathercole (1981) has noted:

(iii) John is two times/*twice older than his son.

Her proposal in terms of contraction (and rightward movement) has a problem with:

(iv) He's older than his son. Alternatively, there's a link to:

(v) John is older than his son by ??two times/*twice.

which may be due to the necessary presence of a postposition here with twice, but not with two times.

On the other hand, we have:

(vi) Nobody should two-time their spouse.

(vii) *Nobody should twice their spouse.

arguably because two time as a (rather complex) verb has an analysis involving 'two N at a time', in which two is not a modifier of time, but rather of a silent N that may be classifier-like (PERSON) in a way comparable with TIME, as discussed later.

restriction in question fails to apply to silent Ps, or, more likely, the silent P in (29) has been stranded.)

4. The singularity of TIME.

Coming back now to the question of singular TIME in (28), we can note that its being a component of twice (as opposed to plural TIMES being a component of twice) receives support from:

(31) Two times are enough.

(32) Two times is enough.

(33) *Twice are enough.

(34) Twice is enough.

Two times allows plural agreement in such sentences (in addition to allowing singular agreement). Twice, on the other hand, allows only singular agreement. This must reflect the fact that twice contains singular TIME, as in (28), and that twice cannot contain plural TIMES.10

A further consideration pointing in the direction of singular TIME for twice, rather than plural TIMES, comes from facts related to those discussed earlier in (7) and (8) concerning (just) this/the once. An initial complication, however, arises from the fact that my English strongly resists combining twice with a definite determiner. I do not accept the following, though I have seen written examples of this sort:

(35) *You should have done it just the twice.

More important, though, for the present discussion are comparable examples with demonstratives. I find the following contrast:

(36) *?You could have done it just that twice.

(37) *You could have done it just those twice.

with the singular demonstrative not quite as bad as the plural demonstrative, in a way that gives comfort to the view that twice contains singular TIME.

A second, rather interesting, complication arises when we consider other instances in (some kinds of) English in which a numeral takes a singular noun. Here I have in mind (monetary) phrases like five pound, which are not possible for me, but are possible for Neil Myler (p.c.), who has the following set of judgments:

10 Note the contrast with:

(i) Fifty head of cattle are enough.

in which the plural verb is presumably keyed to the plural lexical noun cattle, no counterpart of which is present with twice.

(38) Five pounds are/is enough.

With plural pounds, either plural or singular agreement is possible for him (as well as for me) in this kind of sentence. Whereas with singular pound, he has:

(39) Five pound is enough.

(40) *?Five pound are enough.

The fact that five pound for him favors singular agreement here is of some interest. Of even more interest to the present discussion is the fact that he finds (40) slightly less bad than (33), which tried to have plural agreement with subject twice. This difference for him between *Twice are... and *?Five pound are... may be related to his accepting:

(41) He'd better give us back those five pound by next week.

in which, in the presence of a numeral, singular pound is compatible with plural those. Yet for him a plural demonstrative with twice is marginal (for me, those twice is sharply out, as in (37)):

(42) ?(?)We could have agreed (just) those twice.

Moreover, adding those to a sentence like (40) appreciably improves, for him, the status of plural verb agreement:

(43) ?Those five pound are enough (to buy lunch with).

This improvement, i.e., the contrast for him between (43) and (40), recalls phenomena discussed in Collins and Postal (2012), den Dikken (2001), Kayne (1972), and Pesetsky (2014). Adapted to (43), the proposals in those works suggest the following (for the relevant speakers). In (39), the phrase five pound contains no plural morpheme at all.11 A plural morpheme can, however, be introduced above five pound, if a demonstrative is merged, too. That allows (41) and also (43) (though why (43) is not perfectly acceptable remains to be accounted for). Only very marginally can a plural morpheme be introduced above five pound even in the absence of a demonstrative, to yield (40).

