1 Introduction
In the literature on the determiner some one finds a dichotomy between two major types of approach. On the one hand, Israel champions a scalar analysis, arguing that some is a scalar understater (1999) or attenuator (2006), so that a sentence such as Some people enjoy the weather in San Diego ‘makes a weaker claim than one would expect – that most people enjoy the weather in San Diego’ (1999: 177). On the other hand, it is traditional in logic-based analyses to treat some as an existential quantifier, as for example in Partee, ter Meulen & Wall (Reference Partee, Meulen and Wall1990: 138), who state that ‘The universal quantifier . . . is correspondent to English expressions such as all, each and every. The existential quantifier . . . corresponds to some (in the sense of “at least one, possibly more”).’ Farkas (Reference Farkas2002) further specifies that some is an existential quantifier whose variable must be unidentified, as exemplified by uses such as Mary was depressed for some reason which indicates that the speaker doesn't know or won't reveal the nature of the reason for Mary's depression, where the indefinite article would have simply indicated that Mary's depression was not without cause.
As with the opposition between free choice (FC) and negative polarity item (NPI) any, the proponents of these two types of approach are each talking about different uses of some, which makes a direct confrontation between them futile. However, they do have in common the adoption of a quantitative perspective on this determiner: the existential approach explicitly treats some as a quantifier; the scalar analysis views some as evoking a less-than-expected quantity on a pragmatic scale. This raises the question of how to handle uses of some which do not appear to be quantitative, such as (1)–(4) below, and which have generally been passed over or treated as exceptions in the literature:
(1) Some kid spilt a milkshake on the floor.
(2) That was some frittata!
(3) He made some thirty-three snowmen that afternoon.
(4) There were some twenty people present.
Example (1) is not a problem for an existential-quantifier approach such as Farkas' as it denotes an unidentified individual; however, it does represent a clear counterexample to the scalar theory, as there is no impression here that the quantity of kids having spilt a milkshake on the floor is in any way less than expected. Farkas' unidentified-existential approach, on the other hand, runs into difficulty with uses such as (2), where there is no impression of ignorance, indifference or caginess about the identity of the frittata referred to. Utterances like (2) are an even greater problem for the scalar approach which sees some as an attenuator, as the frittata in question is treated in this sentence as being more, and not less, of a frittata than was expected. The use illustrated in (3) poses a similar problem for a scalar treatment in that it also gives the impression that the number of snowmen built was greater, not less, than normal snowman production by one person in one afternoon; and since the exact quantity is specified by the numeral thirty-three, one wonders what an existential quantification signifying ‘at least one, possibly more’ could possibly contribute to the message actually expressed by this sentence. How existential quantification could produce the impression of approximation conveyed by (4) is also a mystery, as is the question of how a scalar attenuator could be compatible with values higher than that denoted by the numeral following some, as shown by the fact that the paraphrase of (4) is ‘more or less twenty’ and not ‘less than twenty’; nor does some convey any impression in this use that the number twenty is less than expected.
The goal of this article will therefore be to try to repair the neglect of the qualitative uses of some and to suggest an explanation which can cover the full range of usage of this determiner – both quantitative and qualitative – showing how a single underlying meaning, modulated by contextual and pragmatic factors, can give rise to the wide variety of messages expressed by some in actual usage. The reason for the neglect of the qualitative uses of one of the most frequent words in English will be suggested to be that they do not square very easily with the assumption that some is a quantifier or a scalar item.
The long-standing treatment of some as a logical quantifier is discussed in the first section of the article, along with approaches that treat some and any as allomorphs of the existential quantifier. The scalar model (Israel Reference Israel1999, Reference Israel, Birner and Ward2006, in press) is considered extensively in the next section. An original analysis of some and the interaction of its meaning with the defining features of the uses in (1)–(4) is then proposed, before extending the discussion to the role of focus and the adverbial modifier quite. What our analysis reveals about the meaning of some, and whether the latter's uses can be explained by a non-scalar model proposing non-identification of a referent assumed to be particular as the crucial semantic feature of some is considered in the conclusion, which summarizes the challenges for a definition of this item that meets the requirements of compositionality.
2 Quantifiers and indefinites
There is a long tradition extending all the way back to the Greeks of applying logical categories to the study of human language. Formal logic pursues the objective of identifying well-formed arguments through the abstract representation of propositions composed of operators and variables, with rules stipulating which inferences are derivable from which premises. Among the class of logical operators, pride of place is occupied by the existential quantifier ∃, which signifies that there is at least one member of a set involved in the proposition and stands in a relation of contrariness to the universal quantifier ∀, which indicates that all the members of a set are concerned by the proposition. These quantifiers are customarily equated with the English determiners some (existential) and each/every/all (universal). As can be seen from the fact that all, each and every are treated all three together as exponents of the very same universal quantifier ∀, the primary goal of logic is not to define the meaning of one linguistic item as opposed to other linguistic items, nor to characterize the meaning of a particular linguistic item in and of itself, but rather to define the logical value of an item's particular use in a proposition. Indeed, as already seen above, treating some as an existential quantifier does not provide any explanation of its use in examples (2)–(4) above: while the meaning of the determiner in Some logicians agree with Frege may be adequately rendered by an existential quantifier, the interpretation of the examples just referred to is not exhausted by the idea that at least one frittata, thirty-three snowmen or twenty people are involved in the propositions corresponding to these sentences.
