Abstract

Appeals to idiom data have played an important role in arguments about the relation between English prepositional datives (Mary gave a present to John) and double object forms (Mary gave John a present).The claimed existence of idioms in both types has led researchers to conclude that the two constructions are independent, with no derivational relation between them. This article shows that the factual claims are mistaken: no English dative idioms exist in either type. Forms like give ~ the creeps/show ~ the ropes are not double object idioms because they are not idioms—they are fully compositional. Forms like throw ~ to the wolves/send ~ to the showers are not dative idioms because they are not datives—they are caused-motion constructions. The former misanalysis arises from a confusion of idioms with collocations. The latter misanalysis arises from a simplistic view about the syntax-semantics mapping, namely, that oblique datives univocally express caused motion. Given that English dative idioms do not exist, arguments about the derivational relatedness of dative forms in English must appeal to other data.

Research into the relation between English PP dative and double object pairs like (1a–b) has debated the full space of possibilities.

(1)

  • a.

    Mary gave a present to John. PP dative construction

  • b.

    Mary gave John a present. Double object construction

Chomsky (1955) introduced the classical transformational analysis of Dative Shift, which derives dative object constructions from PP datives (2a).1Bowers (1981), Dryer (1987), Kitagawa (1994),and Hallman (2015) have proposed an Anti–Dative Shift operation that derives PP datives from double object constructions (2b). And beginning with Oehrle (1976), many have argued that the two constructions are simply independent, with no derivational relation in either direction (2c).

(2)

  • a.

    PPD → DOC Dative Shift

  • b.

    PPD ← DOC Anti–Dative Shift

  • c.

    PPD ←X→ DOC Derivational independence

In this debate, idiom data have played an important role. Idiom availability in a given syntactic form has been taken as evidence for its underived status. Thus, in Larson 1988 I analyze the bold faced items in (3) as “PP dative idioms” and offer them as evidence for the underived status of (at least some) PP datives. This favors analyses (2a) and (2c) over analysis (2b).

(3)

  • a.

    Lasorda sent his starting pitcher to the showers.

  • b.

    Mary took Felix to the cleaners.

  • c.

    Felix threw Oscar to the wolves.

  • d.

    Max carries such behavior to extremes.

Harley (1995, 2002), Bruening (2001), and Richards (2001), among others, analyze the boldfaced items in (4) as “double object construction idioms” and offer them as evidence for the underived status of (at least some) double object constructions. This view favors (2b) and (2c) over (2a).

(4)

  • a.

    The Count gave me the creeps.

  • b.

    John gave Mary flak (about/during her presentation).

  • c.

    His boss gave Max the boot.

  • d.

    Mary gave John a kick.

  • e.

    Alice gave me a piece of her mind.

Taken together, the data in (3) and (4) appear to support (2c), a nonderivational view, and indeed many have drawn this conclusion (see, e.g., Harley 2002).

In this article, I revisit “dative idioms” in English, arguing that discussion in this domain has been confused about both parts of the term. As I show, examples like (4a–e) are not “double object idioms” because they are not idioms, at least not if this term is understood in a way relevant to structure projection. Furthermore, examples like (3a–d) are not “PP dative idioms” because they are not dative constructions; rather, they are expressions of caused motion. The combined result, I show, is that English contains no dative idioms at all, of either the oblique or the double object variety. It follows that this domain of data is not a secure basis from which to argue for the derivational vs. nonderivational status of the dative alternation. Arguments must rely on other sources of empirical evidence.

In section 1, I briefly review the relevance of idioms for various structural analyses of ditransitives. In section 2, I consider the general notion of “idiom” and the related concept of “collocation,” arguing that only the former is relevant for syntactic projection. In section 3, I examine purported cases of double object idioms: datives of the form V–DP1–DP2, where V and DP2 make up an idiomatic unit. I show that examples like (4a–e) involving give are compositional and likewise for purported double object idioms involving other verbs (show, promise, etc.). I conclude that English contains no (exclusively) double object idioms. In section 4, I turn to putative PP dative idioms: datives of the form V–DP–PP, where V and PP make up an idiomatic unit. I argue that idioms of the form give N to DP are not of the relevant type and indeed are arguably not ditransitive constructions at all. I then argue that although examples like (3a–d) are idioms, they are not dative idioms (contra Larson 1988) insofar as they are not (idiomatic) caused possessives. Rather, they are (idiomatic) caused-motion constructions.

1 Dative Constituency and Idioms

Early analyses of English ditransitive constructions assumed “flat” VPs like those in (5a–b).

(5)

graphic

Modern approaches, by contrast, have posited a more complex hierarchical organization among VP-internal elements. Many of these are explicitly proposed to accommodate purported ditransitive idiom data.

1.1 PP Datives

Chomsky (1993) (developing ideas from Larson 1988) proposes that PP datives like (1a) have structure (6a), where “light” v introduces the external argument. Harley (1995, 2002) proposes(6b), which decomposes give in this usage as CAUSE + LOC.

(6)

graphic
graphic

In (6a–b), V/P and the oblique argument form a constituent excluding the theme. In Larson 1988:340, I offer “PP dative idioms” like (7a–d) in support of such structures.2

(7)

  • a.

    Lasorda sent his starting pitcher to the showers.

    ‘Lasorda removed/pulled his starting pitcher (from the game).’

  • b.

    Mary took Felix to the cleaners.

    ‘Mary swindled Felix.’

  • c.

    Felix threw Oscar to the wolves.

    ‘Felix sacrificed Oscar.’

  • d.

    Max carries such behavior to extremes.

    ‘Max overdoes such behavior.’

Assuming that what is interpreted as an idiomatic semantic unit must project into structure as a syntactic unit, (6a) can represent the idiomaticity of send to the showers as in (8), where the relevant constituent is V′.

(8)

graphic

(6b) represents the idiomaticity of (7a) more abstractly with the P′ LOC + the showers, as in (9).

(9)

graphic

The retained figurative sense in (10a–c) is compatible with this more abstract view on the plausible assumption that LOC is also a decompositional constituent of the verb go.

(10)

  • a.

    The starting pitcher went to the showers.

  • b.

    ?Oscar went to the wolves.

  • c.

    Max’s behavior went to extremes.

1.2 Double Object Constructions

The structure of double object constructions has been a rich area of modern syntactic theorizing. Modifying Marantz 1993, Bruening (2001) proposes that double object constructions like (1b) have the structure in (11a), where an APPL(ICATIVE) head introduces the experiencer/goal.3 Harley (1995, 2002) suggests the alternative (11b), which decomposes give (in this usage) into CAUSE + HAVE.

(11)

graphic

Marantz (1993) himself proposes the structure in (12a) for double object constructions, and in Larson 2014, I argue for the derivational variant of Marantz’s view in (12b), where the sequence of verb movements is the same but where, in addition, the goal argument raises to the specifier of APPL from the low position that Marantz (1993) identifies as X.

(12)

graphic

In (11a–b) and in (12a), the V/P and the theme argument form a constituent that excludes the goal argument. Harley (1995, 2002) and Richards (2001) offer the boldfaced items in (13) as “double object idioms” in support of this constituency.

(13)

  • a.

    The Count gave me the creeps.

    ‘The Count unnerved/unsettled me.’

  • b.

    John gave Mary flak.

    ‘John criticized Mary.’

  • c.

    His boss gave Max the boot.

    ‘His boss fired Max.’

  • d.

    Mary gave John a kick.

    ‘Mary thrilled John.’

  • e.

    Alice gave me a piece of her mind.

    ‘Alice upbraided me.’

(11a)/(12a) can represent give the creeps (meaning ‘induce a feeling of apprehension or horror’) as in (14a–b), where either V′ or VP is the relevant idiomatic unit that is projected.4

(14)

graphic
graphic

The decompositional structure (11b) presumably represents (13a) more abstractly, as HAVE + the creeps (15).

(15)

graphic

Note again that the retained figurative sense in (16a–d) is compatible with this view on the plausible assumption that HAVE is a decompositional constituent of the verb get.

(16)

  • a.

    I got the creeps.

  • b.

    Mary got flak.

  • c.

    John got the boot.

  • d.

    I got a piece of her mind.

However, (12b), the derivational variant of Marantz’s (1993) (12a), is not compatible with the claimed idiomaticity of (13a–e), since (12b) makes no constituent available corresponding to the relevant string (17).

(17)

graphic

Thus, if (13a–e) are truly double object idioms, and if idioms are projected as constituents, structures like (12b) will not represent these examples appropriately. An alternative syntax must be assumed.

Arguments like these involving “dative idiom” data evidently presuppose correct understanding of the two parts making up the term: dative and idiom. However, in the context of ditransitives, both terms appear to have been consistently misunderstood, leading to misanalysis. I begin with idiom.

2 Formulaic Language: Idioms, Collocations, and Lexical Bundles

Idioms are widely identified as a subspecies in the broader class of formulaic language: multiword expressions having a single meaning or function that appear to be stored as a unit in memory and retrieved as a unit in processing (Wood 2015).5 Other instances of formulaic language are collocations and lexical bundles.

2.1 Idioms

The notion of idiom typically deployed in linguistics is one also typically offered by dictionaries and lexicographers. (18a) is drawn from a popular introductory syntax textbook and (18b–c) are from standard dictionaries of English.

(18)

  • a.

    “We can define idioms as expressions . . . which have an idiosyncratic meaning that is not a purely componential function of their individual parts.” (Radford 1997: 159)

  • b.

    idiom n. 1. A speech form or an expression of a given language that is peculiar to itself grammatically or cannot be understood from the individual meanings of its elements, as in keep tabs on. (The American Heritage Dictionary of the English Language (AHDEL), 4th ed. 2006)

  • c.

    idiom n. 1. An expression whose meaning is not predictable from the usual meanings of its constituent elements, as kick the bucket, hang one’s head, etc., or from the general grammatical rules of a language, as the table round for the round table, and which is not a constituent of a larger expression of like characteristics. (Random House Unabridged Dictionary of American English, 2016)

The core notion in (18a–c) is (non)compositionality: idioms are formulaic constructions whose interpretations are unpredictable from individual lexical meanings plus normal composition rules.6 Thus, sell ~ down the river is an idiom since its meaning (‘betray’) cannot be predicted from its VP structure together with the meanings of its individual words sell, down, and river.