We can now return to the comparison between five pound and twice (with the analysis 'twi TIME -ce'), both of which contain a singular noun in the context of a numeral. The question is why (33), repeated here:

(44) *Twice are enough.

is worse than (40). A possible answer is that the contrast can be traced back to the difference between the silence of TIME and the non-silence of pound. Thinking of Kayne's (2006) proposal that silent elements are never in exactly the same position

11 Presumably, this is equally true of compound-like examples such as: (i) a five-pound book

that their overt counterparts end up in, it may be that TIME, in the case of twice, actually occurs preceding twi-, i.e., that (28) should be replaced by:12

(45) TIME twi- -ce

If so, then the following comes to mind. The (very marginal) merger of the (silent) plural morpheme above five pound in (40) that yields (very marginal) plural agreement is available only if the numeral precedes (is higher than) the noun. Since twi- does not precede TIME in (45), that merger is blocked, yielding the sharper unacceptability of (44).

This must hold in sentences like (44) in which twice is not associated with a demonstrative. When a demonstrative is present, Neil Myler (p.c.) to some extent accepts:

(46) ?Those twice were enough.

indicating much as before that the demonstrative by itself is, with some degree of marginality, sufficient to license a higher plural morpheme even with twice.

5. The importance of being antecedentless.

TIME is necessarily singular in (45). In (46), a plural morpheme has been merged high in a way dependent on the demonstrative. But TIME itself remains singular even in (46), in a way exactly parallel to the way in which pound remains singular in those five pound in (41) and (43).

TIME is necessarily singular in (45), i.e., when it is a subcomponent of twice. This, however, cannot be a general property of silent TIME as is strongly suggested by the following examples, which are by and large well-formed:

(47) Mary's seen it four times and John five.

(48) We've already been there three times, but we're planning to go another four.

(49) You scolded him three times; (the first) two were enough.

which contrast sharply, in effect, with (44). That is, there is every reason to think that the silent noun in (47)-(49) is plural, just as silent nouns can in general be plural in such contexts:

(50) Mary has written four papers this year, but John has written only three.

(51) Four people I know are interested in your paper, but two are not.

(Note, in particular, the plural agreement licensed by the silent noun in the second part of (51) and (49), again contrasting with (44).)

12 Possibly, TIME has reached the position preceding twi- via movement.

Sentences like (47)-(51), by showing that the language faculty allows for silent plurals (including plural TIMES), make even more pointed the question why twice must contain singular TIME. The key difference would seem to be that the silent plural TIMES of (47)-(49) has an antecedent, namely (overt) times. Whereas the silent singular TIME of twice does not have any antecedent.

6. Classifiers

Continuing to think in terms of 'silent elements' rather than in terms of 'deletion', to keep open the possibility that Kayne (2006) was correct to deny the existence of deletion operations, we might be tempted to formulate a proposal to the effect that a silent plural is licensable only via an antecedent. This does not seem right, however, given often-noted sentences like:

(52) The very poor are in need of help.

in which the plural verb form indicates the presence of a silent (antecedentless) plural noun.13

The absence of an antecedent for the silent TIME of twice is therefore not sufficient to account for its obligatory singularity.14

Thinking of our earlier discussion centering on (39)-(43) of the (partial) parallelism between twice and five pound, with singular pound, it seems likely that a(nother) relevant factor distinguishing twice from the very poor is the presence within twice (and within once) of a numeral. The relevance of the numeral subpart of (once and) twice is brought out by the following consideration. Although the possibility of having a singular noun with a numeral in sentences like (39)-(43) is limited to some varieties of English (not including mine), much more widespread (and perhaps pan-English) is the possibility of numeral + singular noun in compoundlike structures such as in:

(53) They're caught up in a three-year old quarrel.

(54) That three-year old quarrel of theirs has got to stop. At least in my English, a singular here is the only option:

(55) *They're caught up in a three-years old quarrel.

(56) *That three years-old quarrel of theirs has got to stop. Yet I accept:

13 Note the difference between the very poor and: (i) They have two four-year olds.

in which the plural -s is not silent, even in the presence of a silent N.