The logical approach equating some with the existential quantifier thus fails to do justice to the full range of uses of this determiner, its particular distribution and collocations, and how its various uses are related to one another. Such challenges motivated Haspelmath (Reference Haspelmath1997) to avoid making any commitment as to the nature of the relations between the functions of the various indefinite pronouns that he maps in English and other languages. Apart from accounting for the variety of its uses, the semantic characterization of some would need to explain what this item means in itself, and how it is similar to, and yet different from, comparable items. The challenge is such that it is not surprising that it has yet to be met.
This being said, considerable attention has been devoted to exploring the similarities between some and any (see Lees Reference Lees1960; Klima Reference Klima, Fodor and Katz1964; Anderson Reference Anderson1973; Ladusaw Reference Ladusaw1980; Davison Reference Davison and Auwera1980; Israel Reference Israel1999; Huddleston & Pullum et al. Reference Huddleston and Pullum2002). Klima's (Reference Klima, Fodor and Katz1964) highly influential study proposes that some is spelled out in affectiveFootnote 2 contexts as any, which makes it in effect a mere allomorph of an existential archimorpheme. The counterargument was quick to follow (in Lakoff 1969) that there are affective contexts in which some can be used with a nuance of actuality that distinguishes it from any in that it presupposes the existence of the NP's referent:
- (5)
(a) Who wants some coffee?
(b) Who wants any coffee?
- (6)
(a) If you find some mistakes in my translation, let me know.
(b) If you find any mistakes in my translation, let me know.
- (7)
(a) I'm surprised she didn't hire some German musicians.
(b) I'm surprised she didn't hire any German musicians.
Thus to say If you find some mistakes in my translation, don't be angry suggests that ‘the speaker is afraid there will be some mistakes’, whereas with any mistakes ‘the speaker does not know that there are any mistakes’ (see Kruisinga & Erades Reference Kruisinga and Erades1960: 577). Sahlin (Reference Sahlin1979: 29) proposes that the best way to describe the difference in existential import between some and any ‘is to make a distinction between “referential” and “non-referential” NPs’. Such observations have not precluded, however, the subsequent development of approaches such as Postal (Reference Postal, Laporte, Leclère, Piot and Silberztein2004) and Szabolcsi (Reference Szabolcsi2004), which invoke underlying traits to account for the distributional similarities of some and any. Without going into all the technical details, the idea behind these is that the same underlying features are spelled out as any, some or no depending on whether these traits stay in place, move out of the representation or are activated by some external licensor. For instance, Szabolcsi (Reference Szabolcsi2004: 411–12) argues that positive polarity items (PPIs) such as some are in fact double NPIs that simultaneously contain both a strong NPI yet-feature and a weak NPI ever-feature. These features are dormant unless activated by a yet-licensing environment. A yet-licensor such as negation activates both features but licenses only one of them, and so PPIs are excluded from such contexts, which admit only NPIs such as any. The ever-feature requires the presence of a second licenser such as the verb regret for the PPI to be allowed, as in I regret that he didn't say something. While such treatments may achieve the desired distributional predictions, the semantic nuances observed between some and any remain unaccounted for.
More descriptive attempts to situate the meaning of some within the general class of indefinites have also been put forward. Chesterman (Reference Chesterman1991: 182) proposes a scale of definiteness on which some is situated between the zero article, which can be compared to some in usage with plurals and mass nouns, and the indefinite article, which is comparable to some in usage with singulars:
Definiteness is defined (Chesterman Reference Chesterman1991: 2) in terms of three parameters: (i) locatability in a shared set, i.e. familiarity, (ii) quantitative inclusiveness (reference to the totality of locatable referents) and (iii) limited extensivity. Some is distinguished from both zero and the indefinite article by the fact that these two latter items can be generic, whereas some cannot (also noted by Langacker Reference Langacker1991 and Farkas Reference Farkas2002):
- (8)
(a) Printers are quite cheap now.
(b) Some printers are quite cheap now. (≠ generic)
- (9)
(a) A baby requires a lot of attention.
(b) *Some baby requires a lot of attention.
Chesterman compares the non-generic use of zero to some in terms of the opposition between ‘non-quantified’, as in (10) below, and ‘quantified as a non-specific quantity’, as in (11):
(10) I gave her pens.
(11) I gave her some pens.