2.2 Collocations

Distinct from idioms are collocations. (19a–b) give two standard definitions of the latter.

(19)

  • a.

    collocation n. An arrangement or juxtaposition of words or other elements, especially those that commonly co-occur, as rancid butter, bosom buddy, or dead serious. (AHDEL)

  • b.

    collocation n. The act or result of placing or arranging together; specif: a noticeable arrangement or conjoining of linguistic elements (as words). (Webster’s Seventh New Collegiate Dictionary (WSNCD), 1969)

The core concept in collocation is “conventionalized cooccurrence.” Collocations are strings of words used together frequently and recognized as such by speakers.7 Beyond rancid butter, bosom buddy, and dead serious, collocations plausibly include familiar proverbs like (20a–c) as well as the large family of items identified in Jackendoff 1997 as the “Wheel of Fortune corpus,” which comprises phrases, names of all sorts, clichés, titles, and quotations (21a–e).8

(20)

  • a.

    He who laughs last, laughs best.

  • b.

    Every cloud has a silver lining.

  • c.

    A friend in need is a friend indeed.

(21)

  • a.

    frequent flyer program, black and white film

  • b.

    Count Dracula, Beverly Hills, John Deere tractor, American Heart Association

  • c.

    gimme a break, no money down, we’re doing everything humanly possible

  • d.

    All You Need Is Love, The Price Is Right

  • e.

    beam me up, Scotty; may the Force be with you; a day that will live in infamy (Jackendoff 1997:154–155)

All of these constitute “noticeable arrangement[s] or conjoinings of linguistic elements,” as evidenced by the fact that speakers readily recall them as units.9

Idiom and collocation are not coextensive terms. Whereas all idioms are collocations insofar as they are conventionalized or fixed, the vast majority of collocations are not idioms insofar as they are compositional. The last point is plain with the proverbs in (20a–c), as well as many of the phrases identified by Jackendoff (1997) as belonging to the Wheel of Fortune corpus (see(21c–e)).10

Consider the examples mentioned specifically in definition (19a). Rancid butter is an attributive adjective construction. Rancid has the basic meaning of ‘spoiled’ but is understood as applying exclusively to fats or oils, hence rancid butter/oil/cream/lard but not *rancid beer. Here, conventionalized cooccurrence is presumably based upon the semantically restricted class of Ns that rancid can modify given its specific meaning. The construction is fully compositional, however, involving simple intersective modification; a substance is rancid butter if and only if it is rancid and it is butter. Bosom buddy is an also attributive construction. Bosom occurs primarily in English as a noun with the meaning of ‘chest’. But in the specific context of “Ns of intimate association,” it appears to function as an adjective with the sense of ‘close’ or ‘beloved’—hence bosom buddy/friend/chum/companion/comrade but not *bosom acquaintance. As with rancid butter, the collocation involves a semantically restricted class of N heads; but here the collocate bosom itself assumes a new meaning accompanying its change of category.11 Within this context, the construction seems fully compositional, with the head (friend, companion, comrade, etc.) showing its usual meaning and with bosom functioning analogously to close in close friend, close companion, and so on, presumably a form of degree qualification (‘friend to an intimate degree’). Finally, consider dead serious. Dead occurs primarily in English as an adjective meaning ‘no longer alive’ or ‘numb/unresponsive’. In a restricted range of constructions, however, dead appears to function as an intensifying adverb expressing ‘high or ultimate degree’, often with a connotation of gravity. This sense presumably traces to uses like deadly/deathly tired and dead tired with the sense ‘tired to the point of death’/‘tired to a grave degree’. The sense of extremity and gravity recurs in dead(ly)/deathly serious, dead(ly)/deathly earnest, although here the sense is ‘to a high degree’, with the gravity apparently transferring to the topic of discussion (‘highly serious about a grave topic’, etc.).12 That dead is nonpredicative in this context is indicated by the contrast between(22a–d) and (23a–d).

(22)

  • a.

    The nerve is dead.

  • b.

    The nerve deadened (with time).

  • c.

    The dentist deadened the nerve.

  • d.

    The deadness of the nerve was in dispute.

(23)

  • a.

    *The seriousness is dead.

  • b.

    *The seriousness deadened.

  • c.

    *The dentist deadened his seriousness.

  • d.

    *The deadness of his seriousness surprised us.

Once again, relative to the particular context in question, dead serious appears to be fully compositional; the semantics is that of an intensifying adverb coupled with the usual meaning of the AP head (serious, earnest, sure, certain, etc.).

2.3 Lexical Bundles

Distinct from both idioms and collocations are lexical bundles (Conklin and Schmitt 2012, Columbus2013), which are typically identified by corpus search and include expressions like those in (24) (Biber and Conrad 1999, Biber, Conrad, and Cortes 2004, Wood 2015).

(24)

  • a.

    you don’t have to, I’m not going to

  • b.

    it’s going to be, that’s one of the, and this is a

  • c.

    are you going to, do you want to

  • d.

    I want you to, you might want to

  • e.

    to be able to, to come up with

  • f.

    the end of the, a little bit of, a little bit about

Lexical bundles are multiword strings that, although not typically regarded as “phrases of the language” by lexicographers, are nonetheless identifiable statistically and by intuition as recurrent sequences. Lexical bundles have been argued to figure importantly in L1 acquisition by children (Bannard and Matthews 2008, Bannard and Lieven 2012); their frequency has also been shown to distinguish L1 and L2 learners of academic English (Chen and Baker 2010).

2.4 Formulaic Language and Constituency

The idiom/collocation/lexical bundle distinction is significant for linguistic theory with regard to constituency and structure projection.

Idioms have been argued to project as constituents in underived structure, that is, as structure generated only by External Merge in the sense of Chomsky 1995.13 Although controversial, this view can be rationalized in plausible ways. To say that the meaning of a string cannot be computed from the meanings of its parts and their structure is to say that, from a semantic standpoint, the string constitutes a single, unanalyzable object. It is at least plausible to think that what the grammar interprets as a single semantic object it also represents as a single syntactic object—as a constituent. In other words, it is at least plausible to think that the grammar treats idioms as “Saussurean signs.”14

No such reasoning seems available with collocations and lexical bundles, however. Although it is plausible to think that constituent status would be a natural feature of multiword strings as an aid to memory, there is no reason to think it would be a necessary feature—that only multiword strings that form constituents can be collocations and lexical bundles. The latter would be to claim, in effect, that only constituents can be stored and recalled. Note that none of the lexical bundles in (24) are constituents in the adult grammar and plausibly they are not constituents in children’s grammar either, despite their apparently being internalized. Lexical bundling does not require constituency.

Likewise, it does not seem plausible that status as a collocation requires constituency. Tomy knowledge, no one has argued from the collocational status of rancid butter to an analysis of (25) wherein the boldfaced items form an underlying constituent that has been derivationally separated.

(25) rancid yellow creamery butter

Status as a collocation thus is not diagnostic for constituency.15

The distinction among idioms, collocations, and lexical bundles and their contrasting implications for constituency may seem obvious, but in fact the two have been confused in the literature. For example, in an article on ditransitive idioms and the structure of ditransitive sentences in Japanese, Kishimoto (2008) offers the statements in (26).

(26)

  • a.

    “Japanese abounds with dative-V and accusative-V idioms owing to their productivity.” (p. 148)

  • b.

    “Even though the degree of idiomaticity might differ from one idiom to another, the idioms in (11) and (12) are thought to be fixed expressions in the sense that they consist of fixed combinations of lexical items.” (p. 149)

Given idioms as expressions whose meanings precisely cannot be predicted in a rule-governed way from the meanings of their parts, the first statement in (26) is startling. How can a language productively have a class of non-rule-governed expressions? The second statement reveals the source of the confusion. Kishimoto (2008) reasons—illegitimately, in my view—from idioms(“noncompositional phrases”) to collocations (“fixed phrases”), invoking the latter as evidence for constituency when only the former are relevant.16 As we will see, similar confusions haveoccurred in discussions of ditransitive constructions in English.

3 Reexamining “Double Object Idioms”

3.1 Evaluating Idiomaticity

In section 1.2, I noted the claim by Harley (1995, 2002) and Richards (2001) that the boldfaced items in (13a–e) (repeated here as (27a–e)) are “double object idioms” and as such potential evidence for double object structure constituency.

(27)

  • a.

    The Count gave me the creeps.

    ‘The Count unnerved/unsettled me.’

  • b.

    John gave Mary flak.

    ‘John criticized Mary.’

  • c.

    His boss gave Max the boot.

    ‘His boss fired Max.’

  • d.

    Mary gave John a kick.

    ‘Mary thrilled John.’

  • e.

    Alice gave me a piece of her mind.

    ‘Alice upbraided me.’

Given the results in section 2, evaluating this claim is a matter of evaluating whether these forms are compositional. To get clear on what’s involved, let’s consider first the give-constructions in(28a–b).

(28)

  • a.

    John gave Mary the mumps.

  • b.

    The view gave John vertigo.

Our intuition is surely that these are compositional. Alongside gave Mary the mumps (28a), we also have gave Mary the measles, gave Mary chickenpox, gave Mary a cold, and so on. The paradigm is completely productive insofar as any expression denoting a communicable illness can be substituted for mumps. Similarly, alongside gave John vertigo in (28b) we also have gave John the bends, gave John the DTs, gave John heartburn, and so on, where we freely substitute other expressions denoting psychophysical states for vertigo.

How would we actually demonstrate our intuition to be correct—that give Mary the mumps and give John vertigo are in fact compositional? Evidently, we would do so by showing that the meanings of the phrases in question can be computed from the meanings of their parts and the structure in which they occur. Suppose that, following many authors, we take double object give to express ‘caused possession’. Concretely, assume the informal composition rule in (29), which maps a VP consisting of give followed by two DP arguments to a caused-possession meaning.17

(29) [VP give DP1 DP2] → ‘cause DP1 to have/get DP2

Next, let us simply look up the boldfaced nominal elements in (28a–b) in a standard English dictionary. WSNCD, for example, provides the entries in (30a–b) for mumps and vertigo.