14 This point is reinforced by:

(i) Three times are enough, whereas twice is/*are not.

(57) They're caught up in a years old quarrel.

(58) That years old quarrel of theirs has got to stop.

with the interpretation that the quarrel in question is quite a number of years old. This interpretation disappears if plural years here is replaced by singular year:

(59) They're caught up in a year-old quarrel.

(60) That year-old quarrel of theirs has got to stop.

In these, with year-old quarrel, the quarrel must be only one year old.

I conclude, then, that singular year in (53) and (54) is licensed in my English in part by the compound-like structure (to distinguish (53) and (54) from (39)-(43)), but also in part by the preceding numeral, to allow (53) and (54) while prohibiting (59) and (60) from having the interpretation of (57) and (58). This conclusion, combined with the parallelism between twice and five pound (and now with three year), leads in turn to the following proposal:

(61) A necessary condition for silent TIME in twice and once is the presence of the

numeral itself (two, one).

If we now ask why (61) should hold, we are led, I think, to (numeral) classifiers.

The reason is that some languages clearly show that (a noun corresponding to) time has classifier-like behavior even when in English one would have thought it an ordinary (non-classifier-like) noun. This classifier-like behavior of time is discussed in recent work by Cinque (2013) and Simpson (2005), most strikingly for Thai and Khmer, which normally have 'N Num Clf' order, yet with numeral + 'time' have the order 'Num time', as if 'time' itself is a classifier, rather than the order 'time Num'.15

These considerations lead, then, to:16

(62) Antecedentless silent TIME is necessarily classifier-like.

which converges with the proposal in Kayne (2003a) that the silent YEAR found in English in:

(63) At the age of seven, Mary could already speak three languages.

is a classifier. If TIME and YEAR in twice and in (63) are classifiers, and if

15 In English, overt time would appear to fairly straightforwardly act in a classifier-like fashion for those speakers (myself not included) who accept sometime else.

16 Consideration of the question whether all antecedentless silent nouns must be classifierlike is beyond the scope of this paper.

classifiers are universally not pluralizable,17 then it will follow that TIME and YEAR in these cases must be singular, as argued earlier for TIME (and as suggested in Kayne (2003a) for YEAR).

7. Licensing conditions

Antecedentless silent TIME is not always licensed in the presence of a numeral:

(64) Mary is a two*(-time) Olympic champion.

Not surprisingly now, a parallel restriction holds for antecedentless silent YEAR:

(65) John's seven*(-year) stretch in prison is coming to an end.

Comparing (65) with (63), one might think that a left-branch-type restriction is at issue, with YEAR impossible in (65) by virtue of being contained within a left branch (and similarly for TIME in (64)). However, further evidence casts doubt on the viability of a left-branch restriction.

Consider this baseball-related example:

(66) The Yankees won the game with two home runs in the seventh (inning). This contrasts with:

(67) The Yankees won the game with two seventh *(inning) home runs.

in which inning is not allowed to remain silent. The restriction seen in (67) might again appear to be a kind of 'left-branch' constraint, but that cannot be exactly right, given the quite acceptable:18

(68) The Yankees won the game with two top of the seventh home runs.

in which silent INNING is much more readily available than in (67). It seems, instead, that silent INNING is favored by the greater amount of syntactic structure associated with 'top of the seventh INNING' in (68) as compared with just 'seventh INNING' in (67).

This in turn is reminiscent of the well-known pair:

(69) John criticized him.

(70) John criticized himself.

Kayne (2002) proposed, as part of an attempt to account for the existence of reflexives

17 At least classifiers of this sort. For some apparent exceptions to the general statement, see Aikhenvald (2000, 249n). Relevant here is the contrast within English between:

(i) two hundred head of cattle and

(ii) two hundred piece*(s) of furniture

18 The word top in this example modifies a silent counterpart of half: (i) two top HALF of the seventh INNING home runs

in the language faculty, that the extra DP structure associated with self provides an additional (A-bar- like) position in (70) that John can avail itself of in the course of moving from within the complex doubling DP containing him (but not self) up to the subject theta position associated with criticize. In partially similar fashion, we can now take top of the seventh in (68) to make available to INNING a specifier position not available to it in (67), with that specifier position a necessary component of the derivational silence of INNING, along the lines of Kayne (2006).