The inability of some to have generic reference implies, however, that it is not as indefinite as the indefinite article in terms of parameters 1 and 3. This is contrary to the predictions made by Chesterman's scale, on which a is treated as being more definite than some. This may be because only the singular sense of a is referred to in his study, as found in I gave her a pen, which because of the singular noun is quantitatively more definite in reference than the vaguer quantification expressed by the plural noun phrase some pens. A systematic analysis would call for a comparison between some and the indefinite article when both are followed by singular nouns, which is not provided by Chesterman, who discusses some only with mass or plural nouns. We will turn our attention to this comparison in section 4 below when discussing examples such as (1).
The discussion so far seems to point to the value of some as combining elements of both definiteness and indefiniteness, presuming the existence of a particular occurrence but not identifying it. This would set it apart from any and other indefinites, in that unlike any it presupposes the existence of an entity, and would explain its incompatibility with generic interpretations, in that some points to a particular, non-generic but unidentified instance. Whether more is achieved by the scalar analysis proposed by Israel will be considered in the next section.
3 Some and scales
An analysis of the meaning of some in scalar terms has been proposed by Michael Israel, whose overall objective is to account for the phenomenon of polarity sensitivity in natural languages with particular reference to English. This phenomenon is illustrated by any's occurrence in negative and affective contexts, but not normally in assertive ones, in contrast to some's distribution in assertive and affective contexts, but not generally in negative ones (as discussed by Larrivée Reference Larrivée2010, following Szabolcsi Reference Szabolcsi2004).
- (12)
(a) He wanted some coffee.
(b) ?? He wanted any coffee.
- (13)
(a) If you find some mistakes in my translation, let me know.
(b) If you find any mistakes in my translation, let me know.
- (14)
(a) ?? She didn't hire some German musicians.
(b) She didn't hire any German musicians.
This distribution can be explained by the semantic value of the items, polarity-sensitive items such as any encoding a low q(uantitative)-value and a high i(nformative)-value: to take (14) as an example, the quantitative value is low because only a small number of musicians is concerned (‘not even one or two’); the informative value is high because a stronger claim is made with any than with a or two or those – any widens the set beyond particular musicians to include even possible ones. Therefore an emphatic value is involved, which makes these items at home in non-assertive environments that refer to a set of elements ordered from low scalar values to higher ones and thus validate the emphatic value. Contexts where inferences run from high to low such as assertives, on the other hand, validate the understatement expressed by the low i-value of some, which, like any, also has a low q-value. Some is thus a scalar understater (Israel Reference Israel1999) or attenuator (Israel Reference Israel, Birner and Ward2006), so that, as already mentioned above, a sentence such as Some people enjoy the weather in San Diego ‘makes a weaker claim than one would expect – that most people enjoy the weather in San Diego’ (1999: 177). Since some is a low-scalar, attenuating PPI expressing a limited indefinite quantitative value, that explains why it cannot be generic (Israel in press: 373).
This description, however, does not extend to all uses of some. A case comparable to (1), given as (15) below, is treated by Israel (in press: 377) as a separate lexical entry associated only with singular count nouns, although it is considered that understatement is involved here, in that less information is provided than would have been by identifying the person or admitting ignorance of the person's identity.
(15) Apparently some idiot thought this would be a good place for a mall. (Israel in press: 377, example 15c)
Utterances like (2) are also attributed a separate lexical entry, as they are deemed to be ‘radically different’ from all the other uses of some in that the latter ‘forms a definite NP whose referent must be uniquely identifiable’ (Israel Reference Israel1999: 177). We have noted above, however, that they seem the very opposite of attenuation, as the referent in question is treated as being more, and not less, than expected. For similar reasons, the value of some in (3) would also have to be treated as an exception in the scalar approach, as the amount of snowmen built is treated as greater, not less, than normal expectations.
Above and beyond the problems raised in the previous paragraph, the question remains as to whether the understatement perspective can account for the overall weakness of the relation between some and scalarity. Israel (in press: 373) argues that some is the mirror image of any, the latter being a low scalar emphatic NPI and some a low scalar attenuating PPI. Any, however, allows continuations by end-of-scale expressions, as in He can do anything, even sing the Hungarian national anthem in Japanese! This type of behaviour is not observed with some: *He can do something, even that! It has also been claimed (Lee & Horn Reference Lee and Horn1995: 4–5) that it is the scalar status of any that allows it to be complemented by whatsoever: Any kind of paper whatsoever will do. Again, this is not possible with some: *Some kind of paper whatsoever will do. The same difference in behaviour is observed with respect to the scalar particles absolutely and almost, which customarily interact with any, but not with some (see Lee & Horn Reference Lee and Horn1995: 6). This evidence thus suggests that, contrary to Israel's assumptions, some does not in fact inherently represent the lower end of a scale.