(30)

  • a.

    mumps pl.n. (used with a sing. or pl. verb) An acute, inflammatory, contagious disease caused by a paramyxovirus and characterized by swelling of the salivary glands, especially the parotids . . .

  • b.

    vertigo n 1a: a disordered state in which the individual or his environs seem to whirl dizzily . . .

Substituting the results in (30a–b) into (29), adjusting for tense, and suppressing some details, we then derive the outcomes in (31a–b), respectively, for our VPs.

(31)

  • a.

    gave Mary the mumps: ‘caused Mary to have an acute, inflammatory, contagious disease caused by a paramyxovirus and characterized by swelling of the salivary glands, especially the parotids’

  • b.

    gave John vertigo: ‘caused John to have a disordered state in which he or his environs seemed to whirl dizzily’

If these are the correct meanings for (28a–b), then we have demonstrated the compositionality, and hence the nonidiomaticity, of give Mary the mumps and give John vertigo. In my judgment(and that of others), (31a–b) are indeed the correct meanings. Hence, our original intuitions are confirmed.

This reasoning is completely general and extends to other cases where intuitions are less secure. Consider, for example, (32a–c). Are the VPs in question compositional or do the boldfaced elements constitute idioms?

(32)

  • a.

    Students gave Max backtalk.

  • b.

    Alice gave me a (cheery) wave.

  • c.

    My landlord gave me two weeks’ notice.

Proceeding as before, we assume the scheme in (29), look up the boldfaced nominals in a standard dictionary, and compute by substitution. WSNCD provides the entries in (33a–c) for backtalk, wave, and notice. Substituting, we derive the phrasal meanings in (34a–c), again adjusting for tense and suppressing some details.

(33)

  • a.

    backtalk n 1a: an impudent, insolent or argumentative reply

  • b.

    wave n . . . 4: a sweep of hand or arm or of some object held in the hand used as a signal or greeting

  • c.

    notice n . . . 1a . . . (3): notification by one of the parties to an agreement or relation of intention of terminating it at a specified time

(34)

  • a.

    gave Max backtalk: ‘caused Max to get an impudent, insolent, or argumentative reply’

  • b.

    gave me a wave: ‘caused me to get a sweep of hand or arm or of some object held in the hand used as a signal or greeting’

  • c.

    gave me two weeks’ notice: ‘caused me to have notification as the party to an agreement or relation of intention of terminating in two weeks’ time’

Once again, these meanings appear to be correct. Hence, once again we confirm the constructions in (32) as compositional, and not idiomatic.

3.2 Is Give the Creeps an Idiom?

With these results in mind, let us reconsider (13a–e)/(27a–e) (henceforth (27a–e)). If these are truly double object idioms, then attempts to compute their meanings along the lines of (28a–b) and (32a–c) should fail and/or deliver the wrong results. We should not be able to derive the appropriate VP meanings from the meanings of their parts and the structure in which they occur. Is that true?

Again we assume that double object give expresses caused possession, adopting the composition rule in (29). Again we consult a standard dictionary, checking entries for the boldfaced nominal elements in (27a–e). WSNCD and AHDEL offer the entries in (35a–e) for creep, flak, boot, kick, and piece of (one’s) mind:

(35)

  • a.

    creep n 1: a movement of or like creeping 2: a distressing sensation like that caused by the creeping of insects over one’s flesh; esp. a feeling of apprehension or horror—usu. used in the plural. (WSNCD)

  • b.

    flak n . . . 2Informala. Excessive or abusive criticism. b. Dissension, opposition. (AHDEL)

  • c.

    boot n . . . 6Brit a blow delivered by or as if by a booted foot: KICK; also : a rude discharge or dismissal. (WSNCD)

  • d.

    kick n . . . 5Slang a feeling of pleasurable stimulation. (AHDEL)

  • e.

    piece n . . . -idioms a piece of (one’s) mind Frank and severe criticism; censure.

Substituting the results in (35a–e) into (29), adjusting tense, and suppressing some details, we derive the outcomes in (36a–e) for our VPs.

(36)

  • a.

    gave me the creeps: ‘caused me to have/get a feeling of apprehension or horror’

  • b.

    gave Mary flak: ‘caused Mary to have/get excessive or abusive criticism’

  • c.

    gave Max the boot: ‘caused Max to have/get a rude dismissal’

  • d.

    gave John a kick: ‘caused John to have/get a feeling of pleasurable stimulation’

  • e.

    gave me a piece of her mind: ‘caused me to have/get frank and severe criticism; censure’

If (27a–e) are idioms, then (36a–e) should be incorrect. (36a), for example, should fail to render the correct meaning of give me the creeps. But in fact (36a) is correct. In my judgment (and that of others), (36a) does render the meaning of give me the creeps. If so, then (27a) is indeed compositional, and hence give the creeps is not an idiom. The same outcome, and the same conclusion, holds for (27b–e).18

Upon reflection, the results for (36a) are unsurprising. The dictionary entry in (35a) evidently takes the creeps to denote a psychophysical state, broadly comparable to the mumps, the bends, heartburn, or vertigo. And indeed the grammar of English treats these expressions in comparable ways. Thus, the creeps can be modified in ways parallel to the bends or the measles (compare(37a) with (37b–c)).

(37)

  • a.

    The Count gave me an amazingly bad and violent case of the creeps/a case of the creeps that lasted for weeks/creeps that I just couldn’t shake.

  • b.

    The dive gave me an amazingly bad and violent case of the bends/a case of the bends that lasted for weeks/bends that damn near killed me.

  • c.

    My trip abroad gave me an amazingly bad and violent case of the measles/a case of the measles that lasted for weeks/measles that covered my body.

This kind of behavior is at least atypical of idioms, where the presence of a modifier often forces literal interpretation.19

(38)

  • a.

    The thick fur flew.

  • b.

    The shit hit the slowly whirling fan.

  • c.

    John kicked the large bucket.

Genuine idioms also typically resist substitution of subparts, even by elements of similar sense, on pain of reverting to literal meaning; compare (39a–d).

(39)

  • a.

    The cat got out of the bag/#sack/#tote/#pouch.

  • b.

    The fur/#hair flew.

  • c.

    The shit hit/#struck/#collided with the fan.

  • d.

    John kicked the bucket/#pail.

By contrast, give ~ the creeps freely permits substitution in its nominal portion, allowing a broad range of variant forms with similar or related meanings.

(40) The Count gives me the creeps/the willies/the shivers/the shakes/the chills/the jitters/goosebumps/gooseflesh/the fits/the heebee-jeebies/the screamin’ meemies.

These results appear to generalize to (27b–e) as well, as shown in (41)–(42).

(41)

  • a.

    John gave Mary flak (about/during her presentation).

  • b.

    John gave Mary a lot of flak/far more flak than anyone had expected.

  • c.

    John gave Mary flak/static/shit/crap/hassle/trouble.

(42)

  • a.

    Her boss gave Mary the boot.

  • b.

    Her boss gave Mary the royal order of the boot.

  • c.

    We will all be for the boot by Friday.

  • d.

    Her boss gave Mary the boot/the sack/the chop.

Thus, none of these give-constructions cited as double object idioms appear to be idioms infact. All appear to be fully compositional—their meanings are predictable from the meanings of their parts and the structure in which they occur—and all behave comparably to similar nonidiomswith respect to other tests. These results (and standard dictionary practice) support the general conclusion drawn in Larson 1988 that the figurative sense intuited in examples like (27a–e) resides exclusively in the contribution of the final nominal (creeps, flak, boot, kick, etc.) and is not a property of the construction as a whole. Note in particular that no change in meaning occurs for give, the head of the purported VP-idiom. In all cases, give is interpreted uniformly as caused possession (29).

Richards (2001) asks why, if this picture is correct, a nominal like the boot cannot occur freely with its ‘rude dismissal’ meaning in examples like (43a–c).20

(43)

  • a.

    *I was sorry to hear about the boot.

  • b.

    *The boot has ruined many an employee’s Christmas.

  • c.

    *The boot’s legacy is often severe psychological problems.

While this is an interesting question—presumably part of the broader question of why lexical items do not exhibit all their meanings in all their occurrences—the discussion in section 2 shows it to be irrelevant to the issue of idiomaticity (and hence constituency). What Richards is observing is that give ~ the boot is a collocation (that its parts cooccur), not that it is an idiom (that it is noncompositional).21 More precisely, cases like give ~ the boot closely resemble collocations like heavy smoker discussed by Schenk (1995), which consist of a base (smoker) and a collocate(heavy) and wherein “the base in general has an interpretation that it has outside the collocation. The meaning of the collocate, however, is not standard; for example heavy usually refers to weight and not to smoking much” (Schenk 1995:266). Schenk illustrates the distributional effects of this distinction with (44) and (45) from Lakoff 1970. Thus, the adjective hard in its normal meaning of ‘solid’ (as in hard metal) shows the full set of possibilities with respect to inchoative and causative formation and nominalization (44a–d). By contrast, hard in its collocate meaning of ‘difficult’ (as in hard problem) is more restricted (45a–d).22

(44)

  • a.

    The metal is hard.

  • b.

    The metal hardened.

  • c.

    John hardened the metal.

  • d.

    the hardness of the metal

    (Schenk 1995:266, (39))

(45)

  • a.

    The problem is hard.

  • b.

    *The problem hardened.

  • c.

    *John hardened the problem.

  • d.

    *The hardness of the problem surprised us.

    (Schenk 1995:266, (40))

Note, however, that despite the more restricted distribution of hard on its collocate meaning, hard problem is not an idiom. Rather, it shows the normal semantic composition of intersective A-N constructions (‘X is hard and X is a problem’).23 The situation appears to be parallel with all of the putative “double object give idioms.” In all cases, the base (give) has its normal meaning whereas the outer nominal (creeps, boot, flak, kick, etc.) patterns like a collocate. Appeal to datalike (43a–c) in regard to idiom status thus seems to be an instance of the same general confusion noted earlier in Kishimoto 2008. Idioms and collocations (“fixed phrases”) are simply not the same thing.24

3.3 Non-Give “Double Object Idioms”?

Examples like (27a–d) all involve the specific verb give.25 However, candidate double object idioms have been proposed with other dative verbs. (46a–d) are representative examples.