In the same way, TIME in (64) and YEAR in (65), by virtue of not having access to the required specifier position, will fail to be licensed.

Returning to twice, it must now be the case that the silent TIME that is part of twice does have access to an appropriate specifier position, presumably one whose presence is made available by the presence of the postposition -ce.19

I note in passing that a rather different kind of licensing question arises if we ask why once and twice by and large lack (in contemporary English) a counterpart based on three, i.e., if we ask why thrice has become archaic, and if we further ask why no English (that I know of) has ever had a counterpart of once or twice based on a numeral higher than three. There must in all likelihood be a link to the fact that one, two, and three are also special in English in having the corresponding ordinals first, second, and third, rather than the usual ordinal formation with suffixal -th, as in fourth and higher.20 (The fact that thrice has become archaic may be related to the fact that first and second are suppletive,21 whereas third is only partially irregular.) In a more general way, all of this must be connected to the widely attested special behavior of low numerals,22 but I will not pursue this question any further.

19 Cf. the effect of P on French relative pronouns as discussed in Kayne (1994, sect. 8.2); also the effect of P on Italian reflexives discussed in Kayne (2003b, sect. 13).

20 Apart from higher additive ordinals such as twenty-first, twenty-second. Left open is the contrast between these and (ii):

(i) We've been there twenty-one/two times.

(ii) *We've been there twenty-once/twenty-twice. unless singular TIME is incompatible with 21 or 22, etc.

21 For relevant discussion, see Barbiers (2007), whose interesting proposals concerning *oneth lead to the question what exactly distinguishes it from once.

22 Cf. Pesetsky (2014) and references cited there for recent discussion of Russian Case.

Note in addition that couple and pair, despite their interpretation, cannot mimic two here, insofar as:

(i) They arrived late a couple/?pair of times. have no corresponding:

(ii) *They arrived late (a) couple-ce/pair-ce.

The lack of a counterpart to twice with four and higher is probably crucially mediated by - ce, in particular since YEAR in At the age of seven,... is perfectly compatible with higher numerals. Possibly, the absence of once or twice in French and various other languages reduces to the absence of a postposition with the properties of English -ce; a plausible conjecture would be that a counterpart to -ce will be lacking in any language that otherwise entirely lacks postpositions.

8. More on adpositions and TIME.

The idea suggested two paragraphs back to the effect that TIME with twice is in part (indirectly) licensed by postpositional -ce receives support from other cases of TIME involving adpositions. One striking case has to do with soon. Consider:

(71) We'll be there soon.

which has an interpretation involving time such that soon appears to pick out a certain point or interval of time. Yet adding overt time to soon here yields a sharply unacceptable example in:23

(72) *We'll be there at a soon time. which, however, contrasts with:

(73) ?We'll be there at the soonest time possible.

(74) ?You showed up at too soon a time.

The relative acceptability of (73) and (74) supports taking (71) to contain an instance of silent TIME, as well as a silent AT that will play a role in its licensing.24

The difference between (73) and (74), on the one hand, and (72) on the other recalls the discussion in Kayne (2007) of facts concerning few and number:

(75) John has written (a) few papers this year.

(76) *John has written (a) few number of papers this year.

(77) ?John's the student who's written the fewest number of papers this year.

(78) ??John's written too few a number of papers to qualify for a grant.

in which it was proposed that (75) contains silent NUMBER.

Soon and few are modifiers of time/TIME and number/NUMBER, respectively. For some reason (yet to be discovered),25 soon and few can modify overt time and number only if soon and few are raised sufficiently high in the DP, as can happen with too soon and too few (as shown by the following indefinite article),26 and also with superlatives, as suggested for English by:

23 My English does not allow soontime(s), but there are attestations that may ultimately strengthen the text argument.

24 Silent adpositions might at first glance look very different from Larson (1985), but that would change if KP and PP are indeed close.

