当前位置:首页 >> 文学研究 >>

The concepts of constructional mismatch and type-shifting from the perspective of grammaticalization


The concepts of constructional mismatch and type-shifting from the perspective of grammaticalization
ELIZABETH CLOSS TRAUGOTT*

Abstract The concepts mismatch, type-shifting, a

nd coercion are central to much recent work on cognitive linguistics. In a number of papers, Michaelis has investigated entity and event coercion (Michaelis 2003a, b, 2004 a, b). I address her question ‘‘what conditions favor the diachronic development of shift constructions’’ (Michaelis 2004a: 8) from the perspective of grammaticalization, with particular reference to the development of partitive constructions like a bit of apple into degree modi?er constructions like a bit of a hypocrite. I show why these are di¤erent constructions, and conclude that the most important factors have to do with matching quantitative implicatures to already extant quantifying degree modi?ers with NP heads, and with the strategies for expressing how much ‘‘pragmatic slack’’ (Lasersohn 1999) is available in computing denotations. Keywords: type-shifting; coercion; partitive construction; degree modi?er construction; construction grammar; grammaticalization.

1. Introduction The concepts mismatch, type-shifting and coercion have ?gured prominently in work on semantics (e.g., Partee 1987; Pustejovsky 1995; de Swart 1998) and construction grammar (e.g., Langacker 1987; Francis and Michaelis 2003a; Michaelis 2003a, b, 2004a, b). These are essentially synchronic notions. At the end of her paper on type-shifting and aspectual coercion within a construction grammar approach to tense and aspect, Michaelis poses the question of ‘‘what conditions favor the diachronic development of shift constructions’’ (Michaelis 2004a: 83). In the present paper I bring diachronic data to bear on this question and on the larger issue of semantic interpretation of constructions.
Cognitive Linguistics 18–4 (2007), 523–557 DOI 10.1515/COG.2007.027 0936–5907/07/0018–0523 6 Walter de Gruyter

524

Elizabeth C. Traugott

Michaelis is primarily concerned with aspectual type-shifting, the adjustment of the semantics of an event type to that of an expression with which it is combined. For example, in She liked him in a minute a bounded frame adverbial (in a minute) is combined with an unbounded state (like), and the mismatch in boundedness of resolved by a shift that ‘‘involves the addition of the operator BECOME’’ (2004a: 69). She also mentions that partitives such as a piece of bread involve type-shifting, since a mass entity (bread ) is conceptualized in terms of parts (a piece of ) (2003a: 173). Here I investigate some binominal partitive constructions such as a piece/bit/shred of to test whether historically and even synchronically they are as rigidly subject to type-shifting as Michaelis proposes. Further, I discuss their development into adverbial binominal degree modi?ers, as in a bit of a hypocrite (see partially/quite a hypocrite) and show that coercion construed as a strictly formal device is problematic for a theory of constructions as dynamic, partially-productive, and ‘‘contingent, not deterministic’’ elements (Goldberg 2006: 217).1 I address Michaelis’s question from the perspective of grammaticalization (see e.g., Hopper and Traugott 2003 [1993]), and conclude that the most important factors favoring the developments have to do with matching quantitative implicatures to already extant scalar2 quantifying degree modi?ers with NP heads such as quite and all, and with the strategies for expressing how much ‘‘pragmatic slack’’ (Lasersohn 1999) is available in computing denotations.3 The outline of the paper is as follows. First I provide some background on how I view constructions and their hierarchic relationships (Section 2). This is followed in Section 3 by background on concord, mismatch, typeshifting, and coercion as construed by Michaelis (e.g., 2003a, 2004a). The historical developments are discussed in Section 4, and Section 5 contextualizes the particular changes in the larger picture of language change. Section 6 is a conclusion. 2. Background on construction grammar

Outlines of the several currently available di¤erent approaches to construction grammar, as well as their relationship to Cognitive Grammar, appear in various places (see e.g., Croft and Cruse 2004; Langacker ¨ ¨ 2005; papers in Fried and Ostman 2004; Ostman and Fried 2004; Goldberg 2006) and will not be repeated here. Although what follows does not depend crucially on any one theory of construction grammar,4 I draw signi?cantly on the model of Radical Construction Grammar in Croft (2001) and Croft and Cruse (2004), in part because this model was built with issues of grammaticalization in mind (Croft 2001: 7).

The concepts of constructional mismatch and type-shifting

525

The basic assumptions of importance to the present discussion are as follows. A construction is a symbolic form-meaning pairing. It is multidimensional,5 and is associated with more or less detailed information about morphosyntactic, phonological (including prosodic), semantic, pragmatic, and discourse properties. All aspects of conventionalized language are in focus, not only idiosyncracies of form and meaning pairing that is ‘‘not strictly predictable from the properties of the component parts or from other constructions’’ (Goldberg 1995: 4). While conceptual structure and pragmatics of the inferential kind may be at least partially universal,6 semantics and syntax are conventionalized, and therefore language- and period-speci?c. Individual constructions are independent but related in a hierarchic system with several levels of schematicity (Croft and Cruse 2004: 262–265). The number of levels is not ?xed, and indeed may be considered to be on a continuum, depending on the construction in question and the level of granularity needed for the analysis. In this paper I use the levels in (1) (see Traugott 2006, forthcoming a, b; similarly Fried, forthcoming): (1) a. Macro-constructions: high-level schemas, the highest level relevant for the discussion at hand, e.g., ditransitive construction, partitive construction, degree modi?er construction, Meso-constructions: sets of similarly-behaving constructions, e.g., the set a bit/lot (of ), as distinct from the set (a) kind/sort of, etc., Micro-constructions: individual construction-types, e.g., a lot of vs. a bit of, Constructs: empirically attested tokens of micro-constructions.

b.

c. d.

While constructions can be of any size from morpheme to complex sentence, the constructions to be discussed here are phrasal. Langacker (2005) has suggested three factors that a¤ect linguistic structure to be considered in thinking about a construction (see also Bybee et al. 1994): (2) a. b. c. Generality: the extent to which the constructional schema describing the pattern is schematic rather than speci?c, Productivity: the extent to which a constructional schema is accessible for sanctioning new instances, Compositionality: the extent to which the meaning and form of the whole are predictable from those of its parts in accordance with sanctioning schemas.

526

Elizabeth C. Traugott

For any instance of change we can ask: (i) at which level within a construction the change took or is taking place, for example, semantics, morphosyntax, phonology, (ii) which of the three factors in (2) is relevant to the change; and (iii) at which schematic level in (1) it changes, for example, microconstruction or meso-construction, indeed, whether the macroschema itself changes. The ?rst question is traditional in historical linguistics (though without reference to constructions in the construction grammar sense). The second question is typical of work on grammaticalization, and the third is consistent with any theory of grammar that assumes hierarchic structures, but is formulated in ways speci?c to construction grammar. 3. Concord, mismatch, type-shifting, and coercion

It is usually held that in some constructions the parts are in harmony or concord, and unmarked. But sometimes the parts of a construction are incongruent, and marked:
Form-function mappings may be . . . incongruent with respect to more general patterns of correspondence in language. Thus, we adopt the term mismatch to describe linguistic phenomena that involve a crossing of association lines in any cross-modal mapping. (Francis and Michaelis 2003a: 2; italics original)

This statement assumes an ideal of diagrammatic/structural iconicity (Haiman 1980) between components or ‘‘modalities’’ of a construction, e.g., the semantics and morphosyntax are not in competition, but in harmony. Concord and mismatch have been shown to pertain to concepts fundamental to cognitive construal such as boundedness, causation, etc. (Michaelis 2003a: 204). I illustrate the di¤erences between harmony and mismatch with binominal degree modi?ers. Among degree modi?ers are what Quirk et al. (1985: 567) call ‘‘intensi?ers’’. Intensi?ers scale their heads upward or downward, given an assumed norm, and include both ‘‘ampli?ers/ boosters’’, e.g., very, all, a lot/lots (of ), and ‘‘downtoners’’, e.g., almost, a bit (of ), hardly. Other degree modi?ers are ‘‘maximizers’’ (e.g., completely), which place their head at the very top of the scale (Bolinger 1972: 149 refers to ‘‘absoluteness’’ and ‘‘extremity’’). Some are used as both boosters and downtoners (e.g., quite) or as both intensi?ers and maximizers (e.g., all used as an adverb). The head is usually referred to as ‘‘gradable/non-gradable’’, but the terms ‘‘unbounded/bounded’’ (Paradis 2001; Kennedy and McNally 2005) are preferable because of the potential for harmonic relations not only with adjectives (gradable/

The concepts of constructional mismatch and type-shifting

527

non-gradable), but also with verbs (non-telic/telic), and nouns (mass/ count) (see Paradis 1997; Nevalainen and Rissanen 2002). In (3) the degree modi?er is unbounded and is in harmony or concord with its unbounded head. This is usually considered to be the prototypical degree modi?er collocation (which I construe here as a degree modi?er construction), see e.g., Backlund (1973): ¨ (3) very nice/green/interested booster ? unbounded predicate

In (4) the degree modi?er is a bounded maximizer in harmony with its bounded head: (4) completely unique/?nished/extinguished maximizer ? bounded predicate

But (5) shows mismatch of varying degrees: (5) a. b. very unique/dead/extinguished booster ? bounded predicate completely nice maximizer ? unbounded predicate

In (5a) the semantics of unbounded very clashes with that of its bounded head, though the degree of clash is stronger with extinguished than with unique, which has recently come to be used in relatively unbounded fashion. The clash that occurs in non-?gurative as well as ?gurative language (in which is it often exploited) is resolved by reinterpretation of the head as unbounded. By contrast, (5b) is resolved in a di¤erent fashion: the maximizer completely is reinterpreted as a (high) booster. In both cases unbounded interpretations ‘‘win’’. We may note that, historically, maximizers readily become boosters (or even downtoners), cf. very ‘truly’ > ‘to a high degree’, quite ‘completely free’ > ‘entirely’ > ‘very’ > ‘somewhat’ (Sto¤el 1901; Bolinger 1972; Paradis 2000). The latter kind of change is usually said to follow from hyperbolic use of language (Bolinger 1972). But it may be more appropriate to see such changes as following from their function in discourse. To the extent that maximizers are ‘‘slack regulators’’ that ‘‘signal how much detail should be ignored’’ in a particular expression in a particular context (Lasersohn 1999: 526), they are subject to on-line calculations of the appropriate degree to which denotations hold (or can be expected to hold) truthfully in context, and therefore also to societal changes such as the rise of relativism in the twentieth century that have called into question the uniqueness of truth, or technological changes such as medical breakthroughs that have allowed for subcategorizations of death (brain-dead, heart-dead, etc.).

