Multiple Modals in Modern English: Use, History, and Structure of Periphrastic Modal Verbs more

Heather Marie Kosur Wednesday 2 December 2009 Multiple Modals in Modern English: Use, History, and Structure of Periphrastic Modal Verbs Modals and Modality Characteristic of Germanic languages including English, modal verbs are auxiliary verbs that express modality. Modality is loosely defined as the grammaticalized expression of the subjective attitudes and opinions of the speaker including possibility, probability, necessity, obligation, permissibility, ability, desire, and contingency (Bybee et al. 1994:176-181). More specifically, modality is divided into two distinct but related categories: epistemic modality and deontic modality (Palmer 1990:5). Epistemic modality is defined as the expression of the judgment of possibility or necessity of a proposition by a speaker as in That must be him at the door and How much would it cost (Palmer 1990:50-58). Deontic modality is defined as the expression of the judgment of possibility including permissibility or necessity including obligation of an action, state, or event by a speaker as in You can watch the movie now and You will wash the dishes later (Palmer 1990:50,69-74). The fundamental difference between epistemic modality and deontic modality is the judgment on propositions, or knowledge, versus the judgment on actions, states, and events, or performances (Palmer 1990:6, 69). Modality in Modern English is expressed through grammatical mood (indicative, imperative, subjunctive) and modal verbs. English has nine full modals verbs as a well as four quasi-modal verbs, which possess some but not all properties of prototypical modals. The nine English modals are can, could, may, might, must, shall, should, will, and would. The four quasi-modals in English include ought to, used to, would rather, and had better (had best) (Palmer 1990: 5; Palmer 2003:12).1 The verbs need and dare also resemble Kosur 2 modal verbs within negated, but not positive, verb phrases as in He need not whine and She dare not complain (Huddleston 1984:165). Any one modal or quasi-modal verb may express both epistemic and deontic modality depending on the context of the utterance. Modal verbs including quasi-modal verbs differ from prototypical verbs in that modal verbs are both defective and neutral verbs, defectiveness referring to the lack of non-tensed forms and neutralization referring to the lack of separate third person singular simple present forms (Huddleston 1984:126).2 While prototypical English verbs have at least four but up to six separate finite and nonfinite forms depending on the regularity or irregularity of the verb, modals have only base forms as illustrated in (1). Modal verbs lack infinitive, simple present, third person singular simple present, simple past, present participle, and past participle forms. Also unlike prototypical English verbs, modals verbs perform only the grammatical function of modal only within verb phrases performing the grammatical function of predicate (Huddleston 1984:128-131; Hopper 1999:112) as illustrated in (2). Prototypical verbs perform four verbal functions, one adjectival function, and six nominal functions— verb phrase head, predicate, verb phrase complement, adjective phrase complement, noun phrase modifier, subject, subject complement, direct object, object complement, indirect Kosur 3 object, and prepositional complement—while modal verbs perform only the auxiliary function of modal. Modal verbs also appear only in the initial position of the verb phrase as in The room may have been being cleaned but not *The train have will arrived. Multiple Modals Multiple modals (MMs), which are more specifically called double modals and triple modals, are a periphrastic verb construction characterized by the use of two or more modal and/or quasi-modal verbs within a single verb phrase as in might could and used to could. 3 Nagle (2003:350) further theorizes that the "core system of double modals consists of a sequence of two modal auxiliaries, usually arrayed in tiers" with the first tier consisting of the epistemic may, might, and must and the second tier consisting of the deontic can, could, should, will, and would. Although a combination of the nine English modals plus the four English quasi-modals should result in a minimum of 156 possible multiple modals (13*12=156 double modals; 13*12*11=1,716 triple modals), Di Paolo Kosur 4 (1989:197) and Mishoe and Montgomery (1994:3) identify a total of only thirty naturally occurring multiple modal combinations. As illustrated in (3), roughly sixty-three percent (nineteen of the thirty) of the multiple modals have may or might as the initial modal (Mishoe and Montgomery 1994:10; Nagle 2003:350). As Mishoe and Montgomery (1994:10) also discovered from comparisons of their corpus data to elicitation studies by Butters (1973), Coleman (1975), and Boertien (1986), the three core multiple modals identified as most acceptable by all multiple modal speakers are might could, might can, and might would with used to would as the fourth most common combination (Mishoe and Montgomery 1994:3,10,20). Multiple modals allow for intervening adverbs functioning as verb phrase modifiers including negatives (Nagel 2003:350) as in two examples identified by Di Paolo (1989:216): "If we had known ( ), we may still could have done it." and "I don't hear too well ( ) I thought maybe I better put it [her hearing aid] on (or) I might not could understand you, so [rising intonation]." The actual occurrence of may or might as the most frequent initial modal within multiple modal constructions supports the