(46)

  • a.

    Mary showed John the ropes.

    ‘Mary explained to John how to do a particular job or activity.’

  • b.

    John promised Mary the moon.

    ‘John made extravagant/unfulfillable promises to Mary.’

  • c.

    Mary showed John the door.

    ‘Mary ejected John.’

  • d.

    Max read John the riot act.

    ‘Max reprimanded John severely.’

Are (46a–d) double object idioms in fact?

3.3.1 Show ~ the Ropes

The situation with show ~ the ropes in (46a) appears comparable to the give cases in (27a–e) discussed above insofar as the sense of idiomaticity arises through a particular semantic contribution of the object phrase. Thus, AHDEL lists the subentry in (47) under rope.

(47) 7. ropesInformal Specialized procedures or details

Under plausible assumptions about show as meaning ‘teach’ or ‘explain’, it seems clear that the meaning for (46a) could be computed compositionally as (approximately) ‘Mary explained to John the specialized procedures or details’. The particular sense of ropes observed in (46a) recursin a small set of other examples like (48a–c).

(48)

  • a.

    John learned the ropes.

  • b.

    John knows the ropes.

  • c.

    Mary taught John the ropes.

Roughly put, the situation seems to be that the ropes refers (as a collocate) to “specialized procedures or details,” and the semantic field that determines its collocate distribution is the family of verbs describing instruction in and the acquisition, demonstration, and possession of such procedural knowledge. Show the ropes thus appears equivalent to the give-constructions in(27a–d). It is compositional and not an idiom at all.26

3.3.2 Promise ~ the Moon

The situation with promise ~ the moon in (46b) appears somewhat different from that with give ~ the creeps or show ~ the ropes. Here, the “figurative content” that is intuited does not seem to arise from a special collocate meaning of the object phrase (the moon); rather, it seems to arise from a more general source.

Commissive verbs like promise are well-known to impose specific felicity conditions on their use. Concretely, for a successful speech act of promising, the speaker must in principle be capable of carrying out the act promised, and the addressee must in principle be capable of receiving it. Thus, if X promises Y Z, it must be possible for X to give Y Z and it must be possible for Y to have Z.27 Since the moon (the stars, etc.) cannot be given and cannot be owned, any use of examples like (46b) will necessarily involve flouting the felicity conditions on promising—a conspicuous violation that the speaker intends to be perceived as such by the listener. I suggest that in these situations (as in situations of flouting generally), listeners calculate(and are intended to calculate) the figurative ‘make an extravagant/unfulfillable promise’ meaning as an implicature. If this view is correct, then strictly speaking, the additional component of meaning in (46b) does not arise through normal compositional semantic means. But neither is it “listed” lexical content, as with an idiom. Rather, it is calculated or is at least calculable by the listener, as an implicature.28 I should note that there remains a collocational element to this construction insofar as it is the moon or some other astronomical body that is typically invoked (e.g., John will promise you anything: the sun, the moon, and the stars); but the special figurative meaning of this construction nonetheless arises by regular pragmatic means.

3.3.3 Show ~ the Door

AHDEL considers show ~ the door in (46c) to be a genuine idiom, offering the subentry in (49) under the main entry for door.

(49) door: . . . -Idioms: show (someone) the door. 1. To eject (someone) from the premises. 2. To terminate the employment of; fire.

However, the diagnosis of show ~ the door as a double object idiom appears to be based on a misanalysis of the relevant forms.

English show participates in the dative alternation with the approximate meaning of ‘cause to see’.

(50)

  • a.

    Mary showed the table to John.

    ‘Mary caused John to see the table.’

  • b.

    Mary showed John the table.

    ‘Mary caused John to see the table.’

It is from this perspective that show ~ the door looks like an idiomatic double object construction, with the idiomatic meaning attaching to one member of the pair in (51).

(51)

  • a.

    Mary showed the door to John.

    ‘Mary caused John to see the door.’

  • b.

    Mary showed John the door.

    ‘Mary caused John to see the door.’

    ‘Mary ejected John.’ IDIOMATICMEANING

But in fact there is an alternative analysis. In addition to expressing caused perception, show can express caused (accompanied) motion, with a meaning something like ‘conduct’ or ‘guide’. This meaning is realized in a PP structure with to, where the order of DP and PP object arguments inverts vis-à-vis the caused-perception form. Compare (52a) and (52b).

(52)

  • a.

    Mary showed John to the table.

    ‘Mary conducted/guided John to the table.’

  • b.

    Mary showed the table to John.

    ‘Mary caused John to see the table.’

This suggests a very different view of show ~ the door. Rather than taking it as an idiomatic double object construction, we might take it as an idiomatic caused-motion construction, related to this second sense of show (53a–b), but where to is elided (53c).29

(53)

  • a.

    Mary showed John to the door.

    ‘Mary conducted/guided John to the door.’

  • b.

    Mary showed John the door.

    ‘Mary ejected John.’

  • c.

    Mary [VP showed John TO the door].

This analysis seems more plausible than the previous one. Although the meanings of show in(53a–b) are distinct (‘conduct’ vs. ‘eject’), they are plainly much more closely related to each other than either is related to the caused-perception meaning of dative show (show the door to John). Indeed, it seems clear that the idiomatic meaning of show ~ the door is essentially one of caused motion. On this analysis, then, show ~ the door would be an idiom, and even a surface “two-DP idiom,” but it would not be a double object idiom in the sense relevant to the English dative alternation.30

3.3.4 Read ~ the Riot Act

AHDEL also treats read ~ the riot act (46d) as an idiom. A portion of the AHDEL entry, listed under Riot Act, is shown in (54).

(54)

  • Riot Act: . . . -Idiom: read the riot act. To warn or reprimand energetically or forcefully: The teacher read the riot act to the rowdy class.

  • Word History The riot act has been read to far more people than the disturbers of the peace the Riot Act was intended to control . . .

Read ~ the riot act may indeed be a genuine idiom. If so, however, it is one that is problematic for a wide range of modern analyses. Notice that the sample usage provided in (54) gives the phrase read the riot act to the rowdy class. Similarly, the Word History section for the entry begins with the passive of a PP form (The riot act has been read to far more people . . . ). In both cases, the dictionary authors provide uses of the idiom in its PP dative form, not in its double object form. This implies that read ~ the riot act, unlike the other cases that we have surveyed so far, occurs with its idiomatic meaning in both the PP dative form and the double object form. Therefore, read the riot act is not a double object idiom strictly speaking; rather, it is an alternatingidiom.31 Thus, if this form is truly a double object idiom, it is problematic not only for a classic Dative Shift analysis like (2a), but also for modern nonderivational accounts like those in (11a–b)and (12a) that posit a separate source for double object and PP datives. The former fails to provide appropriate constituency for the idiomatic phrase (read the riot act). The latter fail to account for the retention of idiomatic meaning in the PP form.

In conclusion, there appear to be no clear cases of (exclusively) double object idioms in English, either with give or with other ditransitive verbs. Examples cited in the literature are either (a) compositional ((27a–e), (46a)) and hence not idiomatic, (b) accounted for by pragmatics (46b), (c) irrelevant because they are not double object structures (46c), or alternating and hence not double object idioms specifically (46d).

4 Reexamining “PP Dative Idioms”

Let us now consider putative PP dative idioms: datives of the form V–DP–PP, where V and PP purport to form an idiomatic unit.

4.1 Give N to NP Idioms

In her thoroughgoing study of the English dative alternation, Green (1974) labels examples like(27a–e) simply as nonliteral give constructions, reserving the term dative idiom for cases like(55a–d), which show the form give N to NP.

(55)

  • a.

    Mao’s silence has given rise to an absurd rumor.

    (cf. *Mao’s silence hasgivenan absurd rumorrise.)

  • b.

    Sarah gave birth to a son.

    (cf. *Sarahgavea sonbirth.)

  • c.

    Activism gave way to apathy.

    (cf. *Activismgaveapathyway.)

  • d.

    John gave rein to his feelings.

    (cf. *Johngavehis feelingsrein.)

Although these forms do seem to be idioms, and are treated as such by dictionaries, they are not PP dative idioms of the sort we are concerned with since it is V + N that form the idiomatic unit (give + rise) in (55a–d) and not V + PP.32 In fact, on closer examination it is doubtful that (55a–d) are true synchronic ditransitives at all, in the sense of structures with give and two independent nominal arguments.

Observe that in (55a–d), N seems virtually inseparable from V and largely restricted to bare form. Thus, N cannot be passivized (56a–d), wh-extracted (57a–d), pluralized (58a–d), or easily modified (59a–d).33

(56)

  • a.

    *Rise has been given to an absurd rumor (by Mao’s silence).

  • b.

    *Birth was given to a son (by Sarah).

  • c.

    *Way was given to apathy (by activism).

  • d.

    *Rein was given to his feelings (by John).

(57)

  • a.

    A: What has Mao’s silence given to an absurd rumor?

    B: *Rise.

  • b.

    A: What did Sarah give to a son?

    B: *Birth.

  • c.

    A: What did activism give to apathy?

    B: *Way.

  • d.

    A: What did John give to his feelings?

    B: *Rein.

(58)

  • a.

    *Mao’s silences have given rises to absurd rumors.

  • b.

    *Sarah gave births to a son and a daughter.

  • c.

    *Activism and hope gave ways to apathy and despair.

  • d.

    *John gave reins to feelings of hope and feelings of despair.

(59)

  • a.

    *Mao’s silence gave quick rise to an absurd rumor.

    (cf. Mao’s silence quickly gave rise to an absurd rumor.)

  • b.

    ??Sarah gave recent birth to a son.

    (cf. Sarah recently gave birth to a son.)

  • c.

    *Activism gave immediate way to apathy.

    (cf. Activism immediately gave way to apathy.)

  • d.

    *John gave some/far too much rein to his feelings.