528 3.1.

Elizabeth C. Traugott Type-shifting

According to Michaelis, mismatch resolution includes type-shifting. A type-shifted construction is one in which a grammatical construction ‘‘denotes a di¤erent kind of entity or event from the lexical expression with which it is combined’’ (Michaelis 2004b: 29). She says the purpose of type-shifting is to perform a conversion ‘‘to alter the conventional designation of the lexical ?ller’’ (Michaelis 2003a: 173). While most of her examples are from event structure, e.g., type-shifting of state into activity as in I’m liking your explanation (Michaelis 2004a: 75, 2004b) she is to some extent concerned with the partitive construction. She says the prototypical partitive construction involves type-shifting and the partitive
is designed to shift the unbounded value of the (necessarily underdetermined7) lexical complement (say pie, as in a piece of pie) to the bounded value associated with the head ( piece) [and] requires that the nominal complement of the PP headed by of denote a mass entity. (Michaelis 2003a: 173, caps added)

Another claim is that type-shifting typically involves a periphrastic expression in which the semi-closed class head (in the present case, a partitive expression like a piece/bit/shred/slice of ) performs the conversion (ibid.), but other functors may do so as well (Michaelis 2004a: 51–52; see also the de?nition of explicit type-shifting below). Several issues are relevant here. For one, conversion is a problematic concept without further de?nition. It involves word-formation, and in the literature on type-shifting it is not always clear whether authors have in mind the kind of on-line conversion that can occur in a particular situation (e.g., the much-cited examples like There’s rat all over the road ), and those that have become conventionalized, leading to competing homonyms. For example, Michaelis (2003a: 175) appears to assume that pudding is a mass noun that can be converted on-line; however, pudding as substance and pudding as entity have coexisted for a long time and online conversion is unlikely; rather, they appear to have been conventionalized as bounded and non-bounded polysemies: (6) a. and in the composition of a pudding, it was her judgment that mixed the in gradients (sic) (1776 Goldsmith, Vicar of Wake?eld [UVa]) no danger of starving here, I tell you! – plenty of bread-fruit – plenty of water – plenty of pudding – ah! plenty of everything! (1846 Melville, Typee [UVa])

b.

Furthermore, ever since distributors packaged portions of pudding in bounded containers, there has been an arguably lexicalized and hom-

The concepts of constructional mismatch and type-shifting

529

onymous contrast between pudding and packaged portions that can be bought singly as a pudding. The question must be which of these lexical items is accessed in any particular situation. A second issue is whether type-shifting is restricted to lexical, open class items. As noted with respect to (5b), it is the grammatical (semi-closed class) adverb that is shifted. Finally, a third issue is whether partitive ‘‘requires’’ the complement to be a mass entity. 3.2. Coercion

‘‘Coercion’’ has been put to a range of uses (see Ziegeler 2007). I focus here on Michaelis’s approach. She interprets it as a semantic/pragmatic phenomenon accounting for mismatch, in keeping with Partee (1987), Pustejovsky (1995), and de Swart (1998), but from a partially di¤erent perspective. Crucial to Michaelis’s view of coercion is the distinction between explicit and implicit type-shifting. These and the ‘‘override principle’’, which is the accommodation mechanism that allows reconcilations, are de?ned as follows:
Explicit type-shifting. A shift in the designation of a lexical item (or its projection) by a grammatical construction with which that lexical expression is conventionally combined. (Michaelis 2004b: 28, italics original). Implicit type-shifting. A shift in the designation of a lexical item (or its projection) in order to reconcile semantic con?ict between word and construction, in accordance with the override principle (ibid.); an inferential procedure which bridges semantic gaps in morphosyntax. (Michaelis 2004a: 47, italics original). The override principle. If a lexical item is semantically incompatible with its morphosyntactic context, the meaning of the lexical item conforms to the meaning of the structure in which it is embedded. (Michaelis 2004b: 25, italics original)

In cases of explicit type-shifting the external semantics of two or more elements of the construction are mismatched. According to Michaelis, in the English partitive construction there is always semantic mismatch between the partitive expression and its complement, since the ‘part/portion’ is a unit, and, as mentioned above, the partitive construction ‘‘requires’’ that the nominal complement denote an unbounded entity (Michaelis 2003a: 173). This claim will be challenged below. Here it is su‰cient to note that in (7) there is said to be a mismatch or lack of harmony between a piece of (unit/bounded) and the open-class complement bread (mass/ unbounded): (7) She took a piece of bread.

530

Elizabeth C. Traugott

Since bread is prototypically mass, the type-shifting (construal of mass in terms of parts) is explicit, or directly derivable from the uni?cation of the parts of the construction. However, if the open-class complement is, like sheet, typically count/bounded, then the type-shifting is said to be implicit, and the noun is coerced into a mass/unbounded reading by the partitive construction (Michaelis 2003a: 174): (8) Give me a shred of sheet.

Here the internal semantics of the open-class complement is a¤ected, a predictable side e¤ect of the inherent meaning of sheet (the internal semantics) and semantic uni?cation requirements. In the next Section historical data for the development of the partitive into the degree modi?er construction will be used to assess the hypotheses that: (9) a. b. c. Partitives require a mass complement. (Michaelis 2003a: 173) Type-shifting involves the imposition of meaning by a closed class head on an open class complement. (Michaelis 2004a: 28) Implicit type-shifting/coercion in the sense discussed above is su‰ciently important to merit special theoretical status. (Michaelis 2004a: 28)

While construction grammar is typically intended as a usage-based model of language, Michaelis does not appear to have used naturally occurring data in developing these hypotheses. What follows is data-driven in the sense that it relies not on intuition but on data derived from corpora available on-line (see Sources for the references). 4. Some aspects of degree modi?er development

In this Section I outline the development of some partitive constructions into degree modi?er constructions. This assumes that the two constructions are distinct. Reasons for the distinction are given immediately below in 4.1. 4.1. The distinction between partitive constructions and degree modi?er constructions

In the construction grammar literature, NP1s in partitives (heads of the construction) are often discussed in terms of units or portions (Michaelis 2003a), or of quanti?ers (Croft 2001; Yuasa and Francis 2003; Francis and Yuasa 2006). Yuasa and Francis cite such examples as: (10) He’s got a lot of friends. (Yuasa and Francis 2003: 194)

The concepts of constructional mismatch and type-shifting

531

among partitives, and construe the alignment of quanti?cational nouns in Present Day English like a lot/bunch of with binominal NP of NP strings as illustrating unusual distributional properties. They argue (following Sadock 1990) that there is mismatch:
Quanti?cational nouns are categorized as nouns in syntax but as modi?ers in semantics . . . it is clear that syntax and semantics are operating according to di¤erent principles. (Yuasa and Francis 2003: 197–198)

Speci?cally, they analyze a lot/bunch of as being ‘‘just like quanti?cational adjectives with respect to semantics’’ but like nominal heads with respect to syntax (Yuasa and Francis 2003: 193). However, there are several criteria for distinguishing degree modi?er constructions with NP heads from partitive constructions (see Denison 2002).8 These include: (11) a. Agreement patterns with the predicate: e.g., the partitive a lot of has singular agreement (a lot of sheep is for sale ? unit to be sold together), the degree modi?er counterpart has plural (a lot of sheep are for sale ? ‘many’, individuals that can be sold separately), Agreement patterns within the construction: in the partitive construction the initial determiner agrees in number with N1 (these kinds of skill ), but in the degree modi?er construction it can agree with N2 (these kind of skills), at least in colloquial use, In the partitive construction NP2 may be preposed (of an apple a bit), but not in the degree modi?er construction (*of a fraud a bit),9 In the degree modi?er construction a N1 of can be replaced by one word or somewhat of (a bit of a hypocrite ? rather/quite/ somewhat of a hypocrite); but not in the partitive construction (a bit/piece of (a) Christmas pudding A *rather/quite/somewhat of a Christmas pudding), Only the degree modi?er construction may have adverbial properties, therefore only the degree modi?er construction may collocate with adjectives (a bit/*piece green) or Verbs (I sort of/*unit of liked it), and only the degree modi?er construction may have adjunct variants (I liked it a lot/sort of/*unit), Some partitive constructions do not have or have only marginal degree modi?er construction polysemies (a piece/unit of ).

b.

c.

d.

e.

f.

In the next two Sections I show how binominal degree modi?er constructions arose out of partitive constructions.

532 4.2.

Elizabeth C. Traugott The partitive construction in earlier English10

The partitive in the sense ‘unit of ’ was expressed in earlier English by a nominal head with a nominal modi?er in the genitive case (e.g., hlot landes ‘lot land-GEN’). The modern NP of NP partitive construction arose only in Middle English, when of ‘out of ’ came to be the possessive and default preposition of the complement (Denison 2002). In Middle and Early Modern English texts there are several examples of partitives with bounded complements. One of the earlier examples has a de?nite (but pronominal) complement (12a); however, several later examples have a complement with an inde?nite determiner (a, any) (e.g., 12b–d): (12) a. With strengthe of his blast / The white [dragon] brent than rede, / That of him nas founden a schrede./ Bot dust. ‘With the strength of his blast, the white dragon burned the red, so that of him (the red) not a shred was found, only dust.’ (c1300 Arthur and Merlin 1540 [MED shrede a)]; note preposed of him) a piece of a card or paper cut like a crosse. (1598 Florio, Ven? tarello [OED windmill ]) And a certaine woman cast a piece of a milstone vpon Abimelechs head, and all to brake his scull. ‘and a certain woman threw a piece of a millstone on Abimelech’s head and completely broke up his skull.’ (1611 Bible, Judg. ix. 53 [OED all ]) It is a shred of an Italian Letany. ‘It is a small piece of an Italian Litany.’ (1628 Jackson, Ecclesiastes [LION: EEBO])

b. c.

d.

Others are attested with plural (13a) and unbounded complements (13a, b): (13) a. God wole haue rekenyng . . . Of men and cloth, the leste shrede. ‘God wants to-have accounting . . . of men and cloth the smallest part’ (? ‘God wants to have an accounting of every smallest part of men and even of cloth’). (c. 1400 Eche man be ware [MED shrede a]; note the preposing) Take a pece of beef or of mutoun, and wyne. ‘Take a piece of beef or mutton, and wine.’ (a1450 Hrl.Cook .Bk.(1)8 [MED pece 4c])

b.