tiered core multiple modal system in which the Kosur 5 first verb is an epistemic modal and the second verb is a deontic modal (Nagle 2003:351; Fennell 1993:434). In fact, as Fennell (1993:434) concludes in an explanation of the loss of the double epistemic modal shall may from Middle English (Ogura 1993:539) to Modern English, only combinations of epistemic modals followed by deontic modals and never pairs of epistemic or even deontic modals are possible in any variety of Modern English with multiple modals. Similar to the constraint on the position of single modal verbs, epistemic modal verbs always appear first followed by one or more deontic modal verbs in multiple modal constructions. And, just as single modals appear only in the initial position of the verb phrase, multiple modals also appear only in the initial position as in You might could have vacuumed the house but not *You have might could vacuumed the house The ordered combination of epistemic and deontic modals is essential to understanding the use and pragmatics of multiple modals. Users Although multiple modals are proscribed in prescriptive Standard English grammars and stigmatized by some speakers of non-multiple modal Englishes, native speakers of many varieties of English regularly use at least one multiple modal occasionally (Di Paolo 1989:196). Multiple modal constructions are a feature of Southern American Englishes, African American Englishes, Northern British Englishes, Scottish Englishes, Northern Irish Englishes, and Caribbean Englishes (Herndobler and Sledd 1976:196; Di Paolo 1989:195-196; Fennell 1993:430,434-435; Mishoe and Montgomery 1994:3-5; Nagle 2003:349). Regional American multiple modal Englishes more specifically include South Midland American Englishes from Virginia to Oklahoma, Southern American Englishes particularly throughout the Gulf States from Virginia to Kosur 6 Texas, and Utahan Englishes (Herndobler and Sledd 1976: 196; Di Paolo 1989:195-196; Fennell 1993:430, 434-435; Mishoe and Montgomery 1994:3-5; Nagle 2003:349), which are roughly illustrated in the American dialect map in (4). Research by Herndobler and Sledd (1976:196), Di Paolo (1989:196), and Mishoe and Montgomery (1994:4-5) finds that no major division by social class or geographical area exists for multiple modal use within multiple modal Englishes. Every speaker of a multiple modal English uses at least one multiple modal at least occasionally regardless of socioeconomic status, racial or ethnic affiliation, education level, and geographical location (Herndobler and Sledd 1976:196; Di Paolo 1989:196; Mishoe and Montgomery 1994:4-5). That multiple modals are a feature across both vertical and horizontal axes of variation is a result of the history of multiple modal constructions in the history of the English language. History Kosur 7 Multiple modal constructions similar to Modern English multiple modals first appeared in the English language approximately eight hundred years ago shortly after the beginning of the Middle English Period (Herndobler and Sledd 1976:196). Early recorded examples of multiple modals include six occurrences of shal (not) mowe "shall (not) may," the most frequent multiple modal in Middle English, from the Wycliffe Bible as identified by Ogura (1993:539-541): (5) a. b. c. d. e. f. And if Sathanas hath risen aʒeins hym self, he is disparpoilid, and he shal not mowe stoned, but hath an ende. And loo! thou shalt be stille, or doumbe, and thou shalt not mowe speke til in to the day, in which thes thingis schulen be don. Stryue ʒe for to entre by the streit ʒate; for I seye to ʒou, many men seken for to entre, and their shulen not mowe What here I this thing of thee? ʒeld resoun of thi ferme, for now thou schalt not mowe holde thi ferme. forsothe thei that be worthi to that world, and rising aʒen fro deede men, neither ben weddid, nether wedden wyues, nether schulen mowe deye more for I schal ʒyue to ʒou mouth and wisdom, to whiche alle ʒoure aduersaries schulen not mowe aʒenstonde, and aʒenseye The Middle English multiple modal shal mowe, however, differs from Modern English multiple modals in that both modals express epistemic modality, which is no longer possible in Modern English (Traugott 1972:192; Nagle 1989:8-10; Fennell 1993:434; Mishoe and Montgomery 1994:19). Therefore, although the Modern English interpretation of shall may would be "a reasonable inference is that it is permissible," the Middle English understanding of shal mowe is closer to "a reasonable inference is that it is logically possible" (Fennell 1993:434; Palmer 1990:57). Influenced by Old Norse and other Scandinavian languages (Ogura 1993:547; Fennell 1993:433), Old English also produced multiple modal constructions although Old English modals differed significantly from Modern English modals in that Old English Kosur 8 modal verbs operated as full verbs as opposed to the defective, neutral modals of Modern English (Fennell 1993:431). As modals shifted from main verbs to auxiliaries, so, too, did multiple modal structures. In Old English, two different modal verb constructions were possible, the first in which deontic modals functioned as full verbs with objects and complements and the second in which epistemic modals functioned as auxiliary verbs (Fennell 1993:432). Multiple modal constructions were possible so long as one modal functioned as a full verb (Fennell 1993:433). Deontic modals were much more abundant in Old English than epistemic modals, but, due to the gradual loss of the subjunctive mood in Middle and Modern English, epistemic modals expand "to assume the modality that was previously expressed by subjective markers" (Fennell 