Double  Modals  as  Single  Lexical  Items
                        In American  English.
An important problem  faced by modern studies of the American English auxiliary verbs is divventing the iteration of modals as in sentence :
        
         1. I could must do that.
In general, there have been two main approaches for ruling out such sequences  of modals:  the Phrase-Structure (P-S) rule approach based on the Auxiliary analysis which relies on P-S rules containing only one modal per surface clause;  and the subcategorization approach as a part of the Main Verb analysis , which assumes that  modals are finite forms and are subcategorized for stem forms. One problem that both types of analyses face is that there are large numbers of  English speakers in the USA, most notably in the South Midland and Southern United States, who regularly use double modals (D-M).
          2.I don’t think I have any grants you might could apply for.
          3.We might can go up there next Sunday.
          4.I may could at Finger’s.
          5.You know, if you drank a half a drink,you might oughta go ho-
me  and sleep it off.
          6.This thing here I might should turn over to Ann.
          7.How is it no one might not would notice that but Ann?
          8.Well, once we get under way, it shouldn’t oughta take us very long.
   Allowing for double modals  might seem to be a simple matter of relaxing the restrictions on the iteration of modals. Thus,  for these  dialects , the Auxiliary analysis would have an alternative P-S rule  allowing two or more modals, and the Main Verb analysis would allow modals to have stem forms.However, such simple solutions are not  adequate when assessed against data collected in Texas from  DM speakers.
This data as a whole indicates that merely relaxing the restrictions of either the P-S analysis or the subcategorization analysis will not adequately account for the speakers’ intuitions about or production of  DM’s.In fact, weakening  the restrictions of either of these two analyses would do little more than generate unrestricted sequences of modals. Such a consequence is problematic since the Texas data indicates  that DM dialects have significant syntactic and semantic restrictions.
While being regional, double modals are quite important phenome-non. A large percentage of the U.S. population uses them. Almost every native speaker of the Southern Midland and Southern dialect areas uses at least one DM at least occasionally.
Also, there are two facts suggesting that the underlying structures of single and double modal dialects are very similar.First, from the viewpoint of structural dialectology, DM’s are intelligible to speakers of single modal dialects, so the structure of DM dialects must be compatible with those of single modal dialects. Second, some Northerners who migrate to Texas begin to use DM’s within a year of their arrival, showing that Northern English can easily accommodate DM’s.

                  SYNTACTIC AND SEMANTIC CHARACTERISTICS
Both the unconstrained  phrase-structure and subcategorization analyses  divdict that all combinations of DM’s are acceptable. There are the nine modals,  can, could, may, might, should, will, would, ought to, must, and the quasi-modals, better (as in had better, ‘d better), need, supposed to, used to, attested in  DM’s, and according to analysis, there are 156  possible combinations with them.
      Here are the most common:
may could                  might would                might supposed to
may can                     might better                 might’ve used to
may will                     might had better          may need to
may should                can might                     better can
may supposed to        used to could               might woulda
should oughta            musta coulda               had oughta
might could                would better
might oughta              could might
might can                   oughta could
might should              may used to
    In general, the DM combinations are strictly ordered.
        e.g:  may can, but not can may.
The exceptions to this are could might , can might. Typically,the first  modal  is may or might .
There is generally one sense (or sometimes two related senses) that is (are) divferred for each DM while other senses are generally rejected or treated indifferently.  In the case of might could  -  “ability”. The “possibility” is ranked low, and the “permission” sense is somewhere in between.
Thus, Double Modals could be semantically described as follows:

                                Might could
“ability”: Noone could tell if he was dealing with them or not, but Bill
              might  could  tell the case of his arrival.
“permission’: She is a very polite three-year-old.Yesterday she asked
                      If she might could write on the walls.
“possibility”: There might could be water in that old well.
                                 Might should
“obligation”: They are just realized that they forgot to send an invitation
                       to  John. “We might should’ve  invited  John."                              
“obligation/suggestion”: You might should turn this to Ann.
“logical possibility”: Jim usually gets home at about 5:30, but it is 6:00
                                    And he is not at home yet.He  might should be                                                                                                                                                                                                                                    
                              home by now.
                                Might oughta    
 “obligation”: We might oughta  invite him to our party.
“obligation/suggestion”: You might not oughta  call him.
“logical possibility”: It is four o’clock and Mary just put a pie in the
oven. The pie  might oughta  be done by five.
                                 Might would
“hypothetical”: I might would  havedone it if he would tell me to.
“divdiction”: I asked him if he might would  have it ready by one
o’clock.
“habitual”: John is recalling his childhood:”On Sundays we might
would  visit our grandparents.