The lexical items shred and piece (the latter borrowed from French), originally meant ‘a unit/part of ’ an entity that was bounded (coin,

The concepts of constructional mismatch and type-shifting

533

animal) or unbounded (cloth). Other partitive constructions developed from non-partitive sources, e.g., a lot of < ‘a share of ’, by metonymy from drawing lots (lottery): (14) He ne wass nohht wurr?enn mann . . . Forr to forrwerrpenn anig lott O¤ Moys?sess lare. ‘He not was not become man . . . for to overthrow any part of Moses’s teaching’ (? ‘He did not become incarnate to overthrow any part of Moses’s teaching’). (c1200 Orm 15186 [MED lot 2c])

A bit of derives from ‘a bite out of ’ (a bit of is the piece bitten out, a morsel that is metonymically the result of biting, as in (15a)): (15) a. this appyl . . . a bete therof thou take. ‘This apple, a bite/bit of it take.’ (c.1475 Ludus C. 23/220 [MED lot 2c]; note the preposing of this appyl; note also that this is a bridging example, in which both ‘bite’ and ‘bit’ are possible interpretations11) He badd tatt gho shollde himm ec / An bite br?dess brinngenn. ‘He commanded that she should him also a bite/bit of bread bring.’ (c.1200 Orm 8640 [MED bite 3a]; also a bridging example)

b.

A lot/lots of occurred primarily with unbounded complements (14), but the source of a bit of could occur with either bounded (15a) or unbounded (15b) complements. The partitive construction, at least in its earlier instantiations, is not yet a closed class. As (13a) and (14) show, the Ns in NP1 were treated as open class items that could occur with either article, a quanti?er, adjective, or other modi?er. Even if we restrict our examples strictly to strings with inde?nite nominals, they do not necessarily ‘‘require’’ type-shifting as they often do not impose boundedness on their complement. A piece of, though considered colloquial or slang by both the OED and the AHD can still occur with a complement marked for inde?nite ((16) is surely not colloquial!): (16) A piece of a chromosome may break o¤, causing a de?ciency or deletion. (1952 C. P. Blacker, Eugenics 245 [OED deletion])

Therefore Michaelis’s (2003a: 173) claim cited in section 3 that partitives are ‘‘designed to shift the unbounded value of the (necessarily underdetermined) lexical complement (say pie, as in a piece of pie) to the bounded value associated with the head ( piece)’’ is not supported by earlier or even contemporary data on partitives. Furthermore, the very fact that members of certain lexical classes (e.g., vegetables, and cooked foods

534

Elizabeth C. Traugott

such as carrot, endive, potato, pie, pudding, roast) are systematically construable as either unbounded or bounded, i.e., are underdetermined for boundedness, calls the necessity of shifting into question at least in these cases. 4.3. The degree modi?er construction in earlier English

Something that is a piece or shred, or that one can bite out (a morsel) is something of a certain size or quantity. In some contexts small size is associated with inadequacy: (17) a. The fragments, scraps, the bits, and greazie reliques, Of her oreeaten faith. ‘The fragments, scraps, the bits, the greasy remainders of her chewed-over faith.’ (1606 Shakespeare, Troilus and Cressida V.ii.159 [OED]) The spirit which animated her father when he went to housekeeping in a piece of a house without any front window. (1884 Harper’s New Monthly Mag. 69 303 [OED piece d; said to be US regional])

b.

Association with ‘a small quantity of/some’ in the case of a bit of, and with ‘a large quantity of ’ in the case of a lot/bunch of, presumably enabled (but, as discussed below, did not require) the semantic reanalysis of the partitive construction as a polysemous degree modi?er construction, i.e., from a unit to a quanti?er reading especially if the NP2 complement was expressed by a mass/unbounded N. In the case of small quantities, association with negative evaluations may have further enabled the reanalysis. The invited inference of quantity from the N1, e.g., bit, lot, which was a member of a lexical set denoting parts, units, and sets to the complement in N2 was, by hypothesis, the factor that enabled the development of the degree modi?er polysemy.12 This change did not occur with similar binominals that were, and still are, available, e.g., locatives like the front of the house, subjective genitives (the singing of the diva), objective genitives (a painting of the river), and evaluatives like that idiot of a politician because they do not invite the relevant inferences for this kind of change. In some of these cases N1 is expressed by a speci?c relatively open-class lexical item (e.g., painting), of the type unlikely to become a grammatical form because it does not have the appropriate generalized semantics or relevant pragmatics. Others like front/back/side of X involve (body)-part meanings of a general, relatively closed-class type that often grammaticalize to simple prepositions and case markers (especially front and back) (see

The concepts of constructional mismatch and type-shifting

535

Svorou 1993; Heine and Kuteva 2002). What distinguishes all of them from the partitives of quantity under discussion here is that they do not readily invite inferences of quantity.13 By contrast, partitives have the potential for semantic reanalysis within them, since a part or unit of something invites the inference of quantity, and therefore the development of degree modi?er polysemy is not surprising. There are several bridging examples in the data, in which either unit or quantity can be understood, including: (18) a. Mrs. Furnish at St. James’s has order’d Lots of Fans, and China, and India Pictures to be set by for her, ‘till she can borrow Mony to pay for ‘em. (1708 Baker, Fine Lady’s Airs III.i. [LION: English Prose Drama]) Sir, there’s a lot of folks below axing for—are you a Manager, Sir? (1818 Peake, Amateurs and Actors II.i. [LION: English Prose Drama])

b.

In (18a) it seems likely that lots is intended as ‘units for sale’ since money is referred to, and in (18b) that a lot of folks is intended as a group, because of the singular verb (although that is not criterial in representations of dialect). But both at a minimum invite an interpretation of large quantity. Presumably this occurred most readily when the NP2 was unbounded (e.g., china) or plural (e.g., fans). We can hypothesize that there was some pragmatic slack concerning the relative size and composition of units, given the wide array of lexical choices (unit, piece, bit, shred, lot, bunch, etc.). Is a bit of bread a slice, a corner o¤ a loaf, some part roughly torn or bitten o¤ ? In most circumstances it does not matter. Whereas certain expressions can constrain permissible slack (e.g., exactly) on de?nable entities (e.g., three o’clock), others (e.g., loosely speaking) can relax it; they signal the ‘‘pragmatic halo’’ or pragmatically related elements of an expression and hence function as a hedge (Lasersohn 1999: 545). In the case of the partitives in question, it appears that speakers found it useful to recruit the invited inferences of quanti?cation to convert the pragmatic slack associated with partitives into scalar semantic content of a ‘‘loose’’ kind, which often functions as a hedge. Bit and shred came to signal small(ish) amount on a scale, lot and bunch larg(ish) amount, as in: (19) a. I pondered this sentence painfully after he was gone. Is he turning out a bit of a hypocrite, and coming into the country to work mischief under a cloak? I mused. (1847 Emily Bronte, ¨ Wuthering Heights, Chap. 10 [LION: The Bronte Sisters and ¨ Mrs. Gaskell])

536

Elizabeth C. Traugott b. I went right on and furnished them a lot of information voluntarily. (1889, Twain, Connecticut Yankee [UVa])

In sum, the hypothesis is that ?rst there were pragmatic quantitative implicatures arising from binominal partitives in the ?ow of speech in certain linguistic contexts that allowed partial pattern match with the quanti?er/degree modi?er construction based on pragmatic implicatures (a piece of something implies an amount or quantity of something). Then, when individual partitives were construed in enough contexts by enough people with quantitative implicatures, a semantic reanalysis took place, micro-construction by micro-construction.14 In sum, the partitive was construed in relevant contexts as a measure expression that imposes on its head a scalar meaning. This means that a new polysemy arose for some partitives (e.g., a lot of, but not a unit of or, for the most part, a piece of ), evidenced by extension to nominal contexts in which it could not have occurred as a partitive (e.g., a lot of folks/hysteria). The semantic reanalysis resulted in a partial mismatch between form and meaning: the syntax of the partitive was in tension with the semantics/pragmatics. This seems to be what Yuasa and Francis (2003) and Francis and Yuasa (2006) consider the current status of these strings when, after discussing some syntactic factors in the use of a lot/bunch of and lots of, they conclude ‘‘their syntactic category and syntactic function within the noun phrase has not changed’’ (Francis and Yuasa 2006). Francis and Yuasa (2006) consider the binominals to have grammaticalized. But for a string or construction to have grammaticalized, there must have been both a form and a meaning change according to most views of grammaticalization from Meillet (1958[1912]) on (see e.g., Lehmann 1995[1982]), Hopper and Traugott 2003[1993] and references therein). And indeed there was syntactic change, resulting in the distribution and meaning di¤erences cited in (11a–d), which are accountable for in terms of a head shift from [NP1 [of NP2]] to [[NP1 of ] NP2] (see e.g., Tabor 1993; Denison 2002 on (a) kind/sort of; Traugott, forthcoming a, b on these and a bit/lot/shred (of )). By hypothesis, over time the morphosyntactic mismatch mentioned above became apparent, since the new polysemy came to be constrained (e.g., fewer choices of determiner became available for lot, bit, sort, etc.), and was understood as less compositional (a and of ceased to be construed as inde?nite article or as preposition, depending on the expression, cf. sorta, a lotta). This led to matching of the binominal micro-construction with the structural properties of the adverbial degree modi?er macro-construction, i.e., reanalysis as an adverbial, licensing syntactic contexts typical of this new function (e.g., adjectives, as in sort of cold, a lot wiser).

The concepts of constructional mismatch and type-shifting

537

This reanalysis as an adverbial enabled most binominals to be incorporated into the macro-degree modi?er construction. The head of this construction is typically an adjective (very cold ) or adverb (quite quickly). In earlier English there are a few instances with a verb as head (cf. all in all to brake his skull ‘and completely broke up his skull’ in ex. (12c) above). Since the mid eighteenth and early ninteenth centuries it has also had nominal heads (a bit of attention). Degree modifer uses of (a) kind/sort/ bit of are attested by the mid eighteenth century, of lots/a lot of by the early nineteenth century, and of a shred of by the middle of the nineteenth century. The main enabling factor for the morphosyntactic change was no doubt the quantitative pragmatics available in certain contexts. Another possible factor may have been the development in Middle English and retention throughout the centuries of the expressions a great deal of and a good deal of (< Old English d?l ‘part’), as in (21a), and of the adverbs quite and all in Early Modern English (for fuller discussion of possible models, see Traugott 2006): (20) a. ?ey gaderd a gret del of money. ‘They gathered a great deal of money.’ (a1456 (a1402) Trevisa Nicod. [MED del n2, 3b]) so by this means the poore man lyeth in prison till he be quite a beggar. (1596 Hutton, Blacke Dogge [LION: EEBO]) I will not be myself, nor have cognition / Of what I feel. I am all patience.15 (1609 Shakespeare, Troilus and Cressida V.ii.64 [LION: Oxford Shakespeare])

b. c.

