Document Sample
Wells Powered By Docstoc
					            Generative phonology in the late 1940s
                                      John Goldsmith
           This paper oers a careful reading of a paper published by Rulon Wells
       in Language in 1949 on the subject of automatic alternations in phonol-
       ogy. Read with a modern eye, it reveals that phonologists were exploring
       the value and use of phonological derivations, including both abstract rep-
       resentations and intermediate representations in the late 1940s. Contrary
       to what has been suggested in the literature, Bloomeld's explorations
       in rule ordering published in 1939 were not isolated and without inu-
       ence. Our conclusion is the null hypothesis: that there is an intellectual
       continuity from the work of Sapir and Bloomeld, through that of Wells
       and Harris to that of Chomsky and Halle. We conclude by oering some
       suggestions as to why this is not widely recognized in the eld.
    „he spring issue of Language of IWRW ˜rought — rem—rk—˜le new p—per to
its re—ders on the ˜eh—vior of morphophonemesD —nd the te™hniques th—t should
˜e used to un™over themF1 „his p—per w—s entitled eutom—ti™ —ltern—tionsD
@‡ells IWRWA —nd w—s the work of — young professor —t ‰—le …niversityD ‚ulon
ƒF ‡ells @IWIWEAD who would mu™h l—ter"in IWUT"˜e president of the vƒeF st
t—™kled the question of how to de—l with the f—™t th—t if we view the world from —
phonemi™ist9s perspe™tiveD we m—y w—nt to spe—k —˜out — single morphemeD like
the plur—l GEzG in inglishD h—ving sever—l distin™t phonemi™ re—liz—tionsD —ll of
whi™h —re predi™t—˜le from purely phonologi™—l inform—tion in the environment
of the morphemeF ƒu™h gener—liz—tions —re not the purview of phonemi™sD of
™ourseD from — phonemi™ point of view"˜ut they —re the responsi˜ility of the
linguistD for gener—liz—tions of this sort m—y ™onstitute — l—rge portion of —
l—ngu—ge9s stru™tureF2
   1 This   paper will appear in   Phonology   (2008). It is part of a larger project, in progress,

on the history of phonology conducted with Bernard Laks. I am endebted to him for a great

deal of discussion on these topics, as I am to a range of colleagues, including Pierre Encrevé,

Morris Halle, Georey Huck, Charles Hockett, Sidney Lamb, and Jason Riggle.              I am also

grateful for comments from Robert Ladd, François Dell, Peter Daniels, and the editors of this


  I have proted from the views of a number of scholars regarding the spirit and views of the

period treated in this paper, notably Stephen Anderson (1985), Eli Fischer-Jørgensen (1975),

H.A. Gleason (ms.), who kindly gave me a copy of an unpublished manuscript, dated 1988, on

the history of American linguistics in the 20th century, Dell Hymes and John Fought (1981),

and James Kilbury (1976).      Kilbury, for his part, sees the style of Wells's paper as rather

forbidding; this reader had rather the opposite reaction, as we will see below.
   2 We    will recall that Bloomeld's Postulates (Bloomeld 1926) had cleared the way for