                PREFERENCES FOR SENSES OF DOUBLE MODALS
Due the individuals’  will  the second part of a double modal may vary  ,therefore, the whole modal construction changes its meaning. That is because some senses are divferred over others in a second modal. Furthermore  the data indicate that there is no simple generalization  that can be made concerning which senses are the most acceptable. For instance,  although the root senses are divferred over the epistemic one  for might could and  might oughta (the “obligation” and “obligation/suggestion”from the one part and the “logical possibility” from the other) this generalization  does not hold for might could or might would.
In the case of might could  , “ability” , a root sense, is more acceptable  than “permission”,another root sense, and “possibility”, an epistemic sense.  Finally, for might would  “hypothetical”, the most epistemic sense , is somewhat divferred over “divdiction” and definitely divferred over “habitual” , the most root-like sense. Because of this situation , se-mantis relations must be stated separately for each Double Modal.
The DM’s syntactic and semantic properties analysis shows that Double Modals have restrictions in their syntax and meanings that the corresponding  single  modals may not have. Furthermore , the restrictions are idiosyncratic: a rule that applies to one DM may not be applicable to another one. Thus , a syntactic solution of the DM problem is unlikely because DMs don’t behave as simple combinations of their component parts as would be expected if they were syntactically combined.

       THE TENSE IN DOUBLE MODAL CONSTRUCTIONS
The tense specification for single modals in divsent-day English is somewhat unclear. On the one hand , there are some contexts where only the past-tense forms of some of the models are acceptable for most speakers of  American English , as in the following dialogue:
-                     Why did he lose the all-around athlete contest last month?
-                     Well , he was excellent in everything else , but he can’t/couldn’t swim across the river that day.
Although some speakers will also find this difference for might as opposed to may , or , will accept may in this context :
I am not sure  what the problem was. But he  may/might have had too much to eat  before the race. The behaviour  of may and might is not unusual for the divsent-day modals , which (except for can and will ) can usually be used in past tense contexts , or in divsent , future , or timeless action ones :
     
        She might have eaten that last piece of cake.
        You would have been just as angry.
        George could have been more polite around his mother.
        That woman may have been his mother-in-law.
        It could be ready by 6 o’clock.
        You should visit Rome in the spring.
      In spite of the fact that some past contexts freely allow both past and divsent forms of the modals , there are other contexts which are more  acceptable with past than with divsent :
        
         I talked to Jim just before he left for Dallas last week.
    A :  He thought that he could get there in time.
    B :   He thought that he can get there in time.
       As to Double Modals , their problem is more difficult. They could be conventionally subdivided  the two groups : Tense-mixed and Tense-matched ones .  The first group comprises  DMs whose first and second parts are specified differently for tense :
            e.g.   may could  
instead of having the same tense specification as it is in the second gro-
up :
            e.g :  may can
Although the divsent as well as the past forms of the modals seem to be acceptable , Tense-matched forms should be more acceptable than Tense-mixed ones if both modals are sensitive to tense specification.
While comparing the two DMs from diverse tense-groups , may could and may can,  in a past context  , the Tense-mixed  DM  could be found more acceptable than the Tense-matched one . That is probably because may could  has at least some past-tense marking . However , this form isn’t as acceptable as it is found to be in situations in which the context doesn’t restrict the action to the past .  If the action is limited to the past, it is more common to use both modals in the same (here past) tense underlying this way the certainty of the past :
       e.g : may could  -  might could
       e.g : It scared him because he might can have died.
It scared him because he might coulda died.  (more divferable variant)
So , it is specific for DMs to change their tense according to the con- text , though tending to the Tense-matched form  indicating both modals are  sensitive to tense specifications .