Soon itself, whose interpretation is close to that of short in time contexts, may well also be accompanied by silent FROM NOW/THEN, thinking of sentences like: (i) We'll be there a short time from now.

25 Part of the reason might be that time and number, being classifier-like, are high in the DP to begin with.

26 Cf. Hendrick (1990).

(79) They're the best of friends.

(80) *They're good of friends.

and cross-linguistically by the fact that Persian generally has prenominal adjectives only in the case of superlatives.27 If soon and few cannot raise sufficiently high, overt time and number must give way to silent TIME and NUMBER, as in (71) and (75).

9. More on postpositions in English.

English is normally thought of as a prepositional language. Yet if I am correct in taking the - ce of once and twice to be a postposition, then English has at least one postposition. Thinking of Dutch and German,28 there is nothing surprising here. Let me, however, briefly touch on further examples of postpositions in English.

One well-known case is that of:

(81) We have a plan whereby we will read everything a day early.

Whereby here is related to thereby, hereby, therefore, forthwith, whereupon and probably whence, thence and hence, with whereby perhaps being the closest to colloquial English. Somewhat similar is:

(82) His whereabouts are unknown. with about arguably an adposition.29 More surprising, perhaps, is:

(83) We don't have the wherewithal to do it.

in which with is postpositional relative to where. Although wherewithal lends itself to being called 'idiomatic', pieces of an analysis readily come to mind. The -al is all. The definite article in (83) recalls that found overtly with whole, as well as recalling the fact that all is non- initial in:

(84) He gave it his all.

in which there is arguably a silent definite article. In the manner of Dutch and German, wherewith corresponds to with what,30 with the result that (83) can be thought of as very close to:

(85) *We don't have the all with which to do it.

27 Cf. Moshiri (1988, 24) and Ghomeshi (1996, 145).

28 Cf. for example Noonan (2010).

29 The plural here suggests the possibility of a silent PLACE, thinking of: (i) ?The places where he is about are unknown.

30 Cf. van Riemsdijk (1978) and work stemming from his.

even though this sentence is not acceptable. The fact that wherewith precedes al(l) in (83), whereas with which follows all in (85) suggests that in (83) wherewith has raised past al(l) in a way related to the way in which destruct- (remnant-)raises past -ion in the relative clause approach to derived nominals suggested in Collins (2006) and Kayne (2008).

Furthermore, some speakers of English, in particular Bob Frank (p.c.), accept some sentences like:

(86) What about were you guys talking?

(87) Who to are you hoping to talk about that?

(88) Who from are you convinced that John stole the idea?

in which about, to and from look postpositional.31 Possibly, English adpositions are postpositional in the same way in sluicing examples like:

(89) I knew they were talking, but I wasn't sure what about. as is suggested by Bob Frank sharply rejecting:

(90) *What topic about were you guys talking?

with a wh-phrase containing a lexical noun, just as in sluicing:

(91) *I knew they were talking, but I wasn't sure what topic about.

As a final example of an English postposition, we might think of ago, or, more likely, of the a- of ago, especially if the following two sentences are closely related:

(92) They left three days ago.

(93) It's going on three days since they left.

with a- in (92) corresponding to on in (93), with go in (92) corresponding to going in (93), and with three days in (92) preposed to adpositional a- in a way that has something in common with postpositions.32

10. A further instance of TIME.

Alongside (92) one also has:

(94) They left a long time ago.

(95) They left long ago.

31 For Bob Frank, the first of these three is the most fully acceptable.

32 For relevant discussion of postpositions, see Kayne (2003c).

It is hard to see how (95) could fail to contain TIME.33 A related use of long (but one that shows polarity behavior in the absence of overt time) is found in:

(96) You haven't been here very long.