This survey of the scalar analysis of some proposed by Israel shows that the assumption of a symmetrical opposition between some and any, each representing a different way of situating a value on a scale, does not provide a satisfactory answer to the question of the distributional behaviour of some as compared to that of any, nor to the question as to why some can be used in utterances such as (2)–(4), which have to be treated as separate lexical entries in a scalar approach, i.e. as exceptions. The precise contextual conditions that govern these qualitative uses is what we will seek to elucidate in the following sections.
4 Some without scales
4.1 Ignorance, indifference and pejoration
The first use to be examined will be the type illustrated by (1) and (15) above. Huddleston & Pullum et al. (Reference Huddleston and Pullum2002: 381) point out that in this ‘vague count singular use’ some is more natural with emotive expressions expressing annoyance or denigration of the entity referred to than is the indefinite article: while Some idiot has left the oven on expresses more the speaker's opinion of the perpetrator of the deed, the same sequence with the indefinite article would depict idiot as an ordinary descriptive noun. Axiology is not a necessary property of the noun occurring after some in this type of use, however:
(16) Apparently some urban planner thought that this would be a good place for a mall.
Neither is the presence of a singular noun, as the following example with a plural seems quite parallel:
(17) Some idiots must have tracked mud into the lobby.
Nonetheless, the pejorative impression does come through more clearly with singular nouns, as (18) below seems less immediately pejorative than (16) (although, if uttered with a certain intonation, denigratory overtones could indeed be conveyed):
(18) Apparently some urban planners thought that this would be a good place for a mall.
This is confirmed by the following attested use in which the tone is entirely appreciative:
(19) I've always thought about what a neat idea it would be if we could have floating public benches for you to enjoy an aerial view. Well apparently, some urban designers, with their ingenious use of current low-tech, have made my daydream into a reality. Check out some of the photos here. Just imagine the public bench of the future! They might not even need staircases any more, just crank up that invisible lever, baby, and enjoy your lunch in midair.
(www.cybernoeticman.com/2011/01/how-design-can-elevate-present-to.html#more)
This variation in the strength of the axiological value calls for an explanation, which will be proposed below.
This type of usage also often involves indirect sources of information, as indicated by the occurrence of the adverb apparently and the epistemic modal must in (16) and (17) above. However, direct evidence is not necessarily incompatible with this interpretation, as can be seen in:
(20) He was looking for some brand of coffee from the Malabar coast.
(21) Tom was reading some book on Dianetics when I came in.
The affinity between this sense of some and epistemic modalities suggests that some denotes an existing and delimited referent whose identity the speaker is not specifying because he does not know it or judges it to be irrelevant. Christophersen (Reference Christophersen1939: 188) notes regarding this type of use that ‘Some emphasizes the total indefiniteness of the notion. Put the case that a man makes a strange assertion and I ask for his source of information: if he says “I have read it in a book,” he may still remember which book it was; if he says “I have read it in some book,” he probably cannot recall the title.’ This sort of use has been labelled ‘spesumptive’ by Warfel (Reference Warfel, Battle and Schweitzer1972: 42), who says that it conveys the idea that ‘the speaker does not have a particular referent in mind and does not assume a particular referent is known by the hearer, but does assume that the subject (of the verb governing some) has a particular referent in mind’. The fact that the referent is expressly left unidentified by the use of some often gives rise to the impression of disparagement referred to above. This impression can be further illustrated by (22) below, in which the unimportance of the referent is reinforced by the use of the adverb just and the epithet jerk:
(22) ‘Do you see a duck?’ ‘Nope, just some jerk holding a rabbit.’ (schedule.blogspot.com/. . ./one-of-my-favorite-pieces-is-turangalla.html)
Here the impression is that it is not even worth the bother to identify the referent, even though the speaker might be able to do so if she wanted to, an impression also observed by Palmer (Reference Palmer1969: 117) and Farkas (Reference Farkas2002: 11). The axiology comes both from the noun and from the pragmatics of the spesumptive use, which is based on the non-identification of a limited existent occurrence. The derogatory flavour is felt more strongly with a singular noun than with a plural because the idea of non-specification denoted by some can be applied both to the number of referents and to their identity with a plural noun, but only to the identity of the referent with a singular, whence the easier accessibility of the implication of indifference or lack of interest with respect to the referent's identity: if one could have indicated identity of the referent, but chose to leave it unspecified by using some, this can convey the message that one considers that it was not worth the trouble to identify the person or thing in question.
4.2 Approximation
The notion of the non-identification of a possibly identifiable referent has obvious potential for the expression of approximation, and indeed some can be used to convey this impression, as can be seen in (4) above and (23) below:
(23) There were some fifty villages that agreed to the plan.
Although this effect is associated with the presence of a numeral, it should be pointed out that the combination some + numeral does not always produce an impression of approximation – in (24) below, as in (3) above, the speaker's intention is not to suggest that he has not counted the precise number of occurrences:
(24) The phrase is used some twenty-nine times in this book, but not elsewhere. . .