    sn this p—perD ‡ells introdu™es expli™itly —ll of the re—soning th—t would ™h—rE
—™terize the he—rt of gener—tive phonologyX @iA underlying forms @whi™h he ™—lls
˜—si™ formsA whi™h m—y ˜e —˜str—™tD @iiA rules th—t derive surf—™e @phonemi™A
forms from the underlying forms ˜y rules th—t dyn—mi™—lly modify — segment
in the rule9s fo™us when it o™™urs in — p—rti™ul—r phonologi™—l environmentD @iiiA
the ™ru™i—l ™h—r—™ter of rule ordering in some ™—sesD —nd @ivA the ne™essity of
intermedi—te forms in — deriv—tionF
    wu™h of ‡ells9s p—per is thoroughly modern in ™on™eptionD —nd we sh—ll t—ke
the opportunity to go through it in some det—ilD ˜e™—use it foresh—dows"indeedD
presents"the dyn—mi™ —nd ruleE˜—sed ™on™eption of gener—tive phonologyD —nd
—lso ˜e™—use it dire™tly —ddresses the question of how to rel—te ruleE—ppli™—tion
with rep—irs of ™onstr—int viol—tionsD —nother topi™ th—t seems very ™ontempoE
r—ry in its perspe™tiveF „he p—per is org—nized into four p—rtsF sn the (rstD
‡ells dis™usses the dire™tion—lity inherent in some —ltern—tionsY in the se™ondD
some d—ngers th—t —rise from —n—lyzing morphophonemi™ ™h—nges —s h—ving ˜een
™—us—lly triggered ˜y viol—tions of surf—™e phonot—™ti™s"—n impli™it ™riti™ism of
some ƒ—pirEinspired phonologi™—l des™riptionF sn the third p—rtD he o'ers — r—ther
˜—roque —ttempt to ™l—rify how to de—l with morphophonemi™ gener—liz—tions
within — tr—dition—lly floom(eldi—n —nd st—ti™ ™on™eptionD —nd in the fourth
p—rtD — det—iled spelling out of — dyn—mi™ ™on™eption of morphophonemi™s"
wh—t we would tod—y ™—ll — deriv—tion—l —ppro—™hF ‡e will dis™uss e—™h of these
in turnF
    ‡—s ‡ells9s perspe™tive on phonology m—instre—m in the l—te IWRHsc „he
question is —nswered —lre—dy ˜y the f—™t th—t his —rti™le w—s pu˜lished in Lan-
guage X his w—s — view th—t h—d some novelty to itD —nd it w—s not — view th—t
everyone —lre—dy su˜s™ri˜ed toF yn the other h—ndD he w—s defending this ide—
—s one within the st—nd—rd theory of his d—yF re did not view himself —s —
revolution—ryD —nd his form of —rgument—tion w—s not noti™e—˜ly di'erent from
th—t used ˜y other phonologists in Language —t the timeF sn the (n—l se™tionD we
turn to the question whi™h motiv—ted this short p—per itselfX why would it ™ome
—s — surprise to — modernEd—y phonology th—t phonologists in the l—te IWRHs
were sket™hing the outline of —n —ppro—™h to phonologi™—l —n—lysis th—t would
™ome to ˜e known —s gener—tive phonology IH ye—rs l—terc ss there — story to ˜e
told if —ll there is is s™hol—rly ™ontinuityc „he storyD s will suggestD is th—t the
re—lity on the ground w—s intelle™tu—l ™ontinuityD while the story ˜eing told —fter
this discussion; there he makes the assumption 11 that [i]n a construction a phoneme may

alternate with another phoneme according to accompanying phonemes, and he denes such

an alternation as automatic if it is determined by the phonemes of the accompanying form,

rather than by morphological or grammatical information, and Bloomeld gives the example

of the forms of the plural morpheme {-s, -z, -ez} as an automatic alternation.


the f—™t is one of revolution—ry ™h—ngeF fut the story ™—nnot ˜e told without
dipping r—ther deeply into the pu˜lished phonologi™—l workD —nd to th—t we now

1          The directionality of morphophonemic change

‡ells ˜egins with the o˜serv—tion th—t if — morpheme h—s two —ltern—nts e —nd
fD e might ˜e predi™t—˜le from f without f ˜eing predi™t—˜le from eX the
rel—tionship m—y well ˜e —symmetri™—lX

           st does not follow th—t from the knowledge of morph e we ™ould
           predi™t the phonemi™ sh—pe of morph f and converselyF sn genE
           er—lD one of two —utom—ti™—lly —ltern—ting morphs is predi™t—˜le from
           the other ˜ut not the other from the oneD — situ—tion illustr—ted ˜y
           floom(eld9s f—mili—r ex—mple @Language PIVEWA from qerm—nX the
           morpheme for –round9 h—s the —ltern—nts |runt| ˜efore p—useD voi™eE
           less ™onson—ntsD —nd glott—l stopD —nd |rund| elsewhereY where—s the
           morpheme for 9motley9 h—s |˜unt| in ˜oth ™l—sses of environmentsF
           sn view of these f—™tsD |rund| m—y ˜e l—˜eled —s the basic alternant
           —nd |runt| —s derivative @opF™itF PIPAF @‡e sh—ll sym˜olizeX |rund|
           > |runt|Y or d > tFA   3

     ‡ells notes th—t he will use the  > not—tion for —ny —ltern—tion in whi™h the
element to the left of the  > is t—ken —s ˜—si™ to the element to the right of the
 >F „he not—tion thus emph—sizes th—t there ™—n ˜e —n inherent —symmetry
in the rel—tionship ˜etween — ˜—si™ form —nd the morphs @whi™h —re strings of
phonemesA th—t re—lize th—t formF
     xowD it is ™le—r th—t ‡ells is — ˜it un™omfort—˜le with wh—t he h—s just
done"or r—therD he re™ognizes th—t his re—der m—y ˜e — ˜it un™omfort—˜le with
it"—nd mu™h of the —rti™le is spent —n—lyzing the dyn—mi™—l ™h—r—™ter of the
perspe™tive whi™h not—tions like  > —nd terms like t—ken —s ˜—si™ will le—d toF
re is ™on™erned th—t this w—y of spe—king —˜out —n —n—lysis m—y —ppe—rD in his
termsD more pi™turesque th—n —™™ur—te"˜ut with th—t provisoD he —™knowledges
th—t this —n—lysis —llows him to formul—te the notion th—tX