A deal of (which originally meant ‘a part of ’ as in Ic gife ?a twa d?l of Witlesmere ‘I bequeath the two parts of Witlesmere’ (a1121 Peterb. Chron.)) illustrates the fact that invited inferences of quantity or inadequacy did not require reanalysis, or least not one that was long-lived. A deal of had some degree modi?er uses in the ?fteenth to seventeenth centuries, but ceased to be used in either partitive or degree modi?er senses except in ?xed phrases like a great deal of. A piece of appears with unbounded complements (e.g., bread ), and with inadequacy meanings as in (17b) above, and even, for a short period of time from the late sixteenth through the eighteenth centuries, as a degree modi?er in the limited context of scaling downward, indeed denigrating, a person’s abilities or role, as in (21), where piece of a X means ‘somewhat of a X/in part a X’: (21) a. Her ?rst Trade was a cloak to all the rest, vnder color wherof, being withall a piece of a Sempstresse, many young wenches . . . came to her house to worke. (Mabbe, Spanish Bawd I. [LION: English Prose Drama])

538

Elizabeth C. Traugott b. was assailed by that Splenetick Passion, which a Country good fellow that had been a piece of a Grammarian meant, when he said he was sick of the Flatus. (1661 Wotton, Lett. to Edmund Bacon 101, I [OED piece 6c]) but his uncle, Alderman Surcingle the sadler, a piece of a puritan, would not give his consent. (1778 Foote, Maid of Bath I.i. [LION: English Prose Drama])

c.

But a piece of survived as a degree modi?er in only marginal uses (usually sexual), and did not become an adverbial with uses like a bit.16 Each individual micro-construction had a rather di¤erent history. A shred of is now preferred in negative polarity contexts, and with positively oriented heads, cf. (not) a shred of honor/beauty/?treachery/?ugliness. More importantly for our purposes, a shred of came both as a partitive and as a degree modi?er to be used primarily with an unbounded/mass NP2 (e.g., apple (used as a mass N), honesty, evidence). Nevertheless, there are also some uses with bounded heads and in positive contexts. In (22a) shred is treated as equivalent to piece (though presumably smaller), i.e., partitive, with a bounded N, and indeed a piece of a missile is a possible paraphrase. In (22b) we ?nd a degree modi?er use (? ‘some’) with a bounded head; in this case, a piece of a chance is not a possible paraphrase. Both involve NP2 with an inde?nite article, which suggests a bounded construal. If there is any mismatch, it is between the data and the hypothesized ‘‘requirement’’ that partitive a shred of has an NP2 that is externally (sometimes internally) type-shifted (e.g., missile rather than a missile), and (by hypothetical extension from Michaelis’ requirement regarding allegedly prototypical partitives) that the prototypical degree modi?er has a required unbounded NP2. (22) a. it is easy to believe that a missile brought down the airplane. But not a single piece of wreckage shows signs of inward ignition of the fuel tank where the explosion originated. Further, months of sweeping the ocean ?oor turned up not a shred of a missile. (2000 Negroni, Crash site secrets, http:/ /www.pbs.org/ wnet/innovation/experts_qa5.html) I think it has at least a shred of a chance of being accepted. (1993 Walker, PATO: Collective security http:/ /www.fourmilab .ch/auto?le/www/chapter2_105.html)

b.

Other micro-constructions are less restricted than a shred of, and can occur relatively freely with bounded and non-bounded heads in either the partitive or the degree modi?er construction. However, there is in Modern English often a division of labor between partitive with

The concepts of constructional mismatch and type-shifting

539

unbounded complements (type-shifted or not) and degree modi?ers with bounded heads. Minimal pairs can be instructive: a sort of mouse is a type of mouse (a unit in a taxonomy of mice) (partitive, type-shifted) but sort of a mouse is not necessarily a mouse (it approximates one, e.g., it is a computer mouse) (degree modi?er, not type-shifted).17 How subtle such di¤erences may be is emphasized by the fact that during the Modern English period collocations with articles have changed over time, as (23) suggests: (23) He had been a bit ailing like the day before, but naught to signify . . . He began again with a bit of a heaviness in his head . . . and he went to sleep and niver wakened. (1846 Charlotte Bronte, Jane ¨ Eyre, Chap. 48 [LION: The Brontes and Mrs. Gaskell]; note also ¨ the adverbial degree modi?er with an adjective head a bit ailing)

In Present Day English we would expect a bit of heaviness (note this is a degree modi?er (a bit of ? some)). Examples such as those in (22) show that any strong type-shifting hypothesis regarding partitives is overstated. At best there is a tendency of NP2s in partitives to be type-shifted in Present Day English. 4.4. Binominal degree modi?ers as models

Once they had become established, binominal degree modi?ers that arose from partitives in turn served as models along with the adverbs quite and all for other degree modi?ers to appear with NP heads, even though they did not have partitives as their source. Some, like rather, occur with a bounded or unbounded head (24a, b), others like very, pretty, totally select a personal name (24c), or a type-shifted nominal (24d–f; cowboy, prom, and top are usually bounded nominals). (24) a. As you . . . mentioned that Mr Legh had rather an Inclination to purchase S[****}old’s Estate near wargery Moss. (1774 Kerfoot, Letter to Orford [18thCProse, Part 2]) Indeed, she laughed, and said it was rather fun, ‘‘like something out of Sterne.’’ (1896 La Gallienne, Quest of the Golden Girl [UVa])18 This song is very Britney spears (sic) . . . I don’t like it. (Google, forums.ukmix.net/forums/viewtopic.php?p=348701) ‘‘You have to be there to see it.’’ Mr. Carte said of the hurricane work, ‘‘As Mr. Aduddell says, ‘‘It’s pretty cowboy.’’ ’’ (9/ 29/05 New York Times) [About a fashion collection] There were school uniforms and a ?nal sequence that was very prom. (1/12/06 The Independent)

b.

c. d.

e.

540

Elizabeth C. Traugott f. The black scoop-neck sweater looks totally top over a balcony bra. (2001 C. Glazebrook, Madolescents 25 [OED top, draft entry as noun])

In other words, the degree modi?er construction with NP heads appears to be becoming type-productive, with the very schematic structure Degree modi?er/Quanti?er X. However, the construction is not yet conventional and indeed the analysis of it is under discussion. Denison (2001), for example, suggests that the nouns fun, key as in very fun/key, are being used adjectivally. Support for this analysis is provided by such expressions as I had a fun evening, and one of Ashley’s most key appointments. Rather than mismatch resolution that involves ‘‘conforming’’ to a construction, it appears that what we have here is mutual adjustment or ‘‘accommodation’’ (to use Langacker’s 1987: 76 term) in which the increasingly schematic degree modi?er construction intersects with other changes such as increased use of nouns as modi?ers of other nouns if they can be construed as ‘‘resembling/of the type’’, cf. designer jeans, cowboy boots. 5. General characteristics exempli?ed by the changes

In this Section I brie?y discuss some of the mechanisms involved in the changes outlined above (Section 5.1) and some possible motivations for them (Section 5.2). Mechanisms concern the ‘‘how’’ and the structural properties of change, while motivations concern the ‘‘why’’ and reasons for change. Both are language-independent, in that they involve cognitive processes, and, in the case of motivations, also social ones. 5.1. Some mechanisms of the change from binominal partitives into degree modi?ers

Once reanalyzed as members of the degree modi?er construction, the original partitives in question underwent the following changes (they still have partitive polysemies): (25) a. They came to have their own characteristics as microconstructions, e.g., a shred of came to coocur mainly with unbounded nouns, a bit/lot of with both bounded and unbounded nouns. Some came to be uni?ed with meso-degree modi?er constructions, e.g., a bit with very, kind of, sort of with respect to adjective collocations (this is selective analogy since we ?nd a bit hungry, but not *a lot hungry, only a lot hungrier).

b.

The concepts of constructional mismatch and type-shifting c.

541

d.

Together with quite and all with nominal heads, they became models to which other degree modi?ers can be analogized (very fun). The macro-degree modi?er construction was therefore expanded (this is reanalysis of the macro-construction, allowing analogical developments at lower levels).

On this analysis, the changes started with partitive constructs in which the quanti?cational and evaluative implicatures of the construction were particularly salient, as when N2 was plural or unbounded, as in (18). One or two such innovative examples would not have led to a change, but repetition of similar uses across di¤erent speakers would have done so.19 The hypothesis I have put forward is that there were three steps of development: (26) Step I. Pragmatic quantitative implicatures arising from partitives in the ?ow of speech in certain linguistic contexts allowed partial semantic match with quantitative degree modi?ers. When individual partitives were construed with quantitative implicatures by enough people in enough contexts, a semantic reanalysis took place. The resulting morphosyntactic mismatch led to matching with the structural properties of the degree modi?er construction, leading to head shift and in many cases reanalysis as an adverbial.

Step II.

Step III.

In Langacker’s (2005) terms, for the degree modi?er construction there was: (27) a. Increase in generality: at the macro-level, the distributional restriction in Old and Middle English to coocurrence with adjective, adverb, or verb heads no longer applies, Increase in productivity: the meso-level constructions with degree modi?er ? NP syntax have since the eighteenth century become more accessible for sanctioning new micro-construction types, Decrease in compositionality: the extent to which the meaning and form of the whole are predictable from those of its parts have been reduced as constructs have been conventionalized as micro-constructions.

b.

c.