1993: 432). By the end of the Middle English Period, all modal verbs completed the shift from full verbs to auxiliary verbs (Fennell 1993:433; Herndobler and Sledd 1976:185), which allowed for the epistemic-deontic multiple modal combinations of Modern English. Multiple modals resembling the Modern English constructions found in Southern American Englishes, African American Englishes, Northern British Englishes, Scottish Englishes, Northern Irish Englishes, and Caribbean Englishes most likely originated in Northern British and Scottish Englishes (Fennell 1993:434-435; Mishoe and Montgomery 1994:19). During the seventeenth century, Northern British and Scottish settlers brought multiple modal constructions to the Ulster Plantation in Northern Ireland (Fennell 1993:434; Mishoe and Montgomery 1994:19), which explains the retention of multiple modals in Northern Irish Englishes. From Northern Ireland between 1718 and 1775, at least a quarter of a million Ulster Scots, or Scotch-Irish, immigrated to Central Pennsylvania in the United States and moved south through Virginia and Georgia Kosur 9 (Leyburn 1962; Dickson 1988; Fennell 1993:434; Mishoe and Montgomery 1994:19). The Scotch-Irish continued to immigrate to the southern United States as well as the Caribbean and West Indies during the eighteenth and nineteenth centuries (Fennell 1993: 435; Mishoe and Montgomery 1994: 19-20), thus explaining the occurrence of multiple modal constructions in Southern American Englishes, African American Englishes, and Caribbean Englishes. Pragmatics As previously discussed, multiple modals in Modern English occur only in constructions of an initial epistemic modal followed by one deontic modal and sometimes although rarely two deontic modals (Fennell 1993:434). As single modals, deontic modals often function as politeness markers within requests in the form of interrogative sentences in which the subject is the second person pronoun you (Palmer 1990:191): (6) a. b. c. d. Would you please pass the salt? Please pass the salt. Could you clean the bathroom today? Clean the bathroom today. Although the purpose of 6a and 6c differs from 6b and 6d in that 6a and 6c are interrogative sentences and 6b and 6d are imperative sentences, the semantic meaning is the same for 6a and 6b and for 6c and 6d; the speaker is making a request for the listener to pass the salt in 6a and 6b and to clean the bathroom in 6c and 6d. The difference between 6a and 6b and between 6c and 6d instead involves pragmatic meaning; the use of the modal verbs would and could in 6a and 6c results in more polite requests than the use of the imperative mood in 6b and 6d. As deontic modals, would and could nevertheless still convey commands for performances (Palmer 1990:69,71). As single modals, however, epistemic modals express subjective judgments about Kosur 10 the proposition as a whole as opposed to a performance (Palmer 1990:50,69). For example, compare the following two sentences in which the first modal verb expresses epistemic modality and the second modal verb expresses deontic modality: (7) a. b. c. d. You could help out around the house more if you need something to do this summer. (epistemic) You could help out around the house more, or you will be punished. (deontic) You can watch the movie now because the satellite is working again. (epistemic) Because you finished cleaning your room, you can watch the movie now. (deontic) The epistemic modals in 7a and 7c results in the semantic meanings of It is possible for you to help out around the house more and It is possible for you to watch the movie now while the deontic modals in 7b and 7d results in the semantic meanings of It is commanded that you help out around the house more and It is permissible for you to watch the movie now. Both epistemic modality and deontic modality express the subjective judgments of the speaker (Palmer 1990:7); the difference is the focus of the epistemic on beliefs and of the deontic on behavior. Multiple modals allow speakers to combine epistemic and deontic modality. Di Paolo (1989:198) offers the example of a conversation that she experienced between herself and a middle-aged female clerk in a small fabric store. Di Paolo brings in a blouse that she wishes to alter and engages the clerk in a conversation. As a customer, Di Paolo is in a higher status position than the clerk. Therefore, when the clerk wishes to assert her opinion and change the mind of the customer, the clerk employs the multiple modal might could: "You might still could keep the cuff [the way it is] and have French cuffs [on the blouse]." and "You might could keep the cuffs.…" The clerk wants to influence the customer to leave the original style of cuffs on a blouse and softens the command Kosur 11 with the deontic modal could as a politeness marker (Di Paolo 1989:198). However, as the interlocutor of lower status, the clerk further softens the command with the epistemic modal might as if to say, "It is possible that it is suggested that you keep the French cuffs," rather than "I suggest you keep the French cuffs." According to research by Mishoe and Montgomery (1994:6), two pragmatic conditions govern the use of multiple modals: "(1) one-on-one conversation (very often in the form of negotiations); and (2) a threat to the "face" (in the terms of Goffman 1967, etc.) of one or more speakers in a conversation." As defined by Brown and Levinson (1987:61), face is "something that is emotionally invested, and that can be lost, maintained, or enhanced, and must be constantly attended to in interaction." Di Paolo (1989:198) found that, in addition to the spontaneously produced