           knowing the existen™e of |rund| we ™ould predi™t the morpheme |runt|
           —s the one th—t would o™™ur ˜efore p—use et™FD where—s knowing only
           the existen™e of |runt| we ™ould notD in view of the ˜eh—vior of |˜unt|D
    3 p.   101. I have added for clarity, and neutrality, vertical strokes, here and in some other

quotations, which do not appear in the original. The emphasis here, as elsewhere, is in the



           predi™t whether it would ˜e |runt| —g—in or |rund| th—t would —ppe—r
           ˜efore vowels et™F @IHIAF

     r—ving s—id th—tD he tells the re—der th—t he wishes to repl—™e picturesque-
ness with accuracy X to (nd — legitim—te —nd system—ti™ w—y to in™orpor—te these
linguisti™ insightsF „he w—y to do this is ˜y ™omp—ring two r—di™—lly distin™t
™on™eptionsD — static —nd — dynamic ™on™eptionD —s he ™—lls themD —nd the dyE
n—mi™ ™on™eptionD of ™ourseD is the one th—t employs the ™on™eptu—l met—phor
of one element ˜eing ™h—nged into —nother in — p—rti™ul—r environmentF4 es we
h—ve notedD the dyn—mi™ ™on™eption is wh—t would ˜e™ome the domin—nt perE
spe™tive from the midEIWTHs through the mid IWWHsD only to ˜e ™h—llenged ˜y
— num˜er of st—ti™ ™on™eptionsD in™luding de™l—r—tive phonology —nd optim—lity

2          Alternations triggered by constraint violation

‡ells ˜egins ˜y noting th—t he is —w—re th—t there —re some pitf—lls in front of
himD —nd th—t he h—s no intention of f—lling into themF st is ™le—r th—t he knows
th—t the motive for™e ˜ehind the dyn—mi™ ™h—nge isD —t le—st in m—ny ™—sesD the
—ppe—r—n™e of —n illi™it phonemi™ sequen™eD5 ˜ut he —lso is —w—re th—t we must
˜e ™—reful in how we de—l with this f—™tF st would not doD for ex—mpleD to s—yX

             @IA    ‡henD ˜y the pl—™ing of — morpheme in — ™ert—in phonemi™
                    environmentD — phonemi™—lly nonEo™™urrent sequen™e would
                    —riseD —n —ltern—tion or ™h—nge in this sequen™e is ™—lled —uE
                    tom—ti™ if it yields — phonemi™—lly o™™urring sequen™eF @pF

     ‡ells ™onsiders —n ex—mple th—t might seem to work —long these linesF ƒupE
pose one o˜served th—t in qreekD no ™onson—nt ex™ept n, rD or s —ppe—r wordE
(n—llyD —nd th—t —ll other ™onson—nts will ˜e droppedF ƒu™h — view would ˜e
motiv—ted ˜y p—irs in @PAD from galakt —nd stomatD respe™tivelyX wh—t ‡ells
™—lls ˜—si™ to the —utom—ti™ —ltern—tionD whi™h we would tod—y ™—ll underlyE
    4 Incidentally,   Wells's discussion appears to be the rst in which morphophonemic alterna-

tions are explicitly divided up into a focus and an environment (p. 100), where the focus

is the phonemic material that varies, and the environment is the linguistic material nearby

that is relevant to the appropriateness of the alternation.
    5 It   is probably unnecessary to point out that both generative phonological rules and opti-

mality theoretic constraints take illicit phoneme sequences as their point of departure: given

a rule A     →   B / CD, CAD is an illicit sequence, just as OT constraints specify structures

which a language may prefer to avoid.


ingF 6

              nomF sg       genFsgF
  @PA          gál—        gál—ktos       –milk9
              stóm—        stóm—tos     –mouth9