An anonymous reviewer queried whether, formulated as it is, (27) is meant to imply that the greatest degree of compositionality is at the

542

Elizabeth C. Traugott

micro-level. In so far as the meso- and macro-levels are more schematic than the micro-level, they are more general and abstract. This means that to the extent that they are compositional, they are so in a highly schematic way; they are not more or less compositional than the microconstructions. At the micro-level the reduction in compositionality involves reduction of the concrete semantics and across the constituent parts. One question is whether the changes as presented here involve grammaticalization. If grammaticalization is construed as limited to the development of closed classes, it would appear that this is not the case, since the degree modi?er construction is clearly highly receptive to new members. Indeed, Borst (1902: 157) notes 48 degree modi?ers in Old English versus 248 in Modern English, and Peters (1994) comments on the considerable productivity in the Early Modern English period of new degree modi?ers derived from adverbs referring to emotions, e.g., horribly ‘very’ (cf. horribly nice). However, closed class membership is not criterial for grammaticalization, as evidenced by the expansion and relative openness of categories such as auxiliary verbs (Krug 2000; see also Trousdale’s forthcoming analysis of the demise of impersonal constructions in English from the perspective of the emergence of the transitive construction). More important for grammaticalization than closed class membership is the functional status of degree modi?ers, and the general nature of the changes in question. According to these criteria, the changes are clearly instances of grammaticalization. Croft (2001: 127) has suggested steps similar to those outlined in (26) for grammaticalization in general. Furthermore, the examples discussed illustrate the three types of generalization (and hence increased type-productivity), identi?ed by Himmelmann (2004) as criterial for grammaticalization: (28) a. Host-class expansion of NP collocations: NP2s cease to be restricted to concrete lexemes (e.g., expansion from apple, dragon to hypocrite, prejudice), Syntactic expansion: NP1 is syntactically reanalyzed as a modi?er (i.e., the pattern now serves two syntactic purposes); in addition, (i) sort of has been further expanded to pre-adjective and pre-verb contexts, (ii) sort of and a bit, a lot, lots have been expanded to free adjunct status (largely in spoken language or representations of it), Semantic-pragmatic expansion: NP1 is semantically reanalyzed as a quanti?er (i.e., the pattern now has two meanings), and some quanti?ers have shifted to status as approximator (a) sort of, booster a lot (of ), or downtoner a bit (of ).

b.

c.

The concepts of constructional mismatch and type-shifting

543

There were also several changes reducing semantic and syntactic compositionality, including: (29) a. Change from constructions in which NP1 is concrete to constructions in which NP1 is understood in a non-literal, nonreferential, and more abstract way: a bit of as degree modi?er in a bit of a hypocrite does not refer to a small piece of a hypocrite (contrast a bit of bread ), but to the speaker’s assessment of partial similarity to hypocritical behavior. Loss of literal meaning is usually associated with ‘‘bleaching’’, but, as in many instances of grammaticalization, there is signi?cant pragmatic enrichment resulting in semanticization of the salient invited inferences involved. In this case, degree modi?ers have conventionalized quanti?cational measure meanings that their sources did not, they impose a scale on their heads in ways that their sources did not, and they are often evaluative in ways that their sources are not; they have therefore undergone subjecti?cation in the sense of being recruited to speakers’ beliefs and attitudes (see Traugott and Dasher 2002), Change from a construction in which both NPs are free to occur with any determiner and in any number, to a construction with signi?cant constraints in this regard, Construction-speci?c phonological reduction and ultimately loss of a (as in sort of ) or of (as in a bit).

b.

c.

Crucially for grammaticalization, which involves structural as well as meaning change, there was reanalysis: (30) a. b. Rebracketing (reversal of head relationships), Functional shift in which NP1 comes to serve a grammatical modi?er function such as ‘quite’, ‘somewhat’.

Other phenomena typical of grammaticalization include the renewal of already extant categories. Degree modi?ers and free degree adjuncts used as responses have existed from earliest Old English (cf. swi?e ‘very’, and gea ‘yes’). The constructions discussed here added to this inventory. Furthermore, the developments exemplify cross-linguistic replication. In so far as a shred of and other expressions referring to small amounts (a drop of, a bit of ) share the well-known tendency to be associated with negative polarity (Israel 2004), they are reminiscent of similar changes in French, where pas ‘step’, point ‘dot, point’, mie ‘crumb’, gote ‘drop’, etc. came to be understood as reinforcers of the negative (in the case of pas, eventually giving rise to a new negative marker).

544

Elizabeth C. Traugott

The hypothesis is that the grammaticalization occurred initially at the micro-level, individual micro-construction by individual microconstruction; as the new micro-constructions came to be aligned with each other new meso-constructions arose, in turn a¤ecting the constituency of the macro-construction, and that all these changes share characteristics associated with grammaticalization, especially increased entrenchment or routinization (Bybee 2003). For fuller discussion see Traugott (2006, forthcoming b) and Trousdale (forthcoming). 5.2. Some possible motivations for the change from binominal partitives into degree modi?ers

In the search for motivations, or reasons, for change, various factors are often mentioned. The nature of the motivations adduced may depend on the perspective of the researcher. From more formal, generative perspectives, what changes is grammars (Kiparsky 1968; Lightfoot 1979); in so far as motivations are considered, they are structural, internal, or frequently limited to hypotheses about child language acquisition. From more functional perspectives, what changes is use (Croft 2001). In this research paradigm, adults as well as children are acknowledged to be involved in bringing about change, and motivations, both internal and external (e.g., communication practices, or social factors) are considered. There is at least one area of agreement among some proponents of these rather di¤erent perspectives, which is that there is a tendency toward alignment and resolution of mismatch. For example, Faarlund has suggested with respect to the spread of subject properties to expletive subjects such as it and there, that ‘‘the resolution of mismatch is a plausible motivation for a historical change’’ (Faarlund 2004: 134, italics original). Kiparsky (Forthcoming) suggests that grammaticalization results from grammar-internal realignments resulting from UG, all of which lead to unidirectionality. However, such resolution is far from necessary and often does not occur.20 The developments discussed here do not suggest that resolution of mismatch was in fact a signi?cant factor for the macro-constructions in question. If it had been, we would have expected less idiosyncracy. We need to consider not only resolution of mismatch but also matching, and the extent to which it occurs. For example, in one of the pioneer ? works on grammaticalization, Givon (1979) suggested that iconicity, the matching of form and function (cf. Haiman 1980’s diagrammatic iconicity) is a major factor in change, presumably a motivation. Iconicity involves various types of analogical matching. I suggested above that there may have been analogical models for the development of the degree

The concepts of constructional mismatch and type-shifting

545

modi?er construction such as deal of, quite, and all.21 In considering any kind of matching, we must be very clear about the level at which the match is being investigated, and about the extent to which it occurs. My hypothesis is that, with respect to the development of binominal degree modi?ers, the match operated only at a very abstract level, e.g., the quanti?cational inferences derivable from the partitive relation scaled the complement, hence speakers could partially match meaning between partitives and degree modi?ers, and this licensed semantic change. Some degree modi?ers (e.g., quite, all ) were adverbials with nominal heads, so a syntactic similarity was also observable, licensing syntactic change (head shift) as well. But the matches were minimal at the more granular level of the NP, especially whether it was unbounded or not, singular or not. As in the case of De Smet and Cuyckens’s discussion of changes in the distribution of complementation patterns in English, choice of the collocations within a partitive or degree modi?er construction ‘‘is not always determined by principles that operate at an abstract level . . . Instead, all sorts of idiosyncracies and irregularities can be seen to creep into the system’’ (De Smet and Cuyckens 2007: 206). Over the last three decades accounts of motivations for change have often made reference to the concept of ‘‘competing motivations’’: the idea that there is a tension between ‘‘be informative/clear’’ (addresseeoriented), and ‘‘be quick/easy’’ (speaker-oriented) (see e.g., Du Bois 1985; Keller 1994; Radden and Panther 2004). In so far as these motivations have been seriously considered in historical work, ‘‘be clear’’ is usually associated with shifts toward more ‘‘expressive’’ language using redundant or more elaborate structure, while ‘‘be quick and easy’’ is associated with phonological reduction (see Hopper and Traugott 2003 [1993] for some discussion). Michaelis very brie?y suggests that, historically, explicit type-shifting might be motivated by the drive toward transparency (presumably a kind of informativity) while implicit type-shifting or coercion might be motivated by ease:
While the use of an explicit type-shifting devices [sic] can be viewed as a hearerbased accommodation, arising from the drive toward maximal transparency, the use of an implicit type-shifting device can be seen as a speaker-based optimization strategy, involving economy of e¤ort. (Michaelis 2004a: 83)

Assuming for the moment that Michaelis’s analysis is correct and partitive ‘‘requires’’ an unbounded complement that is shifted in the construction to bounded interpretation (by piece/shred of, etc.), then a piece/shred of bread matches the constraints of the construction, and is presumably equally easy for both speaker and addressee. A piece/shred of sheet is,

546

Elizabeth C. Traugott

however, more ‘‘costly’’ as sheet has to be assigned unbounded syntax (no a) (Michaelis assumes conversion, which presumably occurs in morphology), and the syntactically anomalous noun is then semantically shifted back to bounded interpretation in the partitive construction. Absent psychological testing, it is hard to see why this hypothetically costly process would be a case of a speaker-oriented optimization or economizing strategy. It would appear to be costly for both speaker and addressee. If anything, the addressee might be considered to have less work to do than the speaker, since the anomalous sheet is already available and does not have to be formulated. However, it is true that bread in a partitive construction is arguably more transparent semantically than sheet. In any event, since, as we have seen, piece and shred can have bounded complements, as in a piece of a chromosome (16) and a shred of a missile (22a), speakers presumably have a choice between using a normally bounded N with an article or not, and by hypothesis it is easier for a speaker to select the conventional syntax for a lexical item than to violate it only to later undo the violation semantically. A more plausible hypothesis is that the changes were motivated not by transparency vs. ease but by invited inferencing and the possibilities of structural realignment with an already extant degree modi?er construction, given general cognitive construals of entities (and events) as either bounded or non-bounded. This assumes, however, that (most) nominals are lexically underspeci?ed for boundedness, and do not undergo conversion. One would, of course, like to know why this motivation, which was presumably always ambient, and indeed seems to have triggered some early changes such as are exempli?ed by a deal of and a piece of (neither of which became entrenched in the language, the second less so than the ?rst) came to be both so productive and so long-lived only in the eighteenth century. I have suggested that the shift from partitives to degree modi?ers gives expression to strategies for signaling how much pragmatic slack (Lasersohn 1999) is allowable or intended in a particular expression. A hypothesis that might be explored is whether the upsurge of the degree modi?ers in question, and the large numbers of others that developed in the sixteenth to eighteenth centuries, had anything to do with the rise of empiricism during the Enlightenment that Wierzbicka (2006) suggests led to a major shift in ‘‘habits of mind’’ and hence of language use. She suggests that a type of discourse came to be favored that cast doubt on beliefs and opinions, and therefore led to use of relativizing epistemics such as possibly over absolutist in truth, of hedging epistemic parentheticals like I think over con?dent statements like I think that X. Bromhead (2006) tests and supports Wierzbicka’s hypothesis by investigating uses of