(Lancaster–Oslo–Bergen Corpus D03 12 5)
This type of use corresponds rather to the ‘considerable-quantity’ sense (see Huddleston & Pullum et al. Reference Huddleston and Pullum2002: 380), which will be dealt with a bit later on.
When some is used to express approximation, the numeral with which it occurs is usually a round number, i.e. either a simple multiple or a simple fraction of a base number of the decimal system (see Lotz Reference Lotz and Hymes1955), as in (4) and (23) above. In some cases some is combined with other indicators of approximation, such as using two successive numerals joined by or, as in (25):
(25) Some fourteen or fifteen years ago, in an essay I called The Leader Follows – Where? I used his polarity to illustrate what I thought had happened to us in that form of liberalism which we call Progressivism.
(Brown Corpus G21 0260 1)
The impression of approximation can be explained as follows. Some, for its part, evokes a particular but non-identified quantity, as it does in (26) below:
(26) There were some villages that agreed to the plan.
The numeral in (23), for its part, characterizes this quantity as being fifty. The resulting message is to characterize fifty as an approximate figure, by treating it as a number which does not specify the precise quantity referred to. This explains why the numeral following some must be interpretable as a round number, i.e. as a figure that can be interpreted in an approximate manner.
The interplay between some and the numeral in this type of use illustrates both the complexity and the flexibility of the interaction of linguistic meanings: in natural language two quantifications that might appear at first sight to be logically incompatible – the unspecified-quantity meaning evoked by some + a plural/non-count noun, and the precise-quantity meaning denoted by a cardinal numeral – can be combined to convey the notion of quantitative approximation.
4.3 Diminution and augmentation
The unidentified middlish quantity evoked by some with plural and mass nouns, which is exploited in uses like those just discussed in the previous section to convey approximation, can also be contrasted with other quantities, or with the whole category denoted by the noun itself, as has been noted by Israel. The quantity with which some is contrasted can be greater than that expressed by some: thus in Israel's (21a), given as (27) below, the message conveyed is that a better idea of the other person's intended meaning has not yet been achieved, with some denoting a lesser than possible degree of understanding:
(27) I have some idea of what you mean, but it's still a bit hazy.
This accounts for the frequency of the collocations some, but not much (2,870,000 Google hits) and some, but not all (1,970,000 Google hits – searches performed on 5 March 2010).
Because some + plural/mass noun implies a positive, but not necessarily large, amount, it can also be contrasted with zero quantity, in which case some could be argued to express a greater than expected quantity. This is the case in uses such as (28) where the message expressed is ‘at least a little bit’:
(28) There must be some mercy in that guy's heart.
One could perhaps also argue here that there could be a contrast with a lot, which would mean construing (28) like (27), i.e. contrasting the intermediate quantity expressed by some + noun with a larger quantity rather than with zero quantity.
Huddleston & Pullum et al. (Reference Huddleston and Pullum2002: 380) note a different use of stressed some with plural and non-count nouns in which some indicates ‘a quantity significantly above zero’, as in (28) and (29) below:
(29) It was some years before she saw him again.
(30) We discussed the problem at some length.
Here Israel's analysis of some as an attenuator runs into serious difficulty. Huddleston & Pullum et al. point out that quite is often found as a modifier of some in this type of use, and has the same effect as placing accent on some, i.e. that of increasing the quantity denoted (more about the adverb quite below), and that this ‘considerable-quantity’ use is normally restricted to head nouns denoting some kind of measure. The presence of some degree of accent on some is noted, but not explained.
Relevance Theory considerations might be invoked in a case such as this to explain the production of the considerable-quantity impression with some: Sperber & Wilson (Reference Sperber and Wilson1995: 189) analyse the some-phrase in It will take some time to repair your watch as expressing semantically the banally true assertion that a certain amount of time is required to perform the repair, which due to the banality of the literal meaning would be uninformative, and therefore gets reinterpreted pragmatically as an informative assertion that a significant amount of time will be needed to fix the watch. In our view this can only be part of the story, however, as this analysis implies an equivalence between the bare noun time and the NP some time (It will take time to repair your watch also expresses the banally true assertion that a certain amount of time is required to perform the repair). This equivalence is annulled, however, by the existence of contexts where the considerable-quantity sense can only be conveyed by means of a some-phrase and not a bare noun, such as I didn't see her again for some time after that/*for time after that. We would argue that some's meaning of particular, but non-identified, reference is crucial to the production of the considerable-quantity effect. This type of use requires the evocation of a bounded amount of time, which the bare noncount noun is unable to provide all by itself. The role of sentence accent is also important in the production of the considerable-quantity effect: in this type of usage, accent is emphatic rather than contrastive, and has the effect of calling attention to the fact that the unidentified, but bounded, quantity referred to by the some-phrase is a considerable one. This type of use is thus comparable to the accented use of so in utterances like That was SO delicious! The fact that both some and so leave the exact quantity unspecified is exploited, moreover, in such usage to leave the hearer free to imagine however large a quantity he likes.