    ‡ells —grees th—t the d—t— in @PA motiv—tes —n —n—lysis like th—t in @IA @no
™onson—nt ex™ept n, r, or s ™—n st—nd —t the end of — qreek wordY —ll other
™onson—nts —re droppedAY —nd he —grees th—t this —n—lysis is one th—t — linguist
would (nd pl—usi˜le"nonethelessD — theory th—t ™ont—ins the prin™iple in @IA is
not su0™ientD ˜e™—use it would —lso ˜e ™onsistent with — l—ngu—ge like qreekD
˜ut in whi™h di'erent str—tegies were used in the two ™—ses to —void viol—tion of
the ™onstr—int on wh—t ™—n —ppe—r wordE(n—llyF es it isD qreek uses ™onson—ntE
deletion —s its rep—ir str—tegy in —ll ™—sesD ˜ut the theory in @IA does not for™e
th—tY it would ˜e ™onsistent with — di'erent di—le™t of qreek in whi™h some
stems s—tis(ed the ™onstr—int viol—tion ˜y using — su0x −o in the nomin—tive
singul—rF ‡ells un—m˜iguously s—ys th—t ‘w“e would ˜e willing to reg—rd gálakt
—nd stómat —s ˜—si™ to —utom—ti™ —ltern—tions if @—A their nomin—tive singul—rs
were gála —nd stómaD or @˜A if they were gálakto —nd stómatoD or @™A if they were
˜oth di'erent from their ˜—si™ —ltern—nts in —ny other w—yD provided th—t th—t
w—y w—s the s—me or ™omp—r—˜le in ˜oth ™—ses —nd —ll other essenti—lly simil—r
onesY ˜ut not otherwiseF sn ™ontempor—ry terminologyD ‡ells puts the requireE
ment on the ™onstr—intE˜—sed theory th—t the ™h—nge e'e™ted in order to s—tisfy
the ™onstr—int must ˜e the s—me in —ll ™—ses"—nd in even more ™ontempor—ry
termsD he requires th—t the ™onstr—int viol—tion triggers — spe™i(™ ruleF „h—t is
ex—™tly wh—t — gener—tive phonologi™—l rule doesF7
    ‡ells presents —nother —rgument —g—inst — surf—™eE™onstr—int ˜—sed —ppro—™hD
one th—t is perh—ps even more surprising in its pres™ien™eF re s—ys th—t the ™onE
str—ints th—t —re ™riti™—l for triggering morphophonemi™ ™h—nges —re not simply
˜—sed on wh—t sequen™es ™—nnot —ppe—r in — l—ngu—geX they —re ˜—sed on wh—t
sequen™es ™—nnot ˜e found —t morpheme ˜ound—riesF8 e l—ngu—ge ™—n disprefer
sequen™es @enough to trigger — rule to ™orre™t themA even if the l—ngu—ge —™™epts
them stri™tly inside — morphemeF ‡ells exp—nds on this pointX
   6 Wells   says explicitly that his notion of basic is intended to be understood as a statement

of synchronic, not diachronic, analysis in footnote 12a, where he says, the dynamically basic

alternant does not always present a historically older form, and gives an example from ancient

Greek that illustrates this point. McCawley (1979 discusses this point as well in the context

of William Dwight Whitney's perspective on rule application.
   7 Sommerstein      (1974) was an extended argument   against putting these two things together;
see Goldsmith (1993, 1999) for discussion.
   8 This notion would return in the guise of the Alternation Condition as well as strict cyclicity

in the framework of lexical phonology. See Kiparsky (1968).


      ‘„he ™onstr—int viol—tion“ ™ould o™™urD ˜ut not —t — morpheme ˜oundE
      —ryF e perfe™t —™tu—l ex—mple h—s not ™ome to the writer9s noti™eD
      ˜ut some ™lose enough —pproxim—tions h—ve ˜een en™ountered to
      m—ke the possi˜ility worth dis™ussingF sn ƒ—nskritD as + n yields
      on FFF —nd as + n yields an @‡hitney ŸIUSD ŸIUUA yet the phoneme
                 ¯             ¯
      sequen™es written asn D asn o™™urF ynlyD when they o™™ur there
      is no morpheme ˜ound—ry ˜etween the s —nd the nF por inst—n™eD
      the following words ™ont—in the su0xes snu, sna @‡hitney ŸŸIIWRE
      SAX vadhasnu –wielding — de—dly we—pon9D sth¯snu –(xed9D karasna
      –fore—rm9F „he sequen™e sn is then wh—t „ru˜etzkoy @PSTA ™—lls —
      neg—tive qrenzsign—lX — sign th—t there is no morpheme ˜ound—ry
      within itF
      ‡ould weD in this ™—seD s—y th—t the —ltern—tions as > o, as > a
                                                                ¯    ¯
      ˜efore n —re —utom—ti™ in ƒ—nskritc „he —0rm—tive —nswer will
      ˜e —n inst—n™e of wh—t we sh—ll ™—ll the wide st—ti™ ™on™eptionD
      the neg—tive —n inst—n™e of the narrow st—ti™ ™on™eptionF st is not
      in™um˜ent upon us here to ™hoose ˜etween these ™on™eptionsD ˜ut
      only to m—rk their di'eren™eF „hey will ˜e ex—™tly de(ned in ŸIHF
      @pF IHQA