    (cf. John too freely gave rein to his feelings.)

In Larson 1988, I suggest that (55a–d) are not in fact synchronic ditransitives in modern English, but rather complex PP-complement constructions, where give + N forms a compound V (60).Presumably these derive historically from a ditransitive source (e.g., by incorporation of N).34

(60)

graphic

If this proposal is correct, then (55a–d) fail to be ditransitive PP dative idioms on any understanding of the term, except perhaps historical.

4.2 Ditransitive V-PP Idioms

Consider then (7a–d)/(61a–d) (hereafter (61a–d)), offered in Larson 1988:340 as examples of ditransitive PP dative idioms.

(61)

  • a.

    Lasorda sent his starting pitcher to the showers.

  • b.

    Mary took Felix to the cleaners.

  • c.

    Felix threw Oscar to the wolves.

  • d.

    Max carries such behavior to extremes.

Again, standard dictionaries treat these forms as ditransitive V-PP idioms.35 Accepting this judgment, I wish to consider whether they are ditransitive V-PP dative idioms. To do this, we must examine a key point about the semantics of double object vs. PP datives: specifically, whether they are the same or different.

4.2.1 Caused Possession vs. Caused Motion

An often-proposed view of the double object/PP dative distinction in syntax is that it maps to a corresponding distinction in semantics.36 In particular, the double object dative is taken to express caused possession (‘cause to have’) whereas the PP dative form is claimed to express caused motion (‘cause to go’). This view is most persuasive with verbs of transmission and ballistic motion like send and throw. Consider (61a–b), analyzed as in (63a–b), basically following Harley (2002).

(62)

  • a.

    John sent Mary a telegram.

  • b.

    John sent a telegram to Mary.

(63)

graphic

Intuitively, it does seem we can conceive of send as describing two rather different things: (a) a means of bringing something into someone’s possession—by sending it, as opposed to handing or throwing it; or (b) a manner of moving an object from one point to another. Furthermore, this distinction in sense appears to correctly track differences in distribution between the two forms. Thus, the class of possessors is plausibly restricted to animates like persons (only people can get or receive things), whereas the class of motion targets has no similar conceptual restriction. We thus predict the well-known observation that nonanimates are able to occur in PP dative constructions as the object of to, but not in the corresponding double object form.

(64)

  • a.

    Pilar sent a telegram to Lisbon.

  • b.

    *Pilar sent Lisbon a telegram.

Similarly, as Rappaport Hovav and Levin (2008) note, path modifiers like halfway should apply to the PP forms since they describe motion, but not to the double object forms.

(65)

  • a.

    Pilar sent a package halfway to James.

  • b.

    *Pilar sent James a package halfway.

Facts like these suggest a simple syntax-semantics mapping that associates one structure with one meaning.

(66) “One structure – One meaning”

  • a.

    Double object form ⇒ CAUSEDPOSSESSION

  • b.

    PP dative form ⇒ CAUSEDMOTION

4.2.2 PP Datives Also Encode CAUSED POSSESSION

Despite its attractive simplicity, Rappaport Hovav and Levin (2008) argue forcefully against the picture in (66)—specifically, against the unique association of caused possession with double object form. In its place, they propose the more complex mapping in (67) and the corresponding partitioning of verbs in (68).

(67)

  • a.

    Double object form ⇒ CAUSEDPOSSESSION

  • b.

    PP dative form ⇒ CAUSEDMOTIONorCAUSEDPOSSESSION

(68)

  • a.

    Vs expressing only caused possession: give, lend, hand, . . .

  • b.

    Vs expressing only caused motion: drag, carry, convey, . . .

  • c.

    Polysemous Vs expressing both: throw, send, slide, . . .

On this proposal, the double object form is univocal, but the PP dative form is polysemous, with its sense depending on the particular verb that appears. Thus, verbs like give, which are associated strictly with a caused-possession meaning (68a), are predicted to occur in both PP dative and double object forms since, by hypothesis, the latter are compatible with both meanings.

(69)

  • a.

    John gave a book to Mary. CAUSEDPOSSESSION

  • b.

    John gave Mary a book. CAUSEDPOSSESSION

Verbs like drag, which are associated strictly with a caused-motion meaning (68b), are predicted to occur only in PP dative form since the latter is compatible only with this meaning.

(70)

  • a.

    John dragged the chair to Bill. CAUSEDMOTION

  • b.

    *John dragged Bill the chair. *CAUSEDPOSSESSION

Finally, verbs like send, which express both caused motion and caused possession (68c), are predicted to occur in PP dative form with both senses, but in double object form with only the caused-possession meaning.

(71)

  • a.

    John threw a ball to Mary. CAUSEDMOTIONorCAUSEDPOSSESSION

  • b.

    John threw Mary a ball. CAUSEDPOSSESSION

As evidence for their view, Rappaport Hovav and Levin (2008) note that give in both its double object and its PP dative form entails the corresponding possessive; see (72) and (73). (72c) and (73c), for example, are equally contradictory.

(72)

  • a.

    John gave Mary the book.

  • b.

    Mary had/got the book.

  • c.

    #John gave Mary the book, but she never got/received it.

(73)

  • a.

    John gave the book to Mary.

  • b.

    Mary had/got the book.

  • c.

    #John gave the book to Mary, but she never got/received it.

Likewise, although path modifiers of PP are acceptable with Rappaport Hovav and Levin’s caused motion verbs (74a), such modifiers are unacceptable with univocal caused-possession verbs in their PP dative forms (74b), just as they are unacceptable in all double object forms (74c).

(74)

  • a.

    John dragged/carried/slid/sent the chair halfway to Bill.

  • b.

    *John gave/handed/lent the ball halfway to Bill.

  • c.

    *John gave/handed/lent/threw/slid/sent Bill the chair halfway.

In a similar vein, we may note that whereas to-PPs can be questioned with locative/directional where with some verbs expressing caused motion (75a–c), this is never possible with univocal caused-possession verbs (75d).

(75)

  • a.

    John dragged/threw/slid/sent/shipped the chair (over) to Bill.

    Where did John drag/throw/slide/send/ship the chair?

  • b.

    John deposited his money to his checking account.

    Where did John deposit his money?

  • c.

    John lost his money to the stock market.

    Where did John lose his money?

  • d.

    John gave/lent/handed his money to Bill.

    *Where did John give/lend/hand his money?

The latter thus behave like double object forms, which never allow their indirect object to be questioned with where.37

(76)

  • John sent/shipped/mailed Bill the chair.

  • Who/*Where did John send/ship/mail the chair?

Rappaport Hovav and Levin’s (2008) claim that PP dative form is compatible with caused possession seems uncontroversial given the ubiquity of oblique dative possession constructions in the world’s languages. (77) and (78) give examples from Korean and Japanese.

(77)

graphic

(78)

graphic

Rappaport Hovav and Levin’s claim that both location and possession can be expressed by a single verb in oblique frames is also supported by English belong. Belong shows a clear locative meaning in examples like (79a–e), selecting a range of locative PPs. In this use, its sense is something like ‘object or entity X is suitably, customarily, or properly situated at location Y’.

(79)

  • a.

    This chair belongs in the living room/halfway along that wall/there.

  • b.

    A: Where does this chair belong?

    B: It belongs in there/*to Mary.

  • c.

    John belongs in prison.

  • d.

    Memories belong in the past.

  • e.

    This information belongs in the public domain.

At the same time, belong exhibits a distinct possessive meaning in examples like (80a–d), where the dative to is not locative/directional (80b), where the possessed can be material or abstract(80c), and where the notion of possession embraces not only ownership, but the part-whole relations well (80d).

(80)

  • a.

    This chair belongs to Mary.

  • b.

    A: To whom does this chair belong?

    B: It belongs *in there/to Mary.

  • c.

    The book/fault/responsibility/credit belongs to John.

  • d.

    This piece belongs to that puzzle.

In broad respects, then, belong resembles verbs like throw and send under Rappaport Hovav and Levin’s (2008) proposals insofar as it encodes both location and possession senses,38 and in so far as these senses are both projected in an oblique frame.

4.2.3 Ditransitive PP Idioms: PP Datives or PP Locatives?

Rappaport Hovav and Levin’s(2008) results show that the “one structure – one meaning” picture in (66) is too simple and that the oblique class should be further subdivided. (81) offers a revised informal picture of send consonant with their views, in which caused possession is associated with both double object and PP dative forms, the former being represented with underlying HAVE and the latter being represented with underlying BELONG. The latter should not be thought of as equivalent to the surface English verb belong, which (as a possessive) is closer in meaning to English possess/own than to have; rather, it should be thought of like the copula that occurs in the Korean and Japanese examples in (77) and (78).

(81)

graphic
graphic

This revised picture has important consequences for our inquiry into PP dative idioms since we must now identify which class of oblique constructions we are interested in: oblique caused possession constructions, oblique caused-motion constructions, or both?

On reflection, it seems clear that candidate oblique idioms relevant to the dative alternation will need to be caused possessives, for if there is any question of syntactic relatedness, it is surely between the double object construction and the oblique caused-possession construction, whose forms appear to express the same sense and involve the same thematic relations. Putting matters differently, even if we are convinced that double object and oblique caused-motion forms are derivationally independent—that neither derives from the other syntactically—this would still leave another class of oblique forms to which double object forms might be related: oblique caused possessives. Therefore, candidate oblique idioms relevant to the question of derivational relatedness/independence need to be caused possessives.

In this light, let us reconsider (61a–d). Accepting them as idioms, are they oblique caused possession idioms? The answer seems to be no. As noted earlier, possessors/recipients must be animate, but the showers, the cleaners, and extremes in (61a,b,d) are inanimate, thus not potential recipients. In (61c), the wolves is animate and hence a potential recipient. But it is equally plausible that the wolves is a destination in (61c), a counterpart to the bus in the caused-motion idiom throw under the bus, which expresses virtually the same sense (‘sacrifice’).39 If this is correct, then (61a–d) are all caused-motion idioms and hence simply don’t bear on the question of whether(1a–b) (repeated here in (82)) are derivationally related or independently projected.40

(82)

  • a.

    Mary gave a present to John. PP dative construction

  • b.