4.4 Exceptionality
The notion of reference to an occurrence assumed to be particular, and yet left unidentified, is the characterization of the meaning of some emerging from our examination of usage thus far. The question now arises as to how this meaning could underlie the kind of use illustrated in (2) above, which Israel (Reference Israel1999: 177) deems ‘radically different'and most authors refuse to touch at all (see Farkas Reference Farkas2002: fn. 7; Round Reference Round2004: fn. 3; Szabolcsi Reference Szabolcsi2004: fn. 4):
(2) That was some frittata!
This use occurs in certain well-defined environments. The subject that the some-phrase qualifies generally has a specific value, and it is difficult to come up with credible examples with indefinite subjects: *A main course was some frittata! The verb subcategorizes a predicative complement and belongs to the class be, seem, become. The some-phrase is under focal accent – a point to which we will return below – and it can be preceded by the degree determiner quite, a fact that confirms the remarkable character of the referent of the NP.
A relevance-theoretical perspective would argue here that to say of a frittata that it is some frittata is trivially true, and so a more informative interpretation is sought through pragmatic reinterpretation leading to an exceptional-exemplar reading. The problem with this account is that one would expect exceptional-exemplar readings to be producible with the indefinite article as well, for to say of a frittata that it is a frittata is also trivially true; however, one does not find any uses of That was a frittata! to convey the type of message observed in (2) above. The indefinite article, used without focal accent, would convey a non-quantitative sense indicating mere set membership in this type of construction (see Huddleston & Pullum et al. Reference Huddleston and Pullum2002: 372), as in That is a frittata, in case you didn't know. It does this, in our view, through its evocation of the referent of that as a representative exemplar of the category ‘frittata’, a semantic configuration that can be intended by the speaker to simply convey the message that ‘that’ belongs to the category of frittatas. Because of its more complex meaning of ‘particular referent whose identity is expressly left unspecified’, unaccented some, on the other hand, cannot be used to simply ascribe set membership (cf. *Joe is some teacher). Nevertheless, if stressed so as to indicate the exceptional character of the particular referent in question, it does make sense to evoke this referent as being difficult or impossible to specify, because it defies specification due to its exceptionality. This would account for the ability of a noun phrase containing accented some to denote an outstanding exemplar of the category designated by the following noun.
The ability of accented some to imply considerableness or exceptionality is further exploited in some varieties of English in the use of some to modify adjectives illustrated in (31) below:
(31) But oh, that sticky pudding was some good! (www.karlwells.net/mar1211.htm)
Huddleston & Pullum et al. (Reference Huddleston and Pullum2002: 385) characterize this construction as American usage belonging to the very colloquial or non-standard register. Be that as it may, it is obviously an extension of the use illustrated in (2) above: here it is implied that the speaker is unable to specify how good the pudding was, because it was exceptionally good. Focus, predicative position and application of an evaluation to a definite subject are the conditions under which the unidentified dimension of some is reinterpreted as expressing a high-quality reading. These illustrate the relations that a conventionalized reading has with the item itself and the contexts in which it occurs. In the next two sections we take a more systematic look at two factors which interact with the semantic content of some to modulate the effects produced in observable utterances involving this determiner.
5 Modulating factor I: number
One of the main modulating factors interacting with some is the grammatical number of the noun following this determiner. It is significant that Farkas considers only uses of some in which it is followed by a singular count noun (2002: 8 and fn. 7), while Israel's approach seems to work fairly well when applied to plural and mass nouns but breaks down when applied to singulars. This suggests that the impression of non-identification observed by Farkas is connected to the presence of a singular nominal, and that of scalarity perceived by Israel to the occurrence of a plural or mass noun. Based on the hypothesis that the underlying meaning of some is to evoke a particular, but non-identified, referent within the possible range of reference coverable by the noun with which it is used, we argue that the grammatical number of the noun indicates whether what is referred to by the noun phrase containing some is singular or plural. Applied to a singular, the meaning of some indicates that there is a fixed referent to which the utterance applies and the idea of non-identification can only concern the identity of that referent. This explains Farkas' intuition concerning Mary was depressed for some reason that some tells us that there is a definite reason which is not being specified, whence the impression that the speaker is ignorant of the exact nature of the reason, is withholding this information, or just cannot see what could be troubling her.
Applied to a plural referent, the effect is very different. Since a plural noun can denote various numbers of referents falling under the concept expressed by the noun, the plurality of the noun combines with the ‘non-specified’ component of some's meaning in a different way than the notion of singular does – to produce, not the impression of the non-specification of the identity of the referent, but rather that of the non-specification of the number of referents referred to. The effect is analogous with mass nouns, which also involve a possibility of quantitative variation from a small to a large amount of what they denote.