3    Two static conceptions of automatic alterna-


‡ells next develops in some det—il — st—ti™ ™on™eption of —utom—ti™ —ltern—tionD
one whi™h is ™on™eptu—lly p—r—llel to the thenEdomin—nt ™on™eption of —llophonyD
˜y whi™h two —llophones m—y ˜e re—liz—tions of — single phoneme depending on
the environment in whi™h they —ppe—rF ƒu™h — ™on™eption @for —llophones —nd for
—utom—ti™ —ltern—tionsA is nonEpro™essu—l —nd nonEdyn—mi™D —nd ‡ells spends
from p—ge IHS to p—ge IHW spelling it outF st is r—ther ™omplexD —nd we le—ve its
det—ils to — noteF yne re—son—˜le interpret—tion of ‡ells9s str—tegy in this p—per
is th—t he feels o˜liged to present ˜oth — st—ti™ —nd — dyn—mi™ —n—lysis"th—t isD
if he h—d presented only the dyn—mi™ —n—lysisD his profession—l ™olle—gues would
h—ve ™—stig—ted him for not exploring the possi˜ility of wh—t they presum—˜ly
would h—ve preferredD — st—ti™ —n—lysisF re therefore presents — st—ti™ —n—lysisD
˜ut in su™h — w—y th—t its ™omplexities @in — thoroughly pejor—tive sense nowA
—re ˜rought ™le—rly to the foreF
    iven within this st—ti™ ™on™eptionD thoughD ‡ells m—kes the point quite
™le—rly th—t — ™onsistent —n—lysis of —utom—ti™ —ltern—tions requires the positing
of ˜—si™ forms th—t —re never found —s su™h on the phoneti™ levelX —s we would


s—y tod—yD ˜—si™ forms th—t —re —˜str—™tF „his ™—n h—ppen if there —re two
rulesD one of whi™h —pplies in one re—liz—tion of the morphemeD —nd one of whi™h
—pplies in the otherF re illustr—tes this point with some ex—mples from inglishF
     ‡ells points out th—t there isD in his wordsD —n —ltern—tion zero > @ ˜efore
prep—us—l or pre™onson—nt—l r, lD illustr—ted ˜y forms in @QAD ˜ut —lso —n —lterE
n—tion æb @ in unstressed positionF F „his for™es —n —˜str—™t —n—lysis of the
stem in th—t ™—seD ˜utX

          we need only re™ognize th—t theatr > theater exhi˜itsD simult—neE
          ouslyD two independent —ltern—tions"˜oth of them ™—p—˜leD —s it
          h—ppensD of ˜eing ™onsidered —utom—ti™X @—A æ b @ in unstressed
          positionD —nd @˜A zerob @ ˜efore prep—us—l or pre™onson—nt—l rD l @pF

     „he s—me point holdsD he s—ysD for the morphemes anal —nd letharg whi™h
—re ˜—si™—lly ænæl —nd leθargD though these forms never surf—™e —s su™hF
     s h—ve org—nized ‡ells9s ex—mples in @QAD using —n interspersing of st—nd—rd
orthogr—phy —nd phonemi™ represent—tion —s he does @though his ex—mples —pE
pe—r in running textAF

               ˜—si™ form   xo su0x    fefore -ous, -y   fefore -ize, -ist, -y   fefore -ysis, -ic
                            dis—ster       dis—strE
                            —n™estor       —n™estrE
                             —nger          —ngrE
                             no˜le          no˜lE
                  θijætr     θij@tr                                                   θijætr
                  ænæl                                           æn@l                  @næl
                  leθ—r§                                              &
                                                                 leθ@r§                    &

4         A dynamic conception of automatic alternation

‡ells introdu™es the dyn—mi™ ™on™eption in the following termsX

          „he m—nner of des™ri˜ing —ltern—tion th—t is impli™it in most gr—mE
          m—rs is quite di'erent from the st—ti™ one —s presented —˜oveF „he
          ™ustom—ry des™ription of —n —ltern—tion @not ne™ess—rily —utom—ti™A
          presentsD in its most form—lized dressD — qu—siE™hemi™—l equ—tionY eFgF
          rabh + ta = rabdhaF sn wordsX rabh ˜efore ta ˜e™omes rabY ta —fter
          rabh ˜e™omes dhaF „his m—nner of ™on™eption —nd des™ription we
          ™—ll dynamicD ˜e™—use it employs the met—phor of ™h—nge"™h—nge
          in the environment —s well —s in the fo™usF F F F„he di'eren™e ˜etween
          the dyn—mi™ —nd the st—ti™ ™on™eptions is o˜viousX the former t—kes


     —s the relev—nt ™onditioning environment the ˜—si™ —ltern—nt of the
     ™onditioning morphemeD the l—tter t—kes the deriv—tive —ltern—ntF st
     follows th—t there is no di'eren™e in result ˜etween the st—ti™ —nd
     the dyn—mi™ ™on™eptions when the ™onditioning morpheme rem—ins
     inv—ri—ntY for ex—mpleD the p—st p—rti™iple of the ƒ—nskrit root man
     –think9 is mataY if we des™ri˜e in phonemi™ terms the environment
     ™onditioning this —ltern—tion man > maD it would ˜e ta —™™ording to
     either ™on™eptionF fut when — formD eFgF rabdhaD ™ont—ins — deriv—E
     tive —ltern—nt of the environment —s well —s of the fo™us"whenD in
     other wordsD there is wh—t m—y ˜e des™ri˜ed —s re™ipro™—l ™ondiE
     tioning of two morphemesD the st—ti™ —nd the dyn—mi™ des™riptions