multiple modals, she could elicit more MMs from the clerk by acting doubtful or hesitant but not hostile about the suggestions offered, which supports the claim that multiple modals occur most often in facesaving contexts very often in one-on-one face-to-face verbal conversations. The clerk maintains the public self-image of herself as a clerk and Di Paolo as a customer through the use of the multiple modal might could. Because of the pragmatic condition of one-on-one face-to-face conversation, the grammatical subjects of clauses that contain multiple modals are most often first and second person pronouns (Mishoe and Montgomery 1994:13-14). For example, the first five examples of multiple modals in actual speech from the naturalistic sample (see Di Paolo 1989:197 for a description of the data-gathering procedures) from which Di Paolo (1989:195) gathers her data contain only instances of multiple modals with first or second person pronouns as grammatical subjects: Kosur 12 (8) a. b. c. d. e. I don't think I have any grants you might could apply for. We might can go up there next Saturday. I may could at Finger's. You know, if you drank half a drink, you might oughta go home and sleep it off. This thing here I might should turn over to Ann. Of the 236 instances of multiple modals in the North/South Carolina corpus (see Mishoe and Montgomery 1994:26 for a description of the corpus) from which Mishoe and Montgomery (1994:13) gather their data, almost seventy percent (163 of the 236) of the multiple modals occur with first or second pronouns: the first person singular pronoun I with sixty-one multiple modal predicates, the first person plural pronoun we with fifty multiple modal predicates, and the second person pronoun you with fifty-two multiple modal predicates. That multiple modals most often occur with first and second person subjects in actual speech further supports the claim that MMs most often occur in one-onone conversations in which the speaker comments on the interlocutors, both speaker and hearer(s), present (Montgomery and Mishoe 1994:13). Within the pragmatic condition of one-on-one face-to-face conversation, multiple modals most frequently occur in very specific context, most notably negotiations with strangers, service interactions in particular, but also with family and friends (Montgomery and Mishoe 1994:14-15,23). As defined by Montgomery and Mishoe (1994:14), negotiations are interactions in which "one person attempts to resolve an issue or question by seeking action, agreement, information, or approval from another person." For example, in the clerk-customer interaction described by Di Paolo (1989:198), the multiple modal might could occurs during a negotiation over the alteration of a blouse. Out of the 189 (148 and 41) of the 236 examples of multiple modals collected directly by Mishoe and Montgomery (1994:9,14), seventy-four percent (139 of the 189) of the Kosur 13 multiple modals occur in one-on-one interactions with 114 between family and friends and twenty-five between strangers, most often in service interactions as illustrated in (9). (9) a. b. c. Sorry, we don't carry them anymore, but you know, you may might can get one right over there at Wick's. The only thing I might would have like that is an egg-shell mattress. If you like it, I might can sell it. Of the remaining fifty examples of multiple modals, twenty-two percent (42 of the 189) occur in settings with family and friends while only four percent (8 of 189) occur in formal public settings (Mishoe and Montgomery 1994:14). Just as with the use of multiple modals in one-on-one conversations, however, multiple modals most often occur within the context of negotiation in small group and public settings (Mishoe and Montgomery 1994:13-14) as illustrated in (10). (10) a. b. c. I think I may will have me a piece of cake. One of our goals might ought to be to encourage non-member involvement. Are there any other prayer concerns that you may would want to share? That speakers of multiple modal Englishes actually do most often use multiple modals within the very specific context of negotiation, both privately and publicly, further supports the two supposed pragmatic conditions governing the use of multiple modals. In addition to the pragmatic condition of negotiation, multiple modals also frequently co-occur with a variety of other mitigations, hedges, and politeness markers in addition to the "inherent mitigation which modals express" (Mishoe and Montgomery 1994:14). Mitigation, as defined by Labov and Fanshel (1977:84), is a discourse strategy in which a speaker modifies his or her expression to express sensitivity to the hearer and avoid creating offense. A more specific type of mitigation, hedging is a means to Kosur 14 moderate the assertion or commitment of a speaker to a stance through strategies such as tag questions, epistemic markers, discourse markers, and rising intonation (Smith 2009b). For example, the following multiple modals from instances collected by Mishoe and Montgomery (1994) and Di Paolo (1989) co-occur with mitigating adverbs (11a, 11b) or hedging discourse markers (11c, 11d, 11e): (11) a. b. c. d. e. Could you might possibly use a teller machine? (Mishoe and Montgomery 1994:11) If we had known ( ), we may still could have done it. (Di Paolo 1989:216) I reckon I might should better try to get me a little bit more sleep. (Mishoe and Montgomery 1994:9) I don't think I have any grants you might could apply for. (Di Paolo 1989:195) You know, if you drank half a drink, you might oughta go home and sleep it off. (Di Paolo 1989:195) The adverb possibly, which expresses contingency or uncertainty, in 