Consequently, it is argued here that the idea of particular, but non-identified, reference denoted by some produces different resulting impressions according to what it is applied to. With singulars, what is understood to be non-specified is exclusively the identity of the referent. With plurals and mass nouns on the other hand, the possibility of quantitative variation in the referent entails that it is the number of referents or the amount of the reference mass that is understood to be unspecified, although of course the identity of the referents is unspecified as well. This quantitative non-specification gives rise to the evocation of an undefined middlish quantity which, if contrasted with the whole set of referents denotable by the noun following some, can be construed as a lower than expected value on a quantitative scale, whence the scalar analysis applied to some with plural and noncount nouns by Israel.
6 Modulating factor II: sentence accent
We will follow Bolinger's terminology here in referring to sentence accent as the accenting of the stressed syllable of one word, making it phonetically more prominent than usual (Reference Bolinger1972: 644; see also Reference Bolinger1986: 14–15). With some, accent can indicate, first of all, that the latter conveys new information (the ‘narrow focus’ of Carlson Reference Carlson2009: 1194):
(32) – How much exercise did you do this week?
– I did some.
Secondly, stress can be contrastive (Rooth Reference Rooth1992), as in (33):
(33) Some men like housework.
This can be compared to an unaccented use of some such as (34):
(34) Some men were standing around outside the door when I arrived.
The fixed reference of some with a plural noun picks a certain subset of men out of the total category, which opens up the possibility of contrasting these men with the rest of the category, this contrast being what accent signals in a use such as (33) above. When no accent is placed on some as in (34), reference is simply made to a certain unspecified number of men. The possibility of this type of contrastive impression exists only with plural and mass nouns (cf. SomeCanadian wine is drinkable); there are no parallel uses with the singular:
(35) *Some man likes housework.
This is simply because an unidentified individual is not felt to represent part of a whole in the way an unidentified number of individuals does. As expected, a partitive impression is also lacking in the non-accented use of some with a singular noun parallel to (34) above:
(36) Some man was standing around outside the door when I arrived.
We have already observed how the intermediate quantity evoked by some with plural and mass nouns can be contrasted with larger quantities, as in (27):
(27) I have some idea of what you mean, but it's still a bit hazy.
Because some + plural/mass noun implies a positive but not necessarily large amount, it can also be contrasted with zero quantity, as already observed with (28):
(28) There must be some mercy in that guy's heart.
The third role of accent observed with some is found in Huddleston & Pullum et al.'s ‘considerable-quantity’ use, where we have argued it to be emphatic, rather than contrastive, having the effect of calling attention to the fact that the unspecified quantity referred to by the some-phrase is a considerable one. Similarly, in the exceptional-exemplar use, accent signals reference to a significant individual whose identity defies specification.
Accent thus plays an important role in bringing out the qualitative readings of some and blocking the ignorance/indifference interpretation. Although it can signal a contrast between the quantity corresponding to some and some larger quantity, which can in certain cases be the whole category denotable by the noun, it can also be used to place emphasis on the significance of the individual (with singular nouns) or quantity (with plural and mass nouns) referred to by the some-phrase, thereby activating the implication that the latter is so significant or considerable that it is difficult or impossible to identify it precisely. This is the very opposite of ignorance or indifference, and implies a qualitative appreciation of the referent of the NP.
7 Related qualitative uses with quite
The value expressed by stressed some in the qualitative readings illustrated by (2) and (3) is comparable to that conveyed by quite some in the following:
(37) That was quite some party!
(38) We discussed the problem at quite some length.
Quite has been described as having two basic types of use: as a ‘maximizer’ to express ‘reinforcement’, as in Joe is quite right, and as a ‘compromiser’ to express ‘approximation’, as in This novel is quite interesting (see Quirk et al. Reference Quirk, Greenbaum, Leech and Svartvik1985: 447; Huddleston & Pullum et al. 2002: 437). For Paradis (1994: 159) compromisers (quite, rather, fairly, pretty) are situated in the middle of the intensity scale, between diminishers such as slightly and boosters such as very. The literature offers no attempt, however, to show any unity behind the two senses of quite just described (see also Diehl Reference Diehl2004; Palacios Martinez Reference Palacios Martinez2009). We would propose a paraphrase for the underlying meaning of this adverb along the lines of ‘corresponding to a sufficiently substantial degree to x that there is no need to seek a higher degree of correspondence’. Negating quite denies the absence of a need to seek a higher degree of correspondence, and therefore asserts the need for a bit more of some quality, implying an as yet insufficient degree and giving the resulting impression of ‘almost’, as in I am not quite ready. With a gradable category such as interestingness, quite presents the referent as possessing this attribute to a sufficiently substantial degree that there is no need to seek a higher degree of the quality in question, whence the paraphrase by means of fairly. However, the gradable nature of the quality implies that higher degrees of it are nevertheless conceivable. With a yes/no, in-or-out type of category such as right in Joe is quite right, on the other hand, since the referent either possesses the quality or does not, the absence of a need to seek a higher degree of correspondence to the category is inferred to be due to the existence of a perfect correspondence with it, whence the paraphrase with entirely. The use of some in (37) above is predicated on the impossibility of pigeon-holing the referent, due to the latter's exceptionality. As for quite, it plays the role of a maximizer, presenting the referent of the NP as being in no need of any higher correspondence to the notion of ‘partyness’ because it is so exceptional. Uses such as (38) seem open to two possible analyses: they could be treated like (37), with quite signalling that the quantity denoted is in no need of any higher correspondence to the notion of lengthiness, because its length is so considerable; on the other hand, they might also be amenable to a construal in which quite is employed to simply denote the possession of unidentified quantity (some) to a sufficiently substantial degree. We are not in a position at present to adjudicate between these two analyses, nor is it our job in this article to propose a complete treatment of quite, so we will leave this minor loose end untied at this point.