Again, there is presumably a silent TIME. In all likelihood there is also a silent adposition in (96), given the strong similarity to:

(97) You haven't been here for a very long time.

An interesting challenge is to understand why TIME is not compatible with the indefinite article:34

(98) *They left a long ago.

(99) *You haven't been here a very long.

It may be that this is just the same fact—see Kester (1996)—as:

(100) Mary has written a long paper and John has written a long *(paper), too.

Alternatively (or in addition), there is a link to the fact that French longtemps ('long time') is compatible with the absence of an indefinite article:

(101) Marie est restée longtemps à Paris. ('M is remained longtime in P') 11. Conclusion

Both once and twice are complex phrases (containing two visible morphemes and one silent one), rather than simple lexical items. The presence of silent TIME with once and twice (and in other cases mentioned) indirectly reinforces the presence of other antecedentless silent elements in the human language faculty. Since silent elements of this sort are not visible (even via an antecedent) in the primary data available to the learner, study of their properties, for example of their singularity or plurality, and of their licensing conditions, provides us with a privileged window onto the invariant core of the language faculty itself.


Aikhenvald, Y. Alexandra. 2000. Classifiers. A typology of noun categorization devices, New York: Oxford University Press.

33 Similarly, Tsoulas (2013) has argued that before can be followed by TIME; Zamparelli (2004) had suggested TIME for every two days; TIME is clearly called for in the shorter version of:

(i) We'll be there in two hours' (time).

as well as with often, given oftentimes. (Whether often has TIME or TIMES needs to be looked into further.) In addition, Purves (2002, 30) notes that Scots uses this, that and yon for this/that/ yon time/place/person.

34 Another involves:

(i) They left long/*short ago. and, conversely:

(ii) They left shortly/*longly before noon.

Alvarez, Alfredo. 1999. Las construcciones consecutivas. In Ignacio Bosque and Violeta Demonte (eds.), Gramática descriptiva de la lengua española, Volume 3, 3739-3804. Madrid: Espasa.

Amritavalli, Raghavachari & Karattuparambil A. Jayaseelan. 2003. The genesis of syntactic categories and parametric variation. In Huang-Jin Yoon (ed.), Generative grammar in a broader perspective: Proceedings of the 4th GLOW in Asia 2003, 19-41. Hankook: KGGC and Seoul National University.

Barbiers, Sjef. 2007. Indefinite numerals one and many and the cause of ordinal suppletion. Lingua 117. 859-880.

Chomsky, Noam. 1977. On wh-movement. In Peter W. Culicover, Thomas Wasow & Adrian Akmajian (eds.), Formal syntax, 71-132. New York: Academic Press.

Cinque, Guglielmo. 2013. Are all languages 'numeral classifier languages'?. In Guglielmo Cinque (ed.), Typological studies: word Order and Relative Clauses, 107-109. New York: Routledge.

Collins, Chris. 2006. A Note on Derivational Morphology. Ms., New York University.

Collins, Chris. & Paul M. Postal. 2012. Imposters. A study of pronominal agreement. Cambridge, MA: The MIT Press.

Den Dikken, Marcel. 2001. Pluringulars, pronouns, and quirky agreement. The Linguistic Review 18. 19-41.

Gathercole, Virginia C. 1981. Support for a unified QP analysis. Linguistic Inquiry 12. 147-148.

Ghomeshi, Jila. 1996. Projection and inflection: A study of Persian phrase structure. Toronto, Ontario: University of Toronto dissertation.

Hendrick, Randall. 1990. Operator binding in NP. In Aaron Halperin (ed.), Proceedings of the ninth west coast conference on formal linguistics. Chicago: University of Chicago Press.

Jespersen, Otto. 1961. A Modern English grammar on historical principles, part VI. London: George Allen & Unwin and Copenhagen: Ejnar Munksgaard.

Kayne, Richard S. 1972. Subject inversion in French interrogatives. In Jean Casagrande & Bohdan Saciuk (eds.), Generative studies in Romance languages, 70-126. Rowley, Mass: Newbury House.