11a lessens the certainty of the double modal might could, one interpretation of which is it is possible that it is suggested. Similarly, the adverb still, which describes an action or state that is unchanging or uninterrupted, in 11b modifies the double modal may could, which roughly means it is (more) possible that it is suggested. Other adverbs that commonly cooccur with multiple modals include maybe, maybe...just, possibly...just, probably, and even (Mishoe and Montgomery 1994:14). The discourse markers I reckon in 11c, I (don't) think in 11d, and You know in 11e function to distance the speaker from the propositions of the clauses containing the double modals: I might should better try to get me a little bit more sleep, I (don't) have any grants you might could apply for, and you might oughta go home and sleep it off. The co-occurrence of multiple modals with other mitigations, hedges, and politeness markers supports the claim that multiple modals most often occur within the specific context of negotiation. Kosur 15 In addition to distancing the speaker from the proposition, discourse markers like I reckon, I (don't) know, and you know move multiple modals from a main verb clause to an embedded noun clause. Such mitigating hedges as discourse markers create the pragmatic context of indirectness (Mishoe and Montgomery 1994:13). Other syntactic constructions that contain embedded clauses such as stative verbs with noun clause direct objects, noun clauses functioning as complements, conditional adverb clauses, and clauses in indirect questions. Of the 236 instances of multiple modals collected by Mishoe and Montgomery (1994:13), fifty-three percent (125 of the 236) occur in noun or adverb clauses as exemplified in the following examples: (12) a. b. c. d. e. I'm nervous; I'm afraid I may can't do it all. I'll tell you what we might should do. If we could might get a piece of car, things would be better. I wonder if we might could get a copy of last year's test? Sometimes I feel like I might will do all right. The indirectness created by embedded noun and adverb clauses further supports the two pragmatic conditions—negotiation and facesaving—governing the use of multiple modals. Identifying the relationship between indirectness and multiple modals is also essential for understanding the politeness and facesaving functions of multiple modals. In addition to the general definition of public-self image, Brown and Levinson (1987:61) posit two related aspects of face: negative face and positive face. Negative face is defined as "the basic claim to territories, personal preserves, rights to non-distraction— i.e. to freedom of action and freedom from imposition" (Brown and Levinson 1987:61). As Mishoe and Montgomery summarize (1994:15), negative face stems from "the need of people not to look bad or lose self-esteem or self-respect." Positive face is defined as "the positive consistent self-image or 'personality' (crucially including the desire that this self- Kosur 16 image be appreciated and approved of) claimed by interactants" (Brown and Levinson 1987:61). Again summarized by Mishoe and Montgomery (1994:15), positive face stems from "the desire of people to want approval and confirmation." Multiple modals allow speakers to attend to both negative face and positive face. For example, when the clerk in the interaction between Di Paolo (1989:198) and the fabric store clerk uses the multiple modal might could in the statement "You might still could keep the cuff [the way it is] and have French cuffs [on the blouse]," the clerk wants both for Di Paolo to agree with the suggestion and to not look bad (argumentative, aggressive) in front of the customer. Multiple modals in embedded clauses in particular allow speakers to attend to both negative and positive face of both the speaker and the listener simultaneously (Mishoe and Montgomery 1994:15). For example, compare the pragmatic meanings of the following five sentences with similar semantic meanings: (13) a. b. c. d. e. Dye your hair red. You should dye your hair red. You might should dye your hair red. I think that you should dye your hair red. I think that you might should dye your hair red. The semantics of all five sentences is roughly that the speaker wants the hearer to change her hair color to red. In 13a, the imperative mood expresses strong deontic modality; the speaker directly commands the hearer, which potentially threatens the negative face of the hearer. Moving across the pragmatic continuum, the combination of the discourse marker I think and the double modal might should in 13e results in a pragmatic meaning of "My personal judgment is that it is possible that it is desirable for you to dye your hair red." Compared to the strong deontic "I command you" in 13a, the epistemic "my personal judgment is that it is possible" further weakens the already weak deontic "it is Kosur 17 desirable" in 13e, resulting in a proposition that preserves both the positive face and the negative face of both the speaker and the hearer, which further supports the two pragmatic conditions governing the use of multiple modals. Structure Two grammatical analyses for explaining multiple modals, as identified by Di Paolo (1989:195), are the phrase-structure rule approach and the subcategorization approach. Although traditional phrase-structure rules impose the restriction of one modal verb per clause (Di Paolo 1989:195; Nagle 2003:355), modifying the rules allows for the iteration of modals in multiple modal Englishes (Di Paolo 1989:195). Therefore, in addition to the traditional VP→Modal+(Aspect and/or Voice)+Verb, a modified phrase structure rule