    Mary gave John a present. Double object construction

In conclusion, then, we have failed to identify any ditransitive dative idioms of the form V–DP–PP, where V and PP form an idiomatic unit. Idioms like (55a–d) with possessive give are not of the relevant form and arguably not ditransitives at all. Idioms like (61a–d), noted in Larson 1988, are not caused possessives; rather, they are caused-motion constructions and hence irrelevant to the question of syntactic relatedness in the dative alternation.

5 Conclusion

Dative idioms have been claimed to exist in both double object and PP dative form. These claims have been adduced in support of nonderivational accounts of the dative alternation, in which both forms are independently projected. In this article, I have argued that the basic existence claim is not well-founded—that, with perhaps a single exception, there are no dative idioms of either form, and none that fail to alternate. Putative double object idioms are in fact not idioms but collocations, a status that is irrelevant to syntactic projection. Putative PP dative idioms either aren't datives (rather, they are caused-motion constructions) or aren’t ditransitives. Thus, there appears to be no secure argument from this data source against derivational accounts of the dative alternation in English.

Notes

For comments and discussion, I am grateful to audiences at the Center for Research in Syntax, Semantics and Phonology (CRISSP; KU Leuven, Brussels) and at the One-Day Workshop on Syntax and Semantics (Fuji Women's University, Sapporo) where earlier versions of this material were presented. I am particularly indebted to an anonymous LI reviewer whose insightful suggestions and criticisms prompted substantial changes to this article.

1 This analysis was subsequently explored by Fillmore (1965), Jackendoff and Culicover (1971), Emonds (1972),Fischer (1972), and Green (1974), among many others. It was also the standard analysis in early versions of Relational Grammar. See the contributions in Perlmutter 1983 and Perlmutter and Rosen 1984.

2 Examples like (7a–d) were first discussed (to my knowledge) by Emonds (1972).

3 See Bruening 2010a,b for more recent elaboration.

4Marantz (1993) introduces X in order to locate the theme (a present in (12a)) in specifier position. An anonymous LI reviewer points out that this raises questions of the extent to which (12a) really is compatible with claimed double object idioms insofar as it would seem to require X to be part of the idiom content.

5 I am particularly indebted to Guido Vanden Wyngaerd for discussion of issues in this section.

6 The standard view is not without its critics. For example, Gazdar et al. (1985) argue for full compositionality in idioms, with distributional restrictions being an entirely semantic matter. This theory does not (so far as I can tell) entail underlying syntactic constituency for idioms. See also Nunberg, Sag, and Wasow 1994. See Nicolas 1995 for a defense of the traditional position and a careful discussion of apparent idiom modification. See footnote 13 for more on this point.

7 More proposals have been made that sort collocations into different types. See Schenk 1995.

8 As a historical note, this same class of items was the basis of the TV game show Concentration (1985–1991),which predated Wheel of Fortune. Contestants on Concentration were required to guess the identity of various collocations, the latter being concealed in rebus form.

9 Whether the Wheel of Fortune corpus should be included in the lexicon proper is an open question, as linguists understand the lexicon—that is, as part of the human linguistic computational system. Inclusion is not standard dictionary practice and although this might be argued to be an issue of convenience or lexicographic tradition, I suspect there are reasons of principle as well. Humans appear to have knowledge of object cooccurrences of all sorts; consider, for example, “food cooccurrences” like ham and eggs, liver and onions, and cheese and crackers, or “building material cooccurrences” like bricks and mortar or hammer and nails. If object cooccurrences can be stored generally, it’s not clear why cooccurrences where the objects happen to be linguistic ones (words and phrases) shouldn’t fall within this broader domain, and hence outside the lexicon proper. It’s interesting that object cooccurrences like ham and eggs, liver and onions, or bricks and mortar are matched by collocational status in the phrases for them: ham and eggs, liver and onions, bricks and mortar, and so on. Since object collocation (cheese and crackers) seems a nonlinguistic matter—that is, not a matter of specifically linguistic knowledge—it’s plausible that the corresponding word collocation (cheese and crackers) is nonlinguistic as well.

10Jackendoff (1997:155) explicitly identifies idioms as a subclass of Wheel of Fortune items.

11 Note that bosom can be conjoined with an unambiguous adjective in this context (a bosom and trusted friend, etc.).

12Dead takes on a related meaning with close-scale predicates expressing geometric/spatial properties or relations (dead straight, dead (in the) center, dead square, dead left, dead right). Here, only the sense of extremity is present.

13 See Di Sciullo and Williams 1987 and Van Gestel 1995 for arguments in support of this position and Jackendoff 1997 for a dissenting view. Commitment to the principle of “idioms as constituents” depends to a large degree on broader theoretical assumptions. Jackendoff (1997) deploys a level of conceptual representation wherein the unity of idioms is represented; this level obviates the need for phrasal constituency in the syntax proper. Comparison between this theory and more standard ones without conceptual structure is not straightforward.

15 Two observations are relevant here. First, the distinction made here between idiom and collocation appears close to that drawn by Nunberg, Sag, and Wasow (1994:508) between “idiomatic phrasal constructions” and “idiomatically combining expressions.” The former is an I-language concept in the sense of being about how a speaker represents the relevant form; the latter is an E-language concept insofar as it is about regularities in language use by a population. (See Kuiper and Everaert 2000 for discussion on this point.) As Nunberg, Sag, and Wasow discuss at length, there is wide variation in the degree to which idiomatic phrasal constructions/collocations show special meanings. Nunberg, Sag, and Wasow express general skepticism about the connection between idiomaticity and constituency; however, it is unclear (to me) whether their skepticism extends to both “idiomatic phrasal constructions” and “idiomatically combining expressions” or to the latter alone. If syntactic constituency is not a necessary property even of “idiomatic phrasal constructions,” then the use of dative idioms to argue for constituency is forfeit on more general grounds.

Second, the existence of nonconstituent book titles like (ia–d), noted by Pullum (2008), is also evidence against the view that only multiword sequences corresponding to constituents can be remembered.

  • (i)

    • a.

      If On a Winter’s Night a Traveller (Se una notte d’inverno un viaggiatore, Italo Calvino)

    • b.

      The Fire Next Time (James Baldwin)

    • c.

      Sometimes a Great Notion (Ken Kesey)

    • d.

      Dancer from the Dance (Andrew Holleran)

16 See Tsujioka 2011 for further critique of Kishimoto 2008.

17 The demonstration below is presented at an informal level for clarity and simplicity. There would be no difficulty (so far as I can see) with executing it in a more technical semantic framework such as that of Larson and Segal (1995) or Heim and Kratzer (1997).

18 An anonymous LI reviewer notes that (35a–e) are all offered under the entries for the respective nouns, not under the verb give, implicitly supporting the view that the source of figurative meaning in (27a–e) is uniformly the noun alone and not the give + N combination.

19Nicolas (1995), following Arnold and Sadler (1988) and Verhagen (1990), argues that apparent internal modification of idioms as in (ia) is strictly syntactic insofar as the adjective close semantically modifies the whole idiom keep tabs, not the subpart tabs, to which it is apparently adjoined; compare (ib).

Nicolas argues that this is in fact the general pattern for idioms and hence that arguments for semantic compositionality of idioms based on such examples are not well-founded.

20 See Gazdar et al. 1985 for similar points.

21 In the case of the boot, part of the restriction on use with the relevant meaning may be due to the fact that speakers recognize this construction as an Anglicism, comparable to give the sack and the chop. Potentially, American English speakers limit use of the boot with the relevant meaning to that domain. This point gains credibility from the fact that for some British English speakers at least, the equivalents of (43b–c) with the sack are largely acceptable.

  • (i)

    • a.

      ?The sack has ruined many an employee’s Christmas.

    • b.

      ?With government workers, one common legacy of the sack is lasting psychological problems.

22 For related observations, see Dowty 1979:129n4 and Levin and Rappaport Hovav 1995:96. Note that the difficulty in (45b–d) is not obviously semantic insofar as periphrastic forms of similar meaning involving the simple predicate hard appear to be fully acceptable (The problem became harder, John made the problem harder, We were surprised at how hard the problem was, etc.).

23 The adjective hard also shows normal comparison class relativity, as it does in the simple copular predication That problem is hard; that is, hard is understood in the sense ‘hard for a problem’.

24 Confusion on these points seems widespread in linguistics, with significant consequences for analysis. Thus, Brame (1976) analyzes make headway as a VP idiom and cites this status as evidence for a head-raising analysis of relative clauses; see (ia–b). This argument has been much repeated.

  • (i)

    graphic

However, make in (ia) surely possesses the same meaning it does in its synonymous, fully compositional counterpart make progress. Furthermore, dictionaries list headway as having an independent sense identical to progress; see (iia–b). Make headway is thus compositional/nonidiomatic.

  • (ii)

    • a.

      headway n. 2 Progress toward a goal (AHDEL)

    • b.

      headway n. 1b ADVANCE, PROGRESS (WSNCD)

The difference between make progress and make headway thus seems to be that the latter is a collocation—fully compositional but distributionally linked—whereas the former is not.

Whether collocational status should be represented in grammar is an interesting question. The acceptability of both members of the pair in (iiia–b), in addition to providing evidence against the idea that make headway must be represented as an initial constituent in underived structure, suggests that collocations involve a relation that must be checked at some stage in the derivation—either initially or later—for well-formedness.

  • (iii)

    graphic

This suggests some form of local feature checking or agreement. Unfortunately I cannot pursue this idea further here.

25 I am indebted to Beth Levin for discussion of the issues in this section.

26 An anonymous LI reviewer asks whether the meaning of the ropes might have been “backformed” or “backinferred”from an idiomatic show the ropes. It seems to me this question invokes a much broader, prior one regarding idiom acquisition in general. Idioms plainly do not come labeled as such. Furthermore, the base assumption of the learner must surely be that all constructions are compositional, and hence that backforming and back-inferring from phrase meaning to separate word meanings is the general acquisition strategy. This strategy will presumably include various distributional posits for items, ranging from highly restricted domains like those involved with the ropes or rancid(semantic) or the sack (ethnic) to free availability within category. Plausibly, in this picture, learners decide that something is an idiom when they are unable to assign separate word meanings in any coherent distributional domain wider than the phrase itself. The point is that even if the possibility of know the ropes somehow arose by backformation from show the ropes, it’s far from clear this isn’t the trajectory of lexical acquisition generally.