Kayne, Richard S. 1994. The antisymmetry of syntax. Cambridge, MA.: The MIT Press.

Kayne, Richard S. 2002. Pronouns and their antecedents. In Samuel Epstein & T. Daniel Seely (eds.), Derivation and explanation in the minimalist program, 133166. Malden, Mass: Blackwell. (reprinted in 2005 Movement and Silence).

Kayne, Richard S. 2003a. Silent years, silent hours. In Christer Platzack & Lars-Olof Delsing (eds.), Grammar in focus. Festschrift for Christer Platzack Volume 2, 209-226. Lund: Wallin and Dalholm. (reprinted in 2005 Movement and Silence).

Kayne, Richard S. 2003b. Person morphemes and reflexives in Italian, French and related languages. In Christina Tortora (ed.), The syntax of Italian dialects, 102-136. New York: Oxford University Press. (also in 2000 Parameters and Universals).

Kayne, Richard S. 2003c. Antisymmetry and Japanese. English Linguistics 20, 1-40.

(reprinted in 2005 Movement and Silence).

Kayne, Richard S. 2006. On parameters and on principles of pronunciation. In Hans Broekhuis, Norbert Corver, Riny Huybregts, Ursula Kleinhenz & Jan Koster (eds.), Organizing grammar. Linguistic studies in honor of Henk van Riemsdijk, 289-299. Berlin: Mouton de Gruyter. (reprinted in 2010 Comparisons and Contrasts).

Kayne, Richard S. 2007. On the syntax of quantity in English. In Josef Bayer, Tanmoy Bhattacharya and M. T. Hany Babu (eds.), Linguistic theory and south Asian languages. Essays in Honour of K.A. Jayaseelan, 73-105. Amsterdam: John Benjamins. (also in 2005 Movement and Silence).

Kayne, Richard S. 2008. Antisymmetry and the lexicon. Linguistic Variation Yearbook 8. 1-31. (reprinted in 2010 Comparisons and Contrasts; also in Anna Maria Di Sciullo and Cedric Boeckx (eds.), The biolinguistic enterprise: New perspectives on the evolution and nature of the human language faculty, 329-353. London: Oxford University Press (2011)).

Larson, Richard K. 1985. Bare-NP adverbs. Linguistic Inquiry 1. 595-621.

Moshiri, Leila. 1988. Colloquial Persian. London: Routledge.

Noonan, Maire. 2010. A to Zu. In Guglielmo Cinque & Luigi Rizzi (eds.), Mapping spatial PPs. The cartography of syntactic structures, Volume 6, 161-195. New York: Oxford University Press.

Pesetsky, David. 2014. Russian case morphology and the syntactic categories. Cambridge, MA: The MIT Press.

Purves, David. 2002. A Scots grammar: Scots grammar & usage, Edinburgh: The Saltire Society.

Simpson, Andrew. 2005. Classifiers and DP structure in Southeast Asia. In Guglielmo Cinque & Richard Kayne (eds.), The Oxford handbook of comparative syntax, 806-838. New York: Oxford University Press.

Tsoulas, George. 2013. Before-Clauses and the Syntax-Semantics Interface. Paper presented at the University of Stuttgart.

Van Riemsdijk, Henk. 1978. A case study in syntactic markedness. Lisse: The Peter de Ridder Press.

Zamparelli, Roberto. 2004. Every two days. Snippets 9. 19-20.

Address: Department of Linguistics, New York University,

10 Washington Place #403, New York, NY 10003, USA


Receive date: April 1, 2014

Accept date: April 1, 2014

Once Twice Richard S. Kayne

, ^ffi once m twice

WMB^&io once m twice TIME ^/Tv,»® TIME

fiPf№$ twice

toSl TIME a^fi once ffi twice, ^ffiS^TS^B^ NUMBER ^S^^tIS! toS^^TIME m NUMBER , fiPASlffi