allows for VP→Modal+Modal+(Aspect and/or Voice)+Verb and VP→Modal+Modal+Modal+(Aspect and/or Voice)+Verb. Similarly, the traditional subcategorization approach analyzes modals through the main verb analysis (also generalized phrase structure grammar), which assumes that modal verbs are finite forms subcategorized for base forms of non-modal verbs (Di Paolo 1989:195; Nagle 2003:358). To account for multiple modal constructions, the main verb analysis must allow for both finite and nonfinite forms of modal verbs to allow the initial finite modal to subcategorize subsequent base forms of modal verbs (Di Paolo 1989:195,213; Nagel 2003:358). Another, although recently disfavored, analysis of multiple modals is that the initial modal functions as either a verb phrase modifier or an adverbial in the form of an adverb (Herndobler and Sledd 1976:189; Mishoe and Montgomery 1994:7). However, as Di Paolo argues (1989:196), none of the traditional analyses nor the modified analyses adequately account for the intuitions about and the production of multiple modals by Kosur 18 speakers of multiple modal Englishes as well as the pragmatic conditions governing the use of multiple modals. Although traditional linguistic analyses allow for multiple modal constructions after considerable modification to the traditional theories, multiple modals are not phrases generated by a syntactic rule but rather periphrastic verb forms similar to phrasal verbs (look into), noun compounds (blackboard), verb-verb compounds (stir-fry), and other periphrastic idioms (Di Paolo 1989:196,214) as illustrated in (14). As Di Paolo (1989:214) explains, multiple modals share with other periphrastic lexical items three distinguishing properties: "nonproductivity, a combination of both unit-like and non-unit-like behavior, and syntactic and semantic irregularities." Like phrasal verbs, noun compounds, and other periphrastic idioms, the restrictions on multiple modals are not syntactic but rather idiosyncratic (Di Paolo 1989:205). Just as phrasal verbs are not syntactic combinations of any verb plus any preposition but rather idiosyncratic combinations of specific verbs plus specific prepositions, multiple modals are not syntactic combinations of any modal plus any modal but rather idiomatic modal-modal compounds of specific modal verbs. Kosur 19 First, multiple modals are nonproductive in that a syntactic rule such as MM→Modal+Modal results in unacceptable combinations of modal verbs that are indistinguishable in form from acceptable combinations (Di Paolo 1989:214). As previously identified, a total of only thirty multiple modal combinations occur naturally in the speech of all multiple modal speakers (Di Paolo 1989:197; Mishoe and Montgomery 1994:3). Furthermore, not all possible multiple modals are possible in all multiple modal Englishes but rather both individual varieties and individual speakers have restricted sets of multiple modals specific to the variety and speaker (Di Paolo 1989:214-215). If multiple modals were produced through a syntactic rule, then not only would all speakers of all multiple modal Englishes produce identical sets of multiple modals but also the set of possible multiple modal constructions would not be limited to the thirty identified naturally occurring combinations. As Di Paolo concludes (1989:215), multiple modal speakers instead learn (or at least prefer) particular multiple modals, a learning process that is comparable to other forms of vocabulary development. The nonproductivity of multiple modals that limits combinations to restricted, although unpredictable sets, supports the claim that multiple modals are single lexical items similar to phrasal verbs, noun compounds, and other periphrastic idioms. Second, multiple modals exhibit both unit-like and non-unit-like behavior in that English syntax treats multiple modals sometimes as one word lexical items and other times as multiple word lexical items (Di Paolo 1989:215). Multiple modals, therefore, even more closely resemble phrasal verbs syntactically. As periphrastic verb forms formed by a verb followed by one or more prepositions functioning as a particle, phrasal verbs also show both unit-like and non-unit-like behavior in that phrasal verbs may be Kosur 20 nonseparable or separable. In nonseparable phrasal verbs, the preposition must directly follow the verb resulting in a unit-like phrasal verb as in The woman ran into her friend but not *The woman ran her friend into. In separable phrasal verbs, however, the preposition can or must separate from the verb and follow the direct object resulting in non-unit-like behavior as in He looked the information up. Multiple modals similarly function as both units and non-units (Di Paolo 1989:215). For example, adverbs including negatives functioning as verb phrase modifiers may follow the entire multiple modal (unit) as in "The mother might should not put a blanket over her baby" or intervene between two modal verbs (non-unit) as in "I might just couldn't see it" (Di Paolo 1989:215-216). Additionally, auxiliary inversion in interrogative sentences also treats multiple modals as unit-like and non-unit-like constructions (Di Paolo 1989:216). For example, although, in both of the two instances of multiple modals in questions from the naturalistic sample from which Di Paolo (1989:216) gathers her data, auxiliary inversion affects only the second modal verb, related acceptability studies find that three auxiliary inversions—second modal inversion, first modal inversion, and multiple modal inversion—are possible in interrogative sentences containing multiple modals in the verb clause (Di Paolo 