27 See Austin 1972 and Searle 1970 for classic exposition of commissive speech acts.

28 In fact, promise ~ the moon is plausibly a case of what Morgan (1978) calls a “short-circuited implicature,” that is, an implicature derived according to the Cooperative Principle but where the intended effect can be immediately recognized because of conventions of usage. See Morgan 1978 and Green 1996 for fuller discussion.

In a similar vein, see Grant and Bauer 2004, where pragmatic reasoning is applied to derive the purported idiomatic content of kill two birds with one stone.

29 The variety of English spoken in Lancashire permits “alternative double object constructions” like (ia) in addition to the more familiar PP dative construction (ib) (see Gast 2007, Siewierska and Hollmann 2007, Haddican 2010, and Myler 2013 for discussion). Nye (2010) argues persuasively that alternative dative object constructions are derived from the PP datives by deletion of to (ic).

  • (i)

    • a.

      Sarah lent the book the man.

    • b.

      Sarah lent the book to the man.

    • c.

      Sarah lent the book TO the man.

Nye further observes (pers. comm.) that Lancashire dialect shows phenomena similar to that discussed in the text. Thus, the verb take exhibits the for-dative alternation (iia–b). From this perspective, (iic) looks like an idiomatic double object structure.

  • (ii)

    • a.

      Sarah took a bone for the dog.

    • b.

      Sarah took the dog a bone.

    • c.

      Sarah took the dog a walk.

      ‘Sarah took the dog for a walk.’

      (cf. Sarah took a walk for the dog.)

But, as Nye observes, (iic) plainly seems to be a variant of (iiia), formed by for-deletion, even though this process is not productive in her Lancashire dialect; compare (iiib–c).

  • (iii)

    • a.

      Sarah took the dog for a walk.

    • b.

      Sarah took the horse *(for) a ride.

    • c.

      Sarah took the car *(for) a spin.

See also Myler 2013 for a similar proposal regarding null TO in alternative dative object constructions and for additional interesting empirical observations about its content.

30 An anonymous LI reviewer notes that the proposed analysis of show ~ the door as a caused-motion construction will extend to the related show ~ home (as in Mary showed John home), which also involves apparent elision of to. See Myler 2013 for discussion.

31Rappaport Hovav and Levin (2008) cite extensive corpus evidence indicating that many “double object idioms” claimed not to have PP dative alternates actually do so. Bruening (2010a) makes the interesting claim that alternating dative idioms in their to-PP form are not PP datives but rather “disguised” double object forms. This claim is based on “scope-freezing” judgments similar to those noted by David Lebeaux with double object constructions (analyzed by Bruening (2001) in terms of Superiority). Bruening’s scope judgments are, for me, considerably less robust than those noted by Lebeaux. Therefore, given the serious adjustments to current theory that Bruening’s (2010a) analysis requires in terms of rightward specifiers + rightward movement, possible ternary branching, and insertion of to under (for me)rather unclear circumstances, and given the many simple distributional questions that arise for it, such as the placement of adverbs in cases like (i), I am unable to evaluate Bruening’s proposal.

  • (i)

    Boris gave the creeps in those days to everyone who came in contact with him.

My own view of cases like (i) is that they are in fact PP datives and that constraints on no alternating datives are fundamentally pragmatic matters dealing with information structure. See Larson 2014 for some discussion.

32 See Green 1974:224 for an extensive list of such dative idioms.

33 In this respect, give rise/give birth idioms diverge from the familiar pay attention/keep tabs sort, which do allow modification (ia–b) and are susceptible to A-movement (iia–b).

  • (i)

    • a.

      John paid some/a great deal of/far too much attention to the problems.

    • b.

      John kept close/very close/very few tabs on Mary’s activities.

  • (ii)

    • a.

      Some/A great deal of/Far too much attention was paid to the problems.

    • b.

      Close/Very close/Very few tabs were kept on Mary’s activities.

This behavior suggests that the nominal expressions in the latter class are indeed independent elements in argument position, in contrast with those in the dative class. An anonymous LI reviewer notes that the presence of inanimate subjects with give rise and give way is consistent with the idea that these involve a special use of give, which otherwise requires an agent.

34 This analysis seems superior to Richards’s (2001) suggestion that forms like (55a–d) simply correspond to underlying caused-locative idioms in the sense of Harley 1995, 2002. There is no clear sense in which rise goes to a rumor location, rein goes to feelings, and so on. Indeed, these forms barely seem to embed a recognizable ‘give’ meaning atoll.

Note also that, as an anonymous LI reviewer points out, many of the properties cited above for give-N-to idioms do appear to hold of N-incorporation structures and of Ns in light-noun combinations.

35WSNCD lists both throw ~ to the wolves (‘to leave unprotected against fierce opposition or attack’) and take ~ to the cleaners (‘to deprive of money or possessions, clean out’) as VP idioms under their respective verbs. AHDEL lists the related form put ~ through the wringer (‘to subject to a severe trial or ordeal’) under wringer as “slang.”

36 Superficially at least, double object constructions with show do not encode caused possession; rather, they encode caused perception (‘cause to see’). Perhaps the latter can be accommodated under some broader notion of “having a perception.” I ignore this complication for simplicity of exposition.

37 Elsewhere (Larson 2014), I have noted that Rappaport Hovav and Levin’s (2008) partition of verbs into caused possession/caused-motion types is paralleled in some languages (e.g., European Portuguese, Serbian) by a corresponding partition of prepositions. In such languages, univocal caused-motion verbs select a directional preposition that is formally different from the dative that appears with caused-motion verbs. See Levinson 2005 for related points.

38Belong is not the only English verb showing an oblique locative/possessive alternation of this kind. The verb go is analogous in pairs like (ia–b).

  • (i)

    • a.

      John goes/went to that grocery store.

    • b.

      This piece goes/went to that puzzle.

The first use of go is plainly motional/locative, but the second use is possessive, with a meaning very similar to that of (80d) with belong. Counterparts of the English possessive use of go are regularly found in other languages, for example, Spanish.

39O’Grady (1998) also notes that idioms like throw to the wolves do not involve possession between the two objects; wolves do not in any sense get or have the person thrown to them. Hence, there is no sense in which such idioms are possessive constructions. Rappaport Hovav and Levin (2008:154–155) make a similar point in relation to send to the devil, where even though both internal arguments are animate, no possessive relation exists between them. Hence, throw to the wolves and send to the devil are surely caused-motion constructions and not caused-goal constructions.

40 This is not to say that such idioms are irrelevant to syntactic theorizing in general. For example, caused-motion idioms like those in (61a–d) seem to me problematic for PP small clause analyses of caused-motion constructions of the sort proposed by Harley (2002) and Bruening (2010a,b), where V and PP do not form a constituent that excludes the direct object. Consider the caused-motion construction with drag in (ia) under the Larson/Chomsky analysis, and the idiomatic construction drag ~ over the coals. (ib) represents the idiom as a constituent.

  • (i)

    graphic

By contrast, to accommodate this sort of example along the lines indicated in (6b) would seem to require a decompositional analysis of drag that factors it into CAUSE and an appropriate P, locating idiomaticity in P′.

  • (ii)

    graphic

What P is this? The answer is quite unclear, to me at any rate.