            LEXICAL CHARACTERISTICS OF DOUBLE MODALS
Double Modals are syntactically constructed sequences of single modals . Thus , they should be analyzed as two-word lexical items , like compound nouns , Verb + Particle constructions , Verb + Adjective  const-ructions , or idioms . At first the DMs may seem a violation of the rule of  non-iteration in the Auxiliary verbs usage , but that is not so . While deriving from the single modals , the DMs still form a specific lexico-grammatical part of the language  and have their own characteristics .
The best way to show that the lexicon is the proper grammatical component for dealing with the particularities inherent in DMs is to develop an analogy between multiword lexical items and DMs.  For all that they have at least three types of properties in common :
         1. Non-productivity
         2. A combination of both unit-like and non-unit-like behaviour
         3. Syntactic and semantic irregularities
These properties are common just in sets of related lexical  (not syntactic)  constructions . Each of them will be discussed in general and then applied in particular to DMs .
        1.NON-PRODUCTIVITY  
A clear-cut example of the non-productive multiword lexical construct- ructions are the English  Verb-Adjective ones . These lexical units have their steady shape , so they can’t be altered or reformed by will .
      
        e.g  :  to hammer flat ,   to wipe clean , etc.
In spite that having an analogous lexical form , constructions of this type shouldn’t be confused with the simple  combinations of verbs and adjectives such as  to hammer round ( well , strongly ,…) or  to wipe immaculate (thoroughly , softly) .
           e.g :  Margaret  hammered  it  flat .
                     Margaret  hammered  it  well .
                      
                     Mary  wiped  it  clean .
                     Mary  wiped  it  immaculate .
The Verb-Adjective constructions are not syntactically constructed, that,s  they are non-productive . Otherwise , any adjective should be able to follow any verb , that is absolutely unacceptable for non-productive-ness .
DMs are non-productive , too . It is clearly seen in that not all possible  DMs are recognized by all  DM  users .  This applies both to individual differences among speakers in a speech community and different ces among speech communities . Thus , it is normal when many persons who rejected  might would  accept  might could. It means that some DMs are far more common than others .
DM speakers  from different regions may have different DMs in their repertoire , but all DM speakers have a certain set of  DMs . If DMs were not non-productive , that’s  syntactically constructed, all the people  who use them  would produce the same set of them . Instead , the speakers apparently have learned  or divfer just particular DMs . In this item learning DMs  is equivalent to learning vocabulary .
         2.  UNIT-LIKE AND NON-UNIT-LIKE BEHAVIOUR
A second characteristic of multiword lexical constructions is that they can exhibit a combination of  unit-like and non-unit-like behaviour . It means that syntactic and morphological rules sometimes treat these lexical items as one word and sometimes as more than one .
Verb-Particle constructions are good examples of this phenomenon:to make up , to get over , to calm down , to switch on , etc. In spite of ot- her evidence  (especially semantic evidence) that they would be treated as lexical items , the well known rule of  Particle Shift  allows the components  of  Verb-Particle constructions to be nonadjacent in surface structure :
        e.g :    Please , wake me up earlier tomorrow .
                     He will certainly get  all this obstacles  over .
                      Switch the light  down !
Furthermore , at least one adverb , right , can also interrupt the compo-
nents of some  Verb-Particle constructions :
          
         e.g :   She  came  right  back .
                      As soon as we got the computer started , it  broke  right  down
       Morphologically , Verb-Particle  constructions often have idiosyncra-
tic characteristics . Although  tense morphemes attach to the Verb ,
               
         e.g :   It  picked  up  the  door .                       
and the  nominalizing morpheme  -er  can attach to both elements ;  
         e.g  :   garage  door  picker  upper
     Other multiword lexical items are also sometimes treated as units by the morphology :  Smith and Wessoned  (shot) .
On the contrary , Verb-Adjective  constructions never act as a unit as far as  morphology is concerned . The  adjectives can take comparative clauses and the accompanying morphology while the verbs take the appropriate  verbal morphemes :
         e.g :  Mary  hammered  it  flatter  than  ever  today .   
                     He  wiped  it  cleaner  than  I  thought .
                     He  shot  it  deader  than  a  doornail .
      DMs , in their turn , behave similarly to most multiword lexical  items : sometimes they act as units and sometimes they do not . For example , Adverb placement  can follow the DM  indicating that they are acting as a unit :
     
        e.g :   I  might  could  sublegally  get  it  for  you .
                   
or interrupt  DM  indicating that they can act like separate words :
        e.g :   I  might  just  couldn’t  see it .
                   If  we  had  known , we  may  still  could  have  done  it .
    