1989:216). That multiple modals exhibit both unit-like and non-unit-like behavior similar to other periphrastic lexical items like phrasal verbs further supports the claim of multiple modals are single lexical items. Third, multiple modals exhibit both semantic and syntactic irregularities again similar to other periphrastic compounds and idioms like phrasal verbs, verb-verb compounds, and noun compounds (Di Paolo 1989:217). Many compounds and idioms are Kosur 21 semantically noncompositional, meaning that the meaning of a periphrastic lexical item cannot be determined by combining the meanings of the individual parts but rather by considering the compound or idiom as a single whole (Di Paolo 1989:217). For example, the meaning of the phrasal verb squirrel away "hoard" cannot be determined by combining the meanings of squirrel (which alone is a noun referring to a rodent with a bushy tail, not a verb) and away (a preposition meaning "from a place"). Periphrastic structures must be understood as single lexical items. Other periphrastic lexical items like noun-noun compounds and adjective-noun compounds, however, are semantically compositional, meaning that the meaning of compound can be determined by combining the meanings of the individual words (Di Paolo 1989:218). Di Paolo (1989:218) offers the examples of blackboard and baseball glove, both of which whose meanings can be determined by combining the meanings of the adjective black with the noun board and the noun baseball with the noun glove. However, such compositional compounds as blackboard and baseball glove additionally impose semantic restrictions, i.e., blackboard refers only to chalkboards, not just any black-colored board and baseball glove refers only to fielding gloves, not just any glove worn for baseball (Di Paolo 1989: 217-218). Although multiple modals are compositional, the restricted set of naturally occurring multiple modals also restricts the semantics of multiple modals like other compositional compounds. That the nonproductivity of multiple modals restricts the possible combinations of modal verbs syntactically supports the semantic limitations of modal verbs and, therefore, that multiple modals are single lexical items rather than phrases generated by a syntactic rule. Conclusion Kosur 22 Multiple modals are lexical constructions of two or more modal verbs that express both epistemic modality and deontic modality. Strictly ordered, multiple modals are combinations of only an epistemic modal followed by one or more deontic modals, which is attested to by the thirty identified naturally occurring multiple modal combinations. Native speakers of many varieties of English including Southern American Englishes, South Midland American Englishes, Utahan Englishes, African American Englishes, Northern British Englishes, Scottish Englishes, Northern Irish Englishes, and Caribbean Englishes regularly use at least one multiple modal occasionally regardless of vertical variation factors. Developing from changes to the modal system from Old English to Modern English, multiple modals first appeared in English approximately eight hundred years ago, although such modal constructions differed significantly in meaning and form from modern multiple modals. The most significant change—from epistemic-epistemic combinations to epistemic-deontic combinations—established the required pragmatic and grammatical conditions governing the use of multiple modals: one-on-one face-to-face conversations; negotiations; facesaving contexts; first and second person pronouns; hedges, mitigations, and politeness markers; and indirectness. However, traditional linguistic analyses including the phrase-structure rule approach and the subcategorization approach that allow for multiple modals provided that one modal is a full verb with a base form fail to take into consideration the semantic, pragmatic, and syntactic limitations of multiple modal constructions. Instead, multiple modals are single lexical items, an argument supported by the restrictions on multiple modal constructions including the limited number of naturally occurring combinations. Kosur 23 References Boertien, Harmon. 1986. "Constituent structure of double modals." Language variety in the south: Perspectives in black and white, ed. by Michael Montgomery and Guy Bailey, 294-318. Tuscaloosa: University of Alabama. Brown, Penelope and Stephen Levinson. 1987. Politeness: Some universals of language usage. Cambridge: Cambridge University Press. Butters, Ronald. 1973. "Acceptability judgments for double modals in southern dialects." New ways of analyzing variation in linguistics, ed. by Charles-James N. Bailey and Roger Shuy, 276-286. Washington: Georgetown University Press. Bybee, Joan, Revere Perkins, and William Pagliuca. 1994. Mood and modality. The evolution of grammar: Tense, aspect, and modality in the languages of the world, 176-242. Chicago: The University of Chicago Press. Coleman, William. 1975. "Multiple modals in southern states English." Diss. Indiana University. Dickson, R. J. 1988. Ulster emigration to colonial America, 1718-1775, second edition. Belfast: Ulster Historical Foundation. Di Paolo, Marianna. 1989. Double modals as single lexical items. American Speech 34(3).195-224. Fennell, Barbara A. 1993. Evidence for British sources of double modal construction in Southern American English. American Speech 68(4).430-437. Goffman, Erving. 1967. Interaction ritual: Essays on face-to-face behavior. Garden City, NY: Anchor. Herndobler, Robin and Andrew Sledd. 1976. "Black English—Notes on the auxiliary." Kosur 24 American Speech 51(3/4).185-200. Hopper, Paul J. 1999. A short course in grammar. New York: W. W. Norton & Company. Huddleston, Rodney. 