The concepts of constructional mismatch and type-shifting

547

epistemic adverbials such as verily, forsooth, surely, in truth, and expressions like I think, I suppose, I wot, I ween in sixteenth and seventeenth century texts and comparing them with modern uses. It is possible that such a change in ‘‘habits of mind’’ might also have favored non-maximizing degree modi?ers like a bit/lot/shred of that function to scale their heads up or down and to qualify them in some way. 6. Conclusions and directions for further work The data discussed in this paper suggest several conclusions: (a) The patterns associated with both the partitive and the degree modi?er constructions at the highest, macro-construction level are highly schematic. Complements of partitive NP1 may be bounded or not. Therefore type-shifting is not ‘‘required’’. However, some individual micro-constructions appear to be more strongly than others associated in Present Day English with certain complements. Most notably, a shred of favors unbounded complements in either use, a lot of favors unbounded or plural bounded heads, a bit of favors an unbounded head as partitive, but often cooccurs with a singular indefinite bounded head as degree modi?er. No clear examples are attested in my data of coercion as illustrated by Michaelis’s presumably constructed Give me a shred of sheet. This calls into question whether coercion of this type actually occurs, or if it does, whether it occurs frequently enough to make the distinction between explicit and implicit type-shifting theoretically important. Ziegeler queries whether ‘‘coercion’’ is actually needed separately from invited inferencing, and suggests that:

(b)

Coercion is thus a redundant neologism where constructions are concerned, as the processes required to enhance productivity are no di¤erent from cognitivepragmatic processes already well-known to most accounts. (Ziegeler 2007: 1015)

(c)

(d)

Further, this study calls into question the generality of type-shifting itself, at least for the constructions in question. To what extent some expressions are underspeci?ed or not is in part a function of the theory, but a distinction would seem to need to be made between those lexical types that are by convention readily construed as either bounded or unbounded (e.g., apple, cabbage) and those that are not, e.g., box, sheet (bounded), milk, honesty (unbounded). As has been known for a long time, old patterns are reused for related but distinct functions. In the case studied here partitive was reused as degree modi?er. What has been less clear is that the result

548

Elizabeth C. Traugott of such reuse is not necessarily more ‘‘optimal’’ in the sense of being construction-internally more stream-lined. Indeed, as we have seen, the new uses are quite idiosyncratic. What alignments there have been seem to be morphological rather than semantic or pragmatic, cf. the loss of a in the case of sort/kind of (> sorta, kinda) but of of in the case of a bit/lot of (> a bit, a lot) when they became adverbials. But one degree modi?er, a shred of, did not fully grammaticalize to an adverbial (perhaps because it is the most recent in its degree modi?er use?). What has changed is association of particular expressions with: (i) semantic boundedness or non-boundedness, (ii) speci?c types of lexical heads, (iii) fully adverbial use (lots/a bit/lot), (iv) negative polarity (a shred of, a bit), (v) the degree to which certain collocations are entrenched as routinized expressions or ‘‘prefabs’’ (e.g., (not a) shred of evidence, a great deal of ) (see Bybee 2006 drawing on Erman and Warren 2000). To the extent that motivations for the change to degree modi?er can be discerned, it appears that they do not have to do with transparency versus ease. More plausibly they have to do with matching quantitative implicatures to already extant scalar quantifying degree modi?ers (some of which, such as quite and all, already had NP heads) and with the strategies for expressing how much pragmatic slack is available.

(e)

In order to fully come to grips with how constructions like the partitive and degree modi?er constructions have changed and continue to change over time, and hence what the motivations for change might have been, more detailed study of developing collocations is needed. A promising tool for accessing such information is the recently formulated ‘‘collostructional measurement’’: measurement of the strength of association between a schematic slot in a construction and the lexical ?llers of that slot (see e.g., Gries and Stefanowitsch 2004 for synchronic analysis; Ho¤mann 2004; Hilpert 2007 for diachronic analysis). In pursuit of distributional questions, one could ask of selected electronic data bases questions such as: (a) Whether and how often a piece of or a shred of, etc. in partitive or degree modi?er meaning occur with a normally bounded head like sheet, and whether the unbounded head has an inde?nite article or not, How often they appear with a normally unbounded head like bread, What the lexical class of the head is (e.g., abstract N, e.g., honesty, falsehood, or agent imputed to have a certain kind of stereotyped behavior such as liar),

(b) (c)

The concepts of constructional mismatch and type-shifting (d)

549

(e) (f )

Whether there are signi?cant constructional distinctions to be made between expressions where NP2 is a full NP with determiner, or a bare N (see ft. 1), How the changes interact with changes in the determiner system, When and at what rate these collocations have changed over time.

In this research it would be useful to distinguish cases of lexical types that systematically show bounded/unbounded alternations (e.g., terms for fruits, vegetables, and products of cooking) from those that do not. From a diachronic perspective it is a not unreasonable hypothesis that initially all innovations involve mismatch, in other words, some incongruity of correspondence patterns, whether between an old and a novel use by a speaker, or between speaker intention and addressee’s understanding (Kuteva 2001 points to the importance of non-shared as well as shared discourse knowledge in change, especially grammaticalization). In this sense the term ‘‘mismatch’’ is consistent with widely-held views of acquisition (by child or adult) as involving error, mis-learning or some kind of de?cit. However, if we can alter our perspective on change and think of it not as mis-learning, or mis-use, but as the manifestation of our creative ability to symbolize, whether as children or adults, then we can see that far from being a de?cit, change can be an advantage:
the formation of grammar (and other cultural systems) demonstrates an overperformance of human minds, a capacity for forming new symbols for immediate use that surpasses any need to acquire precisely all the details of extant patterns of usage. (Andersen 2006: 81–82)

If speakers adopt an innovative mismatch, by conventionalizing it, they are likely to creatively reanalyze it as a partial match that adds to the repertoire of the language. Received 15 January 2007 Revision received 20 May 2007 Notes
* Earlier versions of this paper were given at Rice University and at the conference Construction Grammar 3, Dusseldorf, March 31st–April 2nd, 2006. I have bene?ted ¨ greatly from comments and discussion, especially with Walter Bisang, Gabriele Diewald, Mirjam Fried, Martin Hilpert, and Graeme Trousdale. Thanks also to the English Grammar reading group at Edinburgh University for suggestions, to Laura Michaelis, Mika Poss, and Ivan Sag for discussing coercion with me, and also to three anonymous reviewers and Ewa Dabrowska for helpful and thought-provoking comments. Contact address: Department of Linguistics, Stanford University, Stanford, CA 94305-2150, USA. Author’s email address: 3traugott@stanford.edu4.

Stanford University, USA

550

Elizabeth C. Traugott

1. An anonymous reviewer pointed out that two separate constructions identi?ed in Selkirk (1976) are being merged in this discussion: a ‘‘true’’ partitive with NP1 of NP2 as in a lot of the leftover turkey, and a ‘‘pseudo-partitive’’ with NP1 of N 0 as in a lot of leftover turkey. Selkirk’s distinctions depend on various extraposition patterns, and are developed within a theory of X-bar syntax that does not allow for the kinds of intermediate categories that appear in textual data and in the development of the determiner in English (see Denison 2006). A detailed investigation of the intersection of changes in the English determiner system and the partitive-degree modi?er developments under discussion here remains to be undertaken (see Section 6). Here and elsewhere I follow Quirk et al. (1985: 249–251) and use partitive as a cover term for ‘‘partition in respect to quantity’’ such as is expressed by a bit/lot/shred of and for ‘‘partition in respect of quality’’ such as expressed by (a) kind/sort of. Croft and Cruse (2004: 150–163), distinguish these as meronymic and taxonomic partitives, but also point out that the distinction is not always easy to maintain. My focus here is on the meronymic expressions. 2. ‘‘Scalar’’ is here used here in the sense of non-logical pragmatic scales that up- or down-tone a gradable head (see Bolinger 1972; Kay 1990). 3. Thanks to Ivan Sag for suggesting the relevance of Lasersohn’s paper to the topic at hand. 4. In using lower case for ‘‘construction grammar’’ rather than initial capitals, I also do not necessarily intend any particular alignment to one theory or another (see Croft and Cruse’s 2004: 257 suggestion regarding use of capitals for work in the tradition of Kay and Fillmore, lower case for work in the tradition of Goldberg, Langacker, etc.). 5. The terms ‘‘multi-modular’’ and ‘‘multi-modal’’ are also used (see Francis and Michaelis 2003b). 6. Even if, by hypothesis, implicatures such as are exempli?ed by Horn-scales (3all, some4), in which all entails some, but not vice versa, are universal, the weighting of pragmatic maxims that regulate their use, such as ‘Say no more than you must’, appear to be language- and period-speci?c (see Wierzbicka 2006 on changing cultural models). 7. What ‘‘undetermined’’ means here is unclear—one kind of indeterminacy might be non-speci?cation with respect to boundedness, due to absence of a determiner, but that does not appear to be what Michaelis has in mind. 8. Denison (2002) and Huddleston and Pullum (2002) call degree modi?ers with the ?rst three properties predeterminers; Aarts (2001) calls them complex speci?ers. 9. Because they can be preposed, Aarts (2001: 264–66) argues that the NP1s of partitives are like collective nouns and are actually not heads, a position that Yuasa and Francis (2001: 193) reject. 10. The periods of English are somewhat arbitrary divisions. The following are useful guideposts: Old English 650–1150, Middle English 1150–1500, Early Modern English 1500–1750, Modern English 1750–1970. Present Day English refers to the last thirty?ve years (1970 on). 11. En?eld (2005: 318) de?nes a bridging context as ‘‘a speech context in which something inferrable as utterance-meaning from an input sentence-meaning happens also to be true, and thus not defeasible in that context’’. 12. For a detailed account of the Invited Inferencing Theory of Semantic Change, see Traugott and Dasher (2002). The basic idea is that speakers and addressees negotiate meaning, not necessarily consciously; the term is designed ‘‘to elide the complexities of communication in which the speaker/writer evokes implicatures and invites the addressee/reader to infer them’’ (Traugott and Dasher 2002: 5).