   ‚est—ted in gener—tive termsD ‡ells9s suggestion is th—t the dyn—mi™ ™onE
™eption is —ppropri—te when the underlying formD r—ther th—n the surf—™e formD
spe™i(es the ™ontext ™orre™tlyD while the st—ti™ ™on™eption is —ppropri—te when
the surf—™e form r—ther th—n the ˜—si™ @th—t isD the underlyingA form is the
—ppropri—te one for spe™ifying the ™ontext of the phonologi™—l ruleF re then
pro™eeds to illustr—te — deriv—tion—l —™™ount of the ƒ—nskrit formD pointing out
th—t we m—y in prin™iple —pply the two rules simult—neouslyD or ™ru™i—lly order
them in either of two w—ysX

     „here is in use — modi(™—tion of the dyn—mi™ ™on™eptionX the ™omE
     promise mentioned ‘—˜ove“F „his modi(™—tion gives — stepwise deE
     s™ription of — re™ipro™—l ™onditioningD —ssigning —n order of su™™esE
     sion to the stepsF …sing the met—phor of ™h—ngeD we might des™ri˜e
     the ™h—nge of rabh + ta to rabdha in —ny of three w—ysX
     @iA in one stepX rabh —nd ta ™h—nge simult—neouslyF „his is the pure
     dyn—mi™ ™on™eptionF
     @iiA sn two stepsX (rst rabh+ta ˜e™omes rabh+dhaY then rabh+dha
     ˜e™omes rab + dhaF
     @iiiA eg—in in two stepsX (rst rabh+ta ˜e™omes rab+taY then rab+ta
     ˜e™omes rab + dhaF
     sn @iiA ta ™h—nges ˜efore rabhY in @iiiA —fter itY —nd in @iA simult—E
     neously with itF „he re—son for ™—lling des™riptions @iiA —nd @iiiA
     ™ompromises ˜etween the pure st—ti™ —nd the pure dyn—mi™ ™onE
     ™eptions is th—t in e—™h of themD the ™onditioning —ltern—nt of one
     morpheme is its ˜—si™ —ltern—nt —nd the ™onditioning —ltern—nt of the
     other morpheme is its deriv—tive —ltern—ntF „he ™hief —dv—nt—ge of
     these ™ompromises is —n expository oneFFFX we m—y ˜e s—ved the need


      of st—ting —n extr— ruleD if the two steps into whi™h — re™ipro™—lly
      ™onditioned —ltern—tion ™—n ˜e ˜roken up —re e—™h of them ™overed
      ˜y — rule whi™h is needed —nyw—yF sn some ™—sesD in order to —™hieve
      this e™onomyD it is ne™ess—ry to spe™ify the order in whi™h the steps
      t—ke pl—™eY in other ™—ses it is notF @pFIHWEIIHA

   ‡ells points out then th—t his ™on™eption of phonologi™—l —n—lysis is found
in ‡hitney9s —™™ount of ƒ—nskritX

      ‡hitney himselfD in des™ri˜ing forms like rabdha @ŸŸISWETHAD uses the
      pure dyn—mi™ des™riptionD for the following re—sonF ris ™onsistent
      pr—™ti™e isD in —ll ™—ses where — pre™eding —nd — following morpheme
      ™ondition e—™h otherD —utom—ti™—lly or otherwiseD never to des™ri˜e
      the following morpheme —s ™h—nging (rstF sn ne—rly —ll these ™—ses
      he des™ri˜es the pre™eding morpheme —s ™h—nging (rst @™ompromise
      of type @iiiA —˜oveAY in those few ex™eptionsD like the ™—se of rabdha
      itselfD whereD for spe™i—l re—sonsD he does not do thisD wh—t he does is
      to revert to pure dyn—mi™ des™riptionD iFeF des™ri˜e the two ™h—nges
      —s simult—neousF @pFIIHA

   ‡ells then turns to some d—t— from v—tin th—t is —men—˜le to — dyn—mi™
—n—lysisD ˜ut whi™hD he will —rgueD is f—t—l for the st—ti™ ™on™eptionF

      vet us ™onsider the hypothesis th—t formul—e su™h —s pat –su'er9 C
      tus @p—st p—rti™ipleD nomF singF m—s™FA a passusD met –h—rvest9 C
      tus a messusD et™FD displ—y two —utom—ti™ —nd re™ipro™—lly ™ondiE
      tioning —ltern—tionsF ƒt—ted in dyn—mi™ termsD —nd redu™ed to the
      phonemes involvedD these —ltern—tions —reX
      @—A t ˜e™omes s ˜etween — short vowel —nd — following t whi™h in
      turn is followed ˜y — vowelY
      @˜A — t followed ˜y — vowel ˜e™omes s —fter — t th—t follows — short
      vowelF @pF IIHA