1984. Introduction to the grammar of English. Cambridge: Cambridge University Press. Labov, William and David Fanshel. 1977. Therapeutic discourse: Psychotherapy as conversation. New York: Academic Press. Leyburn, James. 1962. The Scotch-Irish: A social history. Chapel Hill: University of North Carolina Press. Mishoe, Margaret and Michael Montgomery. 1994. The pragmatics of multiple modal variation in North and South Carolina. American Speech 69(1).3-29. Nagle, Stephen J. 1989. "Double Modals in Early English." International Conference on Historical Linguistics. Rutgers, NJ. —. 2003. "Double modals in the southern United States: Syntactic structure or syntactic structures?" Modality in contemporary English, ed. by Roberta Facchinetti, Manfred Krug, Frank Palmer, 349-372. Berlin: Mouton de Gruyter. Ogura, Michiko. 1993. Shal (not) mowe, or double auxiliary constructions in Middle English. The Review of English Studies 44(176).539-548. Palmer, F. R. 1990. Modality and the English modals. London: Longman. Palmer, Frank. 2003. "Modality in English: Theoretical, descriptive and typological issues." Modality in contemporary English, ed. by Roberta Facchinetti, Manfred Krug, Frank Palmer, 1-17. Berlin: Mouton de Gruyter. Smith, K. Aaron. 2009a. "The history of be fixing to: Grammaticization, sociolinguistic Kosur 25 distribution and emerging literary spaces." English Today 97 25(1).12-18. —. 2009b. Class lecture presented in Sociolinguistics at Illinois State University, Normal, IL. Traugott, Elizabeth Closs. 1972. A history of English syntax: A transformational approach to the history of English sentence structure. New York: Holt. 1 Other catenative verbs like need, want, have (expressing obligation), dare, plan, and intend resemble the quasi-modals ought to and used to in that the form of both verb types within verb phrases is verb-preposition-verb (need to study, ought to study); but, whereas quasi-modals are auxiliary verbs in which the preposition to functions as a particle, catenative verbs function as the heads of verb phrases followed by verb phrase complements or direct objects in the form of infinitives, present participles, or base forms: compare needs a new coat, needs to buy a new coat, and needs you to buy a new coat with ought to buy a new coat and compare had a party and planned a party with had to plan a party, planned to have a party, and planned on having a party. Within the quasi-modal construction ought to buy a new coat, the modal verb ought functions as a modal, the preposition to functions as a particle, the verb buy functions as a verb phrase head, and the noun phrase a new coat functions as a direct object. Within the non-modal and catenative constructions needs a new coat, needs to buy a new coat, and needs you to buy a new coat, the noun phrase a new coat, the verb phrase to buy a new coat, and the noun clause you to but a new coat all function as direct objects with the verb needs functioning as a verb phrase head. Within the catenative constructions had to plan a party and planned to have a party, the catenative verbs had (meaning be obligated, not possess) and planned (meaning intend, not organize) function as verb phrase heads, the infinitives to plan (meaning to organize, not to intend) and to have (meaning to arrange, not to be obliged) function as verb phrase complements, and the noun phrases a party and a party function as direct objects. The difference between catenative verbs with verb phrase complements and catenative verbs with direct objects depends on the structures that can follow the catenative verb; only verbal (verb phrases) structures follow catenative verbs with verb phrase complements while both verbal (verb phrases) and nominal (noun phrase, noun clause) structures follow catenative verbs with direct objects. The periphrastic be going (to) and be fixing (to) also possess some but not all properties of true modal verbs in that both verbs function as futurate auxiliaries (Smith 2009a:12); both be going (to) and be fixing (to) are also grammaticalized constructions (Smith 2009a:12) of the progressive aspect also followed by a verb phrase complement in the form of an infinitive (Smith 2009a:15). Some copula-adjective and copula-participle combinations like be able (to) and be supposed (to) are also sometimes mistaken as quasi-modal verbs; however, the infinitive following the adjective or participle function as adjective phrase or verb phrase modifiers within the higher subject complement function: compare be able to ride a bike and be capable of riding a bike. 2 Catenative and grammaticalized auxiliary verbs differ from modal and quasi-modal verbs in that catenatives and grammaticalized auxiliaries are neither defective nor neutral. For example, the forms of the catenative need are to need, needing, needed*, need, needs, and needed*, the forms of the grammaticalized auxiliary have (to) are to have (to), having (to), had* (to), have (to), has (to), and had* (to), and the forms of the grammaticalized auxiliary be going (to) are to be going (to), been going (to), am/is/are going (to), and was/were going (to). Note that as grammaticalized constructions of the progressive aspect, the auxiliaries be going (to) and be fixing (to) lack present participial forms. 3 Catenative verbs and grammaticalized auxiliary verbs again differ from modal verbs and quasi-modal verbs in that catenatives and grammaticalized auxiliaries may appear in verb phrases that contain a single modal or quasi-modal verb in Standard English: for example, may want to, used to have (to) and would rather be going (to).