The concepts of constructional mismatch and type-shifting

551

13. Other binominal degree modi?ers like sort/kind of derive from terms for ‘type’; again, the source nominal has a rather general semantics, this time feeding into the approximating semantics of these degree modi?ers. 14. An anonymous reviewer was concerned that this statement (and subsequent ones about reanalysis) suggests an abrupt, non-gradual change. An assumption of much work in grammaticalization is that there are no abrupt, catastrophic changes of a whole class at the same time. Instead there are individual reanalyses; they are instantaneous, the steps involved are small, the changes di¤use context by context, and entail the coexistence of older and newer functions of expressions (e.g., Lichtenberk 1991; Haspelmath 1998—however, Haspelmath rejects the term ‘reanalysis’ for these small steps); there may be periods of indeterminacy prior to reanalysis. The same assumption holds here, but the hypothesis is that di¤usion is across micro-constructions rather than lexical and grammatical expressions. 15. It is possible that (20c) exempli?es quanti?er-?oated all, but it seems more likely to be the precursor of eighteenth century expressions concerning manners, e.g., I am all obedience ? ‘I am fully/in every way obedient’), see Buchstaller and Traugott (2006). 16. There is an adverbial use as in go on a piece ‘some distance’, but this concerns extent of distance/time and is not a general degree modi?er use. 17. A similar division of labor is found with cognates, one member of which is inanimate and abstract, the other animate, often denoting a behavioral characteristic, cf. fraud/a fraud, or gossip/a gossip, as in: i) the housemaid, who had done her work for the present, and come into the kitchen for a bit of gossip. (1848 Mary Barton, Tale of Manchester, Chap. 6 [LION: The Brontes and Mrs. Gaskell]) ¨ ii) But between you and me, she’s a bit of a gossip, and will like hearing all how and about the trial. (1848 Mary Barton, Tale of Manchester, Chap. 25 [LION: The Brontes and Mrs. Gaskell]). ¨ On the assumption that these are homonyms, they do not illustrate type-shifting, only distributional preferences. Rather fun has largely been replaced by very fun; there are, however, no instances of very fun in the UVa corpus, which includes texts up to the 1990’s but four of rather fun. In other words, change is associated with replication and spread (see e.g., Weinreich et al. 1968; Milroy 2003), not with innovation (as proposed by e.g., Lightfoot 1979); however, innovation is a necessary step prior to change. Why particular changes do not occur is a standard conundrum for historical linguistics. For discussion of the importance of analogy in grammaticalization and morphosyntactic change, and of its status as local reanalysis, see, from di¤erent perspectives, Traugott (2006); Wanner (2006); Fischer (2007); Kiparsky (forthc.).

18. 19.

20. 21.

Sources of data
18thCProse. John Rylands University Library of Manchester: A Corpus of Late Eighteenth Century Prose, Denison, David, Director, van Bergen, Linda, Principal Collaborator. AHD. American Heritage Dictionary, Pickett, Joseph P. et al., eds., Boston: Houghton Mi?in, 2000, 4th ed. LION. Chadwyck Healey website. http:/ /lion.chadwyck.com. MED. The Middle English Dictionary. 1956–2001. Ann Arbor: University of Michigan Press. (see also http:/ /www.hti.umich.edu/dict/med/).

552
OED. UVa.

Elizabeth C. Traugott
Oxford English Dictionary. In progress. Oxford/New York: Oxford University Press, 3 rd ed. http:/ /dictionary.oed.com/. University of Virginia, Electronic Text Center, Modern English Collection. http:/ / etext.lib.virginia.edu/modeng/modeng0.browse.html.

References
Aarts, Bas 2001 English Syntax and Argumentation. 2nd ed. New York: Palgrave. Andersen, Henning 2006 Synchrony, diachrony, and evolution. In Thomsen, Ole Nedegaard (ed.), Competing Models of Linguistic Change. (Current Issues in Linguistic Theory 279.) Amsterdam/Philadelphia: Benjamins, 59–90. Backlund, Ulf ¨ 1973 The Collocation of Adverbs of Degree in English. (Studia Anglistica Upsaliensia, 13.) Uppsala: Acta Universitatis Upsaliensis. Bolinger, Dwight 1972 Degree Words. The Hague: Mouton. Borst, Eugen 1902 Die Gradadverbien im Englischen. Heidelberg: Carl Winter. Bromhead, Helen 2006 The Reign of Truth and Faith: Adverbial and Exclamatory Expressions and Epistemic Verbs in 16th and 17th Century English. Honors thesis, Australian National University. Buchstaller, Isabelle and Elizabeth Closs Traugott 2006 ‘The lady was al demonyak’: Historical aspects of adverb ALL. English Language and Linguistics 10, 345–370. Bybee, Joan 2003 Mechanisms of change in grammaticalization: The role of frequency. In Joseph, Brian D. and Richard D. Janda (eds.), The Handbook of Historical Linguistics. (Blackwell Handbooks in Linguistics.) Oxford/Malden, MA: Blackwell, 602–623. Bybee, Joan 2006 From usage to grammar: The mind’s response to repetition. Language 82, 529–551. Bybee, Joan L., Revere Perkins, and William Pagliuca 1994 The Evolution of Grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago: University of Chicago Press. Croft, William 2001 Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Croft, William and D. Alan Cruse 2004 Cognitive Linguistics. Cambridge: Cambridge University Press. De Smet, Hendrick and Hubert Cuyckens 2007 Diachronic aspects of complementation: constructions, entrenchment, and the matching problem. In Cain, Christopher M. and Geo¤rey Russom (eds.), Studies in the History of the English Language III. (Topics in English Linguistics 53.) Berlin/New York: Mouton de Gruyter, 187–213.

The concepts of constructional mismatch and type-shifting

553

De Swart, Henriette ¨ 1998 Aspect shift and coercion. Natural Language and Linguistic Theory 16, 347– 385. Denison, David 2001 Gradience and linguistic change. In Brinton, Laurel J. (ed.), Historical Linguistics 1999: Selected Papers from the 14th International Conference on Historical Linguistics, Vancouver, 9–13 August 1999. (Current Issues in Linguistic Theory 215.) Amsterdam/Philadelphia: Benjamins, 119–144. 2002 History of the sort of construction family. ICCG2, Helsinki http:/ /lings.ln .man.ac.uk/info/sta¤/dd/default.html. 2006 Category change and gradience in the determiner system. In Kemenade, Ans van and Bettelou Los (eds.), The Handbook of the History of English. (Blackwell Handbooks in Linguistics.) Oxford/Malden MA: Blackwell, 279–304. Du Bois, John W. 1985 Competing motivations. In Haiman, John (ed.), Iconicity in Syntax. (Typological Studies in Language 6.) Amsterdam/Philadelphia: Benjamins, 343– 365. En?eld, Nicholas J. 2005 Micro- and macro-dimensions in linguistic systems. In Marmaridou, Sophia, Kiki Nikiforidou, and Eleni Antonopoulou (eds.), with the assistance of Angeliki Salamoura, Reviewing Linguistic Thought: Converging Trends for the 21st Century. (Trends in Linguistics: Studies and Monographs 161.) Berlin/ New York: Mouton de Gruyter, 313–325. Erman, Britt and Beatrice Warren 2000 The idiom principle and the open choice principle. Text 20, 29–62. Faarlund, Jan Terje 2004 Old Norse Syntax: With a Survey of the In?ectional Morphology and a Complete Bibliography. New York: Oxford University Press. Fischer, Olga 2007 Morphosyntactic Change: Functional and Formal Perspectives. (Oxford Surveys in Syntax and Morphology.) Oxford: Oxford University Press. Francis, Elaine J. and Laura A. Michaelis 2003a Mismatch: A crucible for linguistic theory. In Francis and Michaelis (eds.), Mismatch: Form-Function Incongruity and the Architecture of Grammar. (Lecture Notes 163.) Stanford, CA: CSLI, 1–27. Francis, Elaine J. and Laura A. Michaelis (eds.) 2003b Mismatch: Form-Function Incongruity and the Architecture of Grammar. (Lecture Notes 163.) Stanford, CA: CSLI. Francis, Elaine J. and Etsuyo Yuasa 2006 A multi-modular approach to gradual change in grammaticalization. Paper presented at the LSA Annual Meeting, Albuquerque, January 5th–8th. Fried, Mirjam forthc. Constructions and constructs: Mapping a diachronic process. In Bergs, Alexander and Gabriele Diewald (eds.), Constructions and Language Change. Berlin/New York: Mouton de Gruyter. ¨ Fried, Mirjam and Jan-Ola Ostman (eds.) 2004 Construction Grammar in a Cross-Linguistic Perspective. (Constructional Approaches to Language 2.) Amsterdam/Philadelphia: Benjamins. ? Givon, Talmy 1979 On Understanding Grammar. New York: Academic Press.

554

Elizabeth C. Traugott

Goldberg, Adele E. 1995 Constructions: A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press. 2006 Constructions at Work: The Nature of Generalization in Language. Oxford: Oxford University Press. Gries, Stefan Th. and Anatol Stefanowitsch 2004 Co-varying collexemes in the into-causative. In Achard, Michel and Suzanne Kemmer (eds.), Language, Culture, and Mind. Stanford, CA: CSLI, 225– 236. Haiman, John 1980 The iconicity of grammar: Isomorphism and motivation. Language 56, 515– 540. Haspelmath, Martin 1998 Does grammaticalization need reanalysis? Studies in Language 22, 315–351. Heine, Bernd and Tania Kuteva 2002 World Lexicon of Grammaticalization. Cambridge: Cambridge University Press. Hilpert, Martin 2007 Germanic future constructions: A usage-based study of grammaticalization. Ph.D. dissertation, Rice University. Himmelmann, Nikolaus P. 2004 Lexicalization and grammaticalization: Opposite or orthogonal? In Bisang, Walter, Nikolaus Himmelmann, and Bjorn Wiemer (eds.), What Makes ¨ Grammaticalization: A Look from its Fringes and its Components. Berlin/ New York: Mouton de Gruyter, 19–40. Ho¤mann, Sebastian 2004 Are low-frequency complex prepositions grammaticalized? In Lindquist, Hans and Christian Mair (eds.), Corpus Approaches to Grammaticalization in English. (Studies in Corpus Linguistics 13.) Amsterdam/Philadelphia: Benjamins, 171–210. Hopper, Paul J. and Elizabeth Closs Traugott 2003 [1993] Grammaticalization. (Cambridge Textbooks in Linguistics.) Cambridge: Cambridge University Press. Huddleston, Rodney and Geo¤rey K. Pullum 2002 The Cambridge Grammar of the English Language. Cambridge: Cambridge University Press. Israel, Michael 2004 The pragmatics of polarity. In Horn, Laurence R. and Gregory Ward (eds.), The Handbook of Pragmatics. (Blackwell Handbooks in Linguistics.) Oxford/Malden, MA: Blackwell, 701–723. Kay, Paul 1990 Even. Linguistics and Philosophy 13, 59–111. Keller, Rudi 1994 On Language Change: The Invisible Hand in Language. Trans. by Nerlich, Brigitte. London: Routledge. [Originally published as Sprachwandel: Von der unsichtbaren Hand in der Sprache. (Uni-Taschenbucher 1567.) Tubingen: ¨ ¨ Francke, 1990.] Kennedy, Christopher and Louise McNally 2005 Scale structure, degree modi?cation, and the semantics of gradable predicates. Language 82, 345–381.