   ‡ells reformul—tes these rules in — st—ti™ ™on™eption —s followsX

      @eA ˜etween — pre™eding short vowel —nd — following s whi™h is in
      turn followed ˜y — vowelD deriv—tive s o™™urs inste—d of ˜—si™ tY
      @fA ˜etween — pre™eding sequen™e of short vowel —nd s —nd — followE
      ing vowelD deriv—tive s o™™urs inste—d of ˜—si™ tF @pF IIIA

   ‡ells ™onsiders sever—l forms th—t demonstr—te th—t rule @fA is in™orre™tD
—nd suggests th—t the strongest ™ounterex—mples —re est¯ –˜e thou39 —nd este


–˜e ye39F ƒimil—rlyD rule @eA is in™orre™tD on the ˜—sis of the form ets¯ –—nd yetD
   ‡ells o˜serves th—t — preferen™e for — dyn—mi™ ™on™eption m—y lie in p—rt
in linguists9 interest in histori™—l —n—lysisD —lthough he m—kes it ™le—r th—t the
˜—si™ forms th—t —re posited —re not ne™ess—rily the histori™—lly —nterior formsF
end he notes th—t the over—ll system ™—n ˜e ™omp—™tly expressed ˜y — not—tion—l
system whi™h fo™uses on morphophonemi™ stru™tureX

      ƒu™h — system is — ™l—ss of letters or other sym˜olsD e—™h of whi™h
      is s—id to design—te — morphophonemeD —nd to whi™h me—nings —re
      —ssigned in su™h — w—y th—t from the morphophonemi™—lly written
      formul— for — given morphemeD ˜y —ppli™—tion of the rules whi™h
      —ssign me—nings to e—™h of its ™omponent sym˜olsD one ™—n dedu™e
      some or —ll of its —™tu—lly o™™urring morphsF „he simplest me—ning
      th—t su™h — sym˜ol ™—n h—ve is simply one single phonemeD under
      —ll ™ir™umst—n™esY it is ™ustom—ry to use for su™h — sym˜ol the s—me
      m—rk —s for th—t phoneme itselfD ˜y — system—ti™ —m˜iguity whi™h
      is gener—lly dispelled ˜y the ™ontext or ˜y some expli™it ™onventionF
      „he next simplest me—ning isD th—t the morphophonemi™ sym˜ol desE
      ign—tes one phoneme in the neigh˜orhood of su™hE—ndEsu™h sym˜ols
      —nd —nother phoneme in the neigh˜orhood of su™hE—ndEsu™h other
      sym˜olsF „he ™onvention in this ™—se is to use the ™—pit—l letter ™orE
      responding to the sm—ll letter th—t design—tes the phoneme whi™h is
      reg—rded —s ˜—si™F @pFIIQA

   ‡ells is well —w—re th—t some import—nt —spe™ts of the ™on™eption th—t he is
dis™ussing h—s —ppe—red in the liter—tureD —nd —mong these he ™ites —re floomE
(eld9s wenomini morphophonemi™s —nd ƒw—desh —nd †oegelin9s @IWQSA e—rlier
—n—lysis of „ü˜—tul—˜—lF yf the l—stD he notes th—t the —n—lysis is very simil—r
in spiritD —nd writesX

      ƒw—desh —nd †oegelin @IHA s—yX –sf it h—s ˜een possi˜leD ˜y the
      re™ognition of — nonp—tent phonology whi™h involves the ™onstru™E
      tion of (™tive formul—eFFFD to redu™e the —pp—rent irregul—rity of
      „u˜—tul—˜—l phonology to systemD this very f—™t gu—r—ntees the
      truth of our theoryF9 roweverD it is not ™le—r wh—t the –theory9 isD
      —s distin™t from their ™onstru™tion of formul—eF yn pF P they reE
      m—rk th—t –the pro™ess of ™onstru™ting morphophonemi™ formul—e
      h—s some resem˜l—n™e to th—t of histori™oEphonologi™—l re™onstru™E
      tionF9 ss their theory the theory th—t their formul—e represent not
      only histori™—l re—lities ˜ut syn™hroni™ re—lities of some sort —s wellc