References

Arnold,
Douglas
, and
Louise
Sadler
.
1988
. Noncompositionality and translation. In
Recent developments and applications of natural language processing
, ed. by
Jeremy
Peckham
,
23
55
.
London
:
Kogan Page
.
Austin,
John
.
1972
.
How to do things with words
.
Cambridge, MA
:
Harvard University Press
.
Bannard,
Colin
, and
Elena
Lieven
.
2012
.
Formulaic language in L1 acquisition
.
Annual Review of Applied Linguistics
32
:
3
16
.
Bannard,
Colin
, and
Danielle
Matthews
.
2008
.
Stored word sequences in language learning: The effects of familiarity on children’s repetition of four-word combinations
.
Psychological Science
19
:
241
248
.
Biber,
Douglas
, and
Susan
Conrad
.
1999
. Lexical bundles in conversation and academic prose. In
Out of corpora: Studies in honour of Stig Johansson
, ed. by
Hilde
Hasselgård
and
Signe
Oksefjell
,
181
190
.
Amsterdam
:
Rodopi
.
Biber,
Douglas
,
Susan
Conrad
, and
Viviana
Cortes
.
2004
.
If you look at . . .: Lexical bundles in university teaching and textbooks
.
Applied Linguistics
25
:
371
405
.
Bowers,
John
.
1981
.
The theory of grammatical relations
.
Ithaca, NY
:
Cornell University Press
.
Brame,
Michael K
.
1976
.
Conjectures and refutations in syntax and semantics
.
Amsterdam
:
North Holland
.
Bruening,
Benjamin
.
2001
.
QR obeys Superiority: Frozen scope and ACD
.
Linguistic Inquiry
32
:
233
273
.
Bruening,
Benjamin
.
2010a
.
Double object constructions as disguised prepositional datives
.
Linguistic Inquiry
41
:
287
305
.
Bruening,
Benjamin
.
2010b
.
Ditransitive asymmetries and a theory of idiom formation
.
Linguistic Inquiry
41
:
519
562
.
Chen,
Yu-Hua
, and
Paul
Baker
.
2010
.
Lexical bundles in L1 and L2 academic writing
.
Language Learning and Technology
14
(
2
):
30
49
.
Chomsky,
Noam
.
1955
.
The logical structure of linguistic theory
.
New York
:
Plenum
.
Chomsky,
Noam
.
1993
. A minimalist program for linguistic theory. In
The view from Building 20: Essays in linguistics in honor of Sylvain Bromberger
, ed. by
Kenneth
Hale
and
Samuel Jay
Keyser
,
1
52
.
Cambridge, MA
:
MIT Press
.
Chomsky,
Noam
.
1995
.
The Minimalist Program
.
Cambridge, MA
:
MIT Press
.
Columbus,
Georgie
.
2013
.
In support of multiword unit classifications: Corpus and human rating data validate phraseological classifications of three different multiword unit types
.
Yearbook of Phraseology
4
:
23
43
.
Conklin,
Kathy
, and
Norbert
Schmitt
.
2012
.
The processing of formulaic language
.
Annual Review of Applied Linguistics
32
:
45
61
.
Di Sciullo,
Anna-Maria
, and
Edwin
Williams
.
1987
.
On the notion of word
.
Cambridge, MA
:
MIT Press
.
Dowty,
David
.
1979
.
Word meaning and Montague Grammar
.
Dordrecht
:
Kluwer
.
Dryer,
Matthew
.
1987
.
On primary objects, secondary objects, and antidative
.
Language
62
:
808
845
.
Emonds,
Joseph
.
1972
.
Evidence that indirect object movement is a structure-preserving rule
.
Foundations of Language
8
:
546
561
.
Fillmore,
Charles
.
1965
.
Indirect object constructions in English and the ordering of transformations
.
The Hague
:
Mouton
.
Fischer,
Susan
.
1972
.
The acquisition of verb-particle and dative constructions
.
Doctoral dissertation, MIT, Cambridge, MA
.
Gast,
Volker
.
2007
.
I gave it him. On the motivation of the “alternative double object construction” in varieties of British English
.
Functions of Language
14
:
31
56
.
Gazdar,
Gerald
,
Ewan
Klein
,
Geoffrey K.
Pullum
, and
Ivan A.
Sag
.
1985
.
Generalized Phrase Structure Grammar
.
Cambridge, MA
:
Harvard University Press
.
Gibbs,
Raymond
, and
Gayle
Gonzalez
.
1985
.
Syntactic frozenness in processing and remembering idioms
.
Cognition
20
:
243
259
.
Grant,
Lynn
, and
Laurie
Bauer
.
2004
.
Criteria for re-defining idioms: Are we barking up the wrong tree?
Applied Linguistics
25
:
38
61
.
Green,
Georgia
.
1974
.
Semantics and syntactic irregularity
.
Bloomington
:
Indiana University Press
.
Green,
Georgia
.
1996
.
Pragmatics and natural language understanding
.
Hillsdale, NJ
:
Lawrence Erlbaum
.
Haddican,
William
.
2010
.
Theme-goal ditransitives and theme passivisation in British English dialects
.
Lingua
120
:
2424
2443
.
Hallman,
Peter
.
2015
.
Syntactic neutralization in double object constructions
.
Linguistic Inquiry
46
:
389
424
.
Harley,
Heidi
.
1995
.
Subjects, events and licensing
.
Doctoral dissertation, MIT, Cambridge, MA
.
Harley,
Heidi
.
2002
. Possession and the double-object construction. In
Linguistic variation yearbook 2
, ed. by
Pierre
Pica
and
Johann
Rooryck
,
31
70
.
Amsterdam
:
John Benjamins
.
Heim,
Irene
, and
Angelika
Kratzer
.
1997
.
Semantics in generative grammar
.
Malden, MA
:
Blackwell
.
Jackendoff,
Ray
.
1997
.
The architecture of the language faculty
.
Cambridge, MA
:
MIT Press
.
Jackendoff,
Ray
, and
Peter
Culicover
.
1971
.
A reconsideration of dative movement
.
Foundations of Language
7
:
397
412
.
Kishimoto,
Hideki
.
2008
.
Ditransitive idioms and argument structure
.
Journal of East Asian Linguistics
17
:
141
179
.
Kitagawa,
Yoshihisa
.
1994
. Shells, yolks and scrambled e.g.s. In
NELS 24
, ed. by
Mercè
Gonzàlez
,
221
239
.
Amherst
:
University of Massachusetts, Graduate Linguistic Student Association
.
Kuiper,
Koenraad
, and
Martin
Everaert
.
2000
.
Phrase structure constraints of (English) phrasal lexical items
.
Ms., University of Canterbury and Utrecht University
.
Lakoff,
George
.
1970
.
Irregularity in syntax
.
New York
:
Holt, Rinehart and Winston
.
Larson,
Richard K
.
1988
.
On the double object construction
.
Linguistic Inquiry
19
:
335
391
.
Larson,
Richard K
.
1990
.
Double objects revisited: Reply to Jackendoff
.
Linguistic Inquiry
21
:
589
632
.
Larson,
Richard K
.
2014
.
On shell structure
.
London
:
Routledge
.
Larson,
Richard K.
, and
Gabriel
Segal
.
1995
.
Knowledge of meaning
.
Cambridge, MA
:
MIT Press
.
Levin,
Beth
, and
Malka Rappaport
Hovav
.
1995
.
Unaccusativity: At the syntax–lexical semantics interface
.
Cambridge, MA
:
MIT Press
.
Levinson,
Lisa
.
2005
. To in two places and the dative alternation. In
Proceedings of the 27th Annual Penn Linguistics Colloquium
, ed. by
Sudha
Arunachalam
,
Tatjana
Scheffler
,
Sandhya
Sundaresan
, and
Joshua
Tauberer
,
155
168
.
Philadelphia
:
University of Pennsylvania, Penn Working Papers in Linguistics
.
Marantz,
Alec
.
1993
. Implications of asymmetries in double object constructions. In
Theoretical aspects of Bantu grammar
, ed. by
Sam
Mchombo
,
113
150
.
Stanford, CA
:
CSLI Publications
.
Morgan,
Jerry
.
1978
. Two types of convention in indirect speech acts. In
Syntax and semantics 9: Pragmatics
, ed. by
Peter
Cole
,
261
280
.
New York
:
Academic Press
.
Myler,
Neil
.
2013
.
On coming the pub in the North West of England: Accusative unaccusatives, dependent case, and preposition incorporation
.
Journal of Comparative Germanic Linguistics
16
:
189
207
.
Nicolas,
Tim
.
1995
. Semantics of idiom modification. In
Idioms: Structural and psychological perspectives
, ed. by
Martin
Everaert
and
Erik-Jan
van der Linden
,
233
252
.
Hillsdale, NJ
:
Lawrence Erlbaum
.
Nunberg,
Geoffrey
,
Ivan
Sag
, and
Thomas
Wasow
.
1994
.
Idioms
.
Language
70
:
491
538
.
Nye,
Rachel
.
2010
.
She showed the data her professor: The alternative double object construction in Lancashire English
.
Ms., Ghent University
.
Oehrle,
Richard
.
1976
.
The grammatical status of the English dative alternation
.
Doctoral dissertation, MIT, Cambridge, MA
.
O’Grady,
William
.
1998
.
The syntax of idioms
.
Natural Language and Linguistic Theory
16
:
279
312
.
Perlmutter,
David M.
, ed.
1983
.
Studies in Relational Grammar 1
.
Chicago
:
University of Chicago Press
.
Perlmutter,
David M.
, and
Carol
Rosen
, eds.
1984
.
Studies in Relational Grammar 2
.
Chicago
:
University of Chicago Press
.
Pullum,
Geoffrey K
.
2008
.
Non-constituent titles
.
Language Log
. Available at http://itre.cis.upenn.edu/~myl/languagelog/archives/005333.html.
Radford,
Andrew
.
1997
.
Syntax: A Minimalist introduction
.
New York
:
Cambridge University Press
.
Rappaport Hovav,
Malka
, and
Beth
Levin
.
2008
.
The English dative alternation: The case for verb sensitivity
.
Journal of Linguistics
44
:
129
167
.
Richards,
Norvin
.
2001
.
An idiomatic argument for lexical decomposition
.
Linguistic Inquiry
32
:
183
192
.
Schenk,
André
.
1995
. The syntactic behavior of idioms. In
Idioms: Structural and psychological perspectives
, ed. by
Martin
Everaert
and
Erik-Jan
van der Linden
,
253
282
.
Hillsdale, NJ
:
Lawrence Erlbaum
.
Searle,
John
.
1970
.
Speech acts
.
Cambridge
:
Cambridge University Press
.
Siewierska,
Anna
, and
Willem
Hollmann
.
2007
. Ditransitive clauses in English with special reference to Lancashire dialect. In
Structural-functional studies in English grammar
, ed. by
Mike
Hannay
and
Gerard J.
Steen
,
83
102
.
Amsterdam
:
John Benjamins
.
Siyanova-Chanturia,
Anna
,
Kathy
Conklin
, and
Norbert
Schmitt
.
2011
.
Adding more fuel to the fire: An eye-tracking study of idiom processing by native and non-native speakers
.
Second Language Research
27
:
1
22
.
Siyanova-Chanturia,
Anna
,
Kathy
Conklin
, and
Walter J. B.
van Heuven
.
2011
.
Seeing a phrase “time and again” matters: The role of phrasal frequency in the processing of multiword sequences
.
Journal of Experimental Psychology: Learning, Memory and Cognition
37
:
776
784
.
Swinney,
David
, and
Anne
Cutler
.
1979
.
The access and processing of idiomatic expressions
.
Journal of Verbal Learning and Behavior
18
:
523
534
.
Tsujioka,
Takae
.
2011
.
Idioms, mixed marking in nominalization and the base-generation hypothesis for ditransitives in Japanese
.
Journal of East Asian Linguistics
20
:
117
143
.
Van Gestel,
Frank
.
1995
. En bloc insertion. In
Idioms: Structural and psychological perspectives
, ed. by
Martin
Everaert
and
Erik-Jan
van der Linden
,
75
96
.
Hillsdale, NJ
:
Lawrence Erlbaum
.
Van Lancker,
Diana
,
Gerald
Canter
, and
Dale
Terbeek
.
1981
.
Disambiguation of ditropic sentences: Acoustic and phonetic cues
.
Journal of Speech and Hearing Research
24
:
330
335
.
Verhagen,
Marc
.
1990
.
Support verbs in Disneyland
.
Eurotra-Essex Internal Memoranda No. 20. Essex University, Wivenhoe
.
Wood,
David
.
2015
.
Fundamentals of formulaic language
.
London
:
Bloomsbury
.