The concepts of constructional mismatch and type-shifting

555

Kiparsky, Paul 1968 Linguistic universals and linguistic change. In Bach, Emmon and Robert T. Harms (eds.), Universals in Linguistic Theory. New York: Holt, Rinehart and Winston, 171–202. forthc. Grammaticalization as optimization. In Jonas, Dianne (ed.), The Emergence of Grammatical Structures. Oxford: Oxford University Press. Krug, Manfred G. 2000 Emerging English Modals: A Corpus-based Study of Grammaticalization. (Topics in English Linguistics 32.) Berlin/New York: Mouton de Gruyter. Kuteva, Tania 2001 Auxiliation: An Enquiry into the Nature of Grammaticalization. Oxford/New York: Oxford University Press. Langacker, Ronald W. 1987 Foundations of Cognitive Grammar. Volume I: Theoretical Perspectives. Stanford: Stanford University Press. 2005 Construction Grammars: Cognitive, radical, and less so. In Ruiz de Men? ?ez, Francisco J., and M. Sandra Pena Cervel (eds.), Cognitive doza Iban ? Linguistics: Internal Dynamics and Interdisciplinary Interaction. (Cognitive Linguistics Research 32.) Berlin/New York: Mouton de Gruyter, 101– 159. Lasersohn, Peter. 1999 Pragmatic halos. Language 75, 522–562. Lehmann, Christian 1995 [1982] Thoughts on Grammaticalization. (LINCOM Studies in Theoretical Linguistics 1.) Munchen/Newcastle: LINCOM EUROPA. ¨ Lichtenberk, Frantisek 1991 On the gradualness of grammaticalization. In Traugott, Elizabeth Closs and Bernd Heine (eds.), Approaches to Grammaticalization, vol 1. (Typological Studies in language 19: 1.) Amsterdam/Philadelphia: Benjamins, 37–80. Lightfoot, David 1979 Principles of Diachronic Syntax. (Cambridge Studies in Linguistics 23.) Cambridge: Cambridge University Press. Meillet, Antoine ? 1958 [1912] L’evolution des formes grammaticales. In Meillet, Antoine, Linguistique his? ? torique et linguistique generale. Paris: Champion, 130–148. [Originally publ. in 1912 in Scientia (Rivista di Scienza) 12, No. 26(6)]. Michaelis, Laura A. 2003a Word meaning, sentence meaning, and syntactic meaning. In Cuyckens, ? Hubert, Rene Dirven, and John R. Taylor (eds.), Cognitive Perspectives on Lexical Semantics. (Cognitive Linguistics Research 32.) Berlin/New York: Mouton de Gruyter, 163–210. 2003b Headless constructions and coercion by construction. In Francis, Elaine J. and Laura A. Michaelis (eds.), Mismatch: Form-Function Incongruity and the Architecture of Grammar. Stanford, CA: CSLI, 259–310. ¨ 2004a Entity and event coercion in a symbolic theory of syntax. In Ostman, Jan-Ola and Mirjam Fried (eds.), Construction Grammars: Cognitive Grounds and Theoretical Extensions. Amsterdam/Philadelphia: Benjamins, 45–88. 2004b Type shifting in construction grammar: An integrated approach to aspectual coercion. Cognitive Linguistics 15, 1–67.

556

Elizabeth C. Traugott

Milroy, James 2003 On the role of the speaker in language change. In Hickey, Raymond (ed.), Motives for Language Change. Cambridge: Cambridge University Press, 143–157. Nevalainen, Terttu and Matti Rissanen 2002 Fairly pretty or pretty fair? On the development and grammaticalization of English downtoners. Language Sciences 24, 359–380. ¨ Ostman, Jan-Ola and Mirjam Fried (eds.) 2004 Construction Grammars: Cognitive Grounds and Theoretical Extensions. (Constructional Approaches to Language 3.) Amsterdam/Philadelphia: Benjamins. Paradis, Carita 1997 Degree Modi?ers of Adjectives in Spoken British English. (Lund Studies in English 92.) Lund: Lund University Press. 2000 It’s well weird: Degree modi?ers of adjectives revisited: The nineties. In Kirk, John M., (ed.), Corpora Galore: Analyses and Techniques in Describing English: Papers from the Nineteenth International Conference on English Language Research on Computerised Corpora (ICAME 1998). (Language and Computers 30.) Amsterdam/Atlanta: Rodopi, 147–160. 2001 Adjectives and boundedness. Cognitive Linguistics 12, 47–65. Partee, Barbara 1987 Noun phrase interpretation and type shifting principles. In Groenendijk, Jereon, Dick de Jong, and Martin Stokhof (eds.), Studies in Discourse Representation Theories and the Theory of Generalized Quanti?ers. Dordrecht: Foris, 115–143. Peters, Hans 1994 Degree adverbs in Early Modern English. In Kastovsky, Dieter (ed.), Studies in Early Modern English. (Topics in English Linguistics 13.) Berlin/New York: Walter de Gruyter, 269–288. Pustejovsky, James 1995 The Generative Lexicon. Cambridge, MA: The MIT Press. Quirk, Randolph, Sidney Greenbaum, Geo¤rey Leech, and Jan Svartvik 1985 A Comprehensive Grammar of the English Language. London/New York: Longman. Radden, Gu ¨nter and Klaus-Uwe Panther 2004 Introduction: Re?ections on motivation. In Radden, Gunter and Klaus-Uwe ¨ Panther (eds.), Studies in Linguistic Motivation. (Cognitive Linguistics Research, 28.) Berlin/New York: Mouton de Gruyter, 1–46. Sadock, Jerrold M. 1990 Parts of speech in autolexical syntax. In Hall, Kira, Jean-Pierre Koenig, Michaal Meacham, Sondra Reinman, and Laurel A. Sutton (eds.), Proceedings of the Sixteenth Annual Meeting of the Berkeley Linguistics Society. Berkeley: Berkeley Linguistics Society, 269–281. Selkirk, Elisabeth 1976 Some remarks on Noun Phrase structure. In Culicover, Peter W., Thomas Wasow, and Adrian Akmajian (eds.), Formal Syntax. New York: Academic Press, 285–316. Sto¤el, C. 1901 Intensives and Down-toners: A Study in English Adverbs. (Anglistische Forschungen 1.) Heidelberg: Carl Winter.

The concepts of constructional mismatch and type-shifting

557

Svorou, Soteria 1993 The Grammar of Space. (Typological Studies in Language 25.) Amsterdam/ Philadelphia: Benjamins. Tabor, Whitney 1993 The gradual development of degree modi?er sort of and kind of: A corpus proximity model. In Beals, Katharine, Gina Cooke, David Kathman, Sotaro Kita, Karl-Erik McCullough, and David Testen (eds.), Papers from the 29th Regional Meeting of the Chicago Linguistic Society. Chicago: Chicago Linguistic Society, 451–465. Traugott, Elizabeth Closs 2006 Constructions, emergent constructions, and the notion of ‘‘newness’’. Paper presented at HDLS, Albuquerque, NM, November 9th–11th. German translation to appear as Grammatikalisierung und emergente Konstruktionen. In Stefanowitsch, Anatol, and Kerstin Fischer (eds.). Konstruktionsgrammatik und grammatische Konstruktionen. Tubingen: Stau¤enburg. ¨ forthc. a The grammaticalization of NP of NP constructions. In Bergs, Alexander and Gabriele Diewald (eds.), Constructions and Language Change. Berlin/ New York: Mouton de Gruyter. forthc. b Grammaticalization, constructions and the incremental development of language: Suggestions from the development of degree modi?ers in English. In Eckardt, Regine, Gerhard Jager, and Tonjes Veenstra (eds.), Variation, Se¨ lection, Development: Probing the Evolutionary Model of Language Change. Berlin/New York: Mouton de Gruyter. Traugott, Elizabeth Closs and Richard B. Dasher 2002 Regularity in Semantic Change. (Cambridge Studies in Linguistics 97.) Cambridge: Cambridge University Press. Trousdale, Graeme forthc. Words and constructions in grammaticalization: The end of the English impersonal construction. In Minkova, Donka and Susan Fitzmaurice (eds.), Empirical and Analytical Advances in the Study of English Language Change. (Topics in English Linguistics.) Berlin/New York: Mouton de Gruyter. Wanner, Dieter 2006 The Power of Analogy: An Essay on Historical Linguistics. (Trends in Linguistics; Studies and Monographs 170.) Berlin/New York: Mouton de Gruyter. Weinreich, Uriel, William Labov and Marvin I. Herzog 1968 Empirical foundations for a theory of language change. In Lehmann, W. P., and Yakov Malkiel (eds.), Directions for Historical Linguistics. Austin: University of Texas Press, 95–189. Wierzbicka, Anna 2006 English: Meaning and Culture. Oxford/New York: Oxford University Press. Yuasa, Etsuyo and Elaine J. Francis 2003 Categorial mismatch in a multi-modular theory of grammar. In Francis, Elaine J. and Laura A. Michaelis (eds.), Mismatch: Form-Function Incongruity and the Architecture of Grammar. (Lecture Notes 163.) Stanford, CA: CSLI, 179–227. Ziegeler, Debra 2007 A word of caution on coercion. Journal of Pragmatics 39, 990–1028.


相关文章:
12. 学The Concept shift of expression in trans analysis
12. 学The Concept shift of expression in trans analysis_英语学习_外语学习_...The second type demands from the translator that he should aspire to match...
更多相关标签:
constructional | downshifting | string shifting | risk shifting | shifting cultivation | shifting | tense shifting | phase shifting |