8 Some conclusions
Based on the observations made by previous authors, we have proposed that the underspecified meaning of some corresponds to the notion of a real, extensionally limited, but non-identified referent. This general, abstract meaning has been reconstructed through a process of inference, based on the observation of some's uses in discourse and comparison with words which have closely related meanings. When utilized to express particular messages, this meaning has been shown to be modulated by certain discourse factors which bring into play Gricean principles and the inferences to which they give rise. The scalar-attenuator impression perceived by Israel in certain uses of some has been shown to be a product of the application of the notion of non-identification which is part of some's underlying meaning to a quantitatively variable referent. This gives rise to the impression of a reference to a non-specified middlish quantity, and since this quantity is smaller than the whole category, it can be felt to be situated lower on a scale of quantities than a maximum or close-to-maximum amount. The absence of any impression of scalarity with singular head nouns has been argued to be due to the fact that some's semantic component of non-specification applies in this case to the identity and not the quantity of the NP's referent. This accounts for Farkas' impression that in this type of use some is an existential quantifier whose variable is unidentified.
The treatment of some proposed here also opens up interesting perspectives regarding Haspelmath's (1997) cognitive map of indefinite expressions. A first point that could be made is that our study highlights the need to recognize that Haspelmath's ‘functions’, which in the current state of research are a mix of types of context (e.g. indirect negation) and of semantic readings (e.g. specific indefinite), emerge from the tight relation that holds between readings and contexts. Consequently, it is perhaps the case that a tri-dimensional map is needed rather than a bi-dimensional one to cover the third dimension of ‘context + reading’ (see Larrivée, in press). Secondly, our study also shows that the kind of map proposed by Haspelmath is neither sufficiently precise nor sufficiently nuanced to handle the qualitative uses of some, as three of the four uses dealt with in this article cannot be located on his cognitive map, viz.:
– evocation of a particular exceptional exemplar which defies precise identification (That was SOME frittata!)
– evocation of a particular considerable quantity which defies precise quantification (That was SOME time ago)
– evocation of a particular approximate quantity which defies precise quantification (There were some fifty people at the party).
As Dahl (Reference Dahl1999: 665) has pointed out: ‘the map simplifies reality in that it leaves out a number of relevant distinctions’.
The role of sentence accent with some has also been examined in greater depth and brought into clearer focus in this study. It has been seen that placing accent on some can indicate new information, contrast or emphasis. When it is contrastive, accent can signal an opposition between the part of the whole referred to by some+ plural/noncount noun and (i) the rest of the whole (some vs the others), (ii) the whole itself (some vs all), or (iii) nothing (some vs none). When emphatic, according to whether the noun following it is plural/noncount or singular count, accent can signal the considerableness of the quantity or the exceptionality of the singular exemplar. The non-specificatory dimension of some's meaning also comes into play in such cases, as the speaker's inability or refusal to specify the quantity or identity of the referent is implied in these cases to be due to the latter's considerableness or exceptionality.
Finally, the use of some as an approximator has also been shown to be derivable from its underlying meaning. In this case, the notion of non-identification inherent in some's semantics is exploited to signify that the numeral following some must be understood as a non-precise value, which explains why the latter must be construable as a round number. Apparent exceptions to this such as (24) were shown to correspond to another exploitation of some's meaning – one in which some's non-specification of the quantity referred to is implied to be due to a difficulty in specifying it attributable to the quantity's considerableness. What is curious about such uses is the fact that in the very same breath the speaker actually specifies what he implies there is no way to specify. Indeed, the specification of the precise number even provides evidence for the existence of a difficulty in specifying the exact quantity, by showing that the speaker has actually counted the occurrences and that they do add up to a considerable amount which is not easy to count up precisely. Such are the intricate convolutions of the pragmatic exploitation of natural language – but we hope to have shown that it does all make some sense in the end.