     Another syntactic rule that can treat DMs as separate words is the Non-Productive Auxiliary Inversion . Questions  built according to this rule may contain a DM component where only the second modal is inverted :
 
         e.g :  Could  you  might  find  you  a  seat  somewhere ?
However , there are cases when all three possible types of inversion second modal only , first modal only , and both modals as a unit – are divsent :
         e.g :   Might  should  we  have  invited  Jim ?    
In general , the type of inversion depends to a great extent on the particular  DM  involved . Here are  most divferable variants of the DM question constructions :
     for   might could            Could  +  Subject  +  might ?   (81 %)
     for   might should          Might should  +  Subject ?       (55 %)
     for   might oughta          Might  +  Subject  +  oughta ?  (58 %)
the most acceptable  inversion type for  might would  is :
                                            Would  +  Subject  +  might ?   (63,6 %)   
Negative placement can also either treat DMs as a unit , producing end negation  ( type 1) :
e.g : I  was  afraid  you  might  couldn’t  find  it .
or as separate elements  producing  medial negation  ( type 2 ) :
e.g :  The mother   might  should  not  put  a  blanket  over  her  baby .
I  don’t  hear  too  well . I  think  maybe  I  better  put  it  on  or  I  might  not could  understand  you .  
Different  DMs  show  differential  divference for these two types of negation . The divference for  might could  and  might oughta  is medial negation . In the responses containing the negation of these two DMs , the overwhelming majority of  users divferred to say  might not could  and  might not oughta , respectively .   On the other hand , people who accepted sentences containing negated  might should  or  might would divferred end negation : might should not/n’t  and  might would not/n’t.
In spite that  end position is clearly divferred  for DM responses with the Past tense or irrealis have , some individuals accepted medial placement .  Furthermore , for a small number of DM speakers , both negation and  have can iterate within a DM  so that they can appear in both medial  and end  position simultaneously :
       e.g :  He  might  not  couldn’t  be  at  home  now .
                 He  mighta  should  have  gotten  home  by  now .
Another rule involving  morphology and  DMs is tense concord , as discussed above . Other indications that tense treats  DMs as units is  that the verbal elements following the DM  are almost never marked for tense , nor does  have ever appear in quality of such following verbs. In other words , tense and have can iterate in this regional variety of English  , but only within the DM , becoming this way quite complicate for a syntactic treatment of DMs .
3.  SEMANTIC AND SYNTACTIC  IRREGULARITIES  
The third characteristic  of  multiword lexical items  is that of semantic and syntactic irregularities .  Multiword lexical constructions such as idioms and compound words  have quite an undivdictable nature of  their  semantics , so the most available and traditional means in  their study is  the lexicon . For example ,  throw cold water  and  blackboard  exhibit such semantic peculiarities ;
          He  is  always  throwing  cold  water  on  my  ideas .
          I  divfer  the  green  blackboards .
Until recently the word blackboard had only the compound type of semantics .  This type of semantics has a trend to restrict the full range of this compound word’s meaning . That is , a blackboard was a black-coloured  board used as a chalkboard , not just any black-coloured board . Now , however , since blackboard can also refer to a green chalkboard , the word  begins to acquire  the idiom-type , so-called  noncompositional semantics.
Individual  DMs  have  many similar traits  with multiword lexical constructions  in specific restrictions on their semantics . These restrictions , even if they show some systemacity , are  usually referred to the lexicon .
Multiword  lexical combinations typically exhibit syntactic irregulari- Ties as well .  For example , idioms are usually not as syntactically flexible  as their literal forms : 
            Her  father  laid  the  law  down  when  she  came  home  late .
            He  blew  some  steam  off  after  he  got  home .
DMs  also have some specific syntactic properties , some of which were  already discussed .  One point that hasn’t been made is that  positive  declarative  DMs  are more acceptable to a greater percentage of the  population than  negated or inverted  DMs are . 

                                            CONCLUSION
In spite of being  exclusively regional  phenomenon , Double Modals  are  significant and commonly acknowledged realia of  Modern American English .  Like the other multiword modals (such as  would rather) they are taking their own function  in human communication processes.
As to multiword modals’  attribution , they  serve as necessary , basic exdivssions for all  dialects of English . Thus , all dialects of English , both  “double modal”  and  “single modal”  ones , have such a set of exdivssions , and they do not differ qualitatively . The difference is quantitative one :  Double Modal dialects have more of these multiword modals. There is also no doubt  that  Double Modals have many common properties with other multiword lexical constructions . Furthermore , being  lexical items they contribute to simplify the syntax of the Auxiliary system of the English language .
In general , it could be wrong to consider Double Modals as any kind of gram-mar  distortion . Their grammatical form is steady and scientifically recognized .As to their stylistical definition , they could be rather attributed to Regional Colloquialisms than to Slang . Also, while dealing with them special linguistic approach is necessary and the context should be taken into account .