   ‡ells ™on™ludes with sever—l striking rem—rksD —ll of whi™h —re import—nt in
underst—nding the intelle™tu—l ™ontinuity ˜etween phonologi™—l theory in this
period —nd th—t of the next two de™—desF pirstD two phonemes ‘m—y“ yield one
in — dyn—mi™ st—tementD —nd he gives —n ex—mple from uore—n —nd from uot— in
whi™h two morphophonemes merge to — single phonemeF ƒe™ondD in the se—r™h
for not—tion—l ™on™isenessD p—ying —ttention to the pre™ise formul—tion of — rule
m—y —llow us to dispense with — rule ˜y de™omposing it into two independently
needed rulesD —nd he illustr—tes this with —n ex—mple from ƒ—nskritF9 „hirdD the
™omplete spe™i(™—tion of — phonemi™ —ltern—tion requires four itemsX
   @iA the phonemes involvedY
   @iiA their phonemi™ environmentY
   @iiiA the dire™tion of the —ltern—tionY —nd
   @ivA in the ™—se of nonE—utom—ti™ —ltern—tionsD the environments in whi™h
e—™h mem˜er of the —ltern—tion o™™ursF @IISA
end fourthD intermedi—te forms"whi™h ‡ells ™—lls ev—nes™ent forms"m—y ˜e
™onvenient in wh—t we tod—y would ™—ll the deriv—tionX

      st is sometimes ™onvenient to m—ke use of ev—nes™ent forms in st—ting
      —ltern—tionsF por inst—n™eD the v—tin —ltern—tions ‘dis™ussed —˜ove“FFF
      ™ould ˜e less stringently st—ted so —s to ™over the formul—X ut –use9
      C tus a ususF ‡e would postul—te —n ev—nes™ent form ussusD —nd
              ¯                                           ¯
      would then —dd —nother rule —™™ording to whi™h this ussus immeE
      di—tely ™h—nges to ususF „he formul— –ut + tus = ussus = usus9
                         ¯                  ¯          ¯       ¯
      re)e™ts histori™—l f—™tFFF˜ut the point we wish to m—ke is th—t there
      w—s — period @sometime —fter the ˜eginning of the smperi—l periodA
      when usus o™™urred —nd ussus did notD —nd th—t in des™ri˜ing the
           ¯                 ¯
      l—ngu—ge of this period the sym˜ol –ussus9 h—s neither —™tu—l nor
      hypotheti™—lD ˜ut only (™tive me—ningF @pF IISA

   „r—nsl—ting into ™ontempor—ry terminologyD ‡ells is s—ying th—t the form
u:ssus is neither — predi™ted surf—™e form @—™tu—lAD nor — hypothesized underE
lying form @hypotheti™—lAD ˜ut only — (™tive formX in modern terminologyD
—n intermedi—te formF
  9 The importance of the increasing concern with simplicity of description, in both philosophy

and linguistics during this period, cannot be over-emphasized.     For recent discussion, see



5        A few remarks on rule ordering in the late


„he prim—ry go—l of the present p—per h—s ˜een to look —t — single pu˜li™—tion
in some det—ilF sn the (n—l se™tionD we will —ddress the question —s to why
s™hol—rs —nd histori—ns of gener—tive phonology h—ve seen (t to ex—gger—te the
dis™ontinuity ˜etween phonologi™—l studies ˜eing pu˜lished in Language in the
IWRHs —nd th—t whi™h w—s developed in e—rly gener—tive phonology some ten to
(fteen ye—rs l—terF fut we need to spend —t le—st — ˜rief moment looking —t
the l—nds™—pe in the IWRHs ˜eyond ‡ells9s p—perF „here is re—lly no dou˜t th—t
some linguists did not like phonologi™—l —n—lyses with ordered rules"thoughD to
˜e sureD the s—me thing ™ould ˜e s—id —˜out the phonologi™—l ™ommunity during
the IWUHsD10 —nd s d—res—y th—t there is no single domin—nt view tod—yD —s s
write these wordsD with reg—rd to how issues of op—™ity should ˜e tre—ted in —
form—l theoryF fut returning to the IWRHsD the su˜je™t w—s — live oneD with —
r—nge of positions held ˜y di'erent peopleD for di'erent re—sonsF11
    „he eviden™e does not suggest th—t the p—per ˜y ‡ells whi™h we h—ve disE
™ussed here w—s p—rti™ul—rly un™h—r—™teristi™ of its timeF ‡ells expl—ined in the
(rst footnote th—t he h—d written the p—per in IWRTD —nd th—t he h—d gotten
™omments from g—rl †oegelinD gh—rles ro™kettD —nd fern—rd flo™hD —nd th—t
ro™kett h—d sent him — summ—ry of points th—t he h—d m—de on —utom—ti™ —lE
tern—tions in — p—per presented —t the IWRU vinguisti™s snstitute in enn er˜orF
es we h—ve —lre—dy seenD ‡ells w—s —w—re of the in)uenti—l p—per ˜y †oegelin
—nd ƒw—desh @IWQSA on „ü˜—tul—˜—lD he—vily —nd dire™tly in)uen™ed ˜y ƒ—pirF
sn this p—perD †oegelin —nd ƒw—desh —rgued th—t the form—l simpli™ity —nd efE
fe™tiveness of their —™™ountD with intermedi—te st—gesD w—s the ˜est —rgument
for the v—lidity of su™h —n —ppro—™hF €resum—˜lyD their theory w—s th—t their
formul—e represent not only histori™—l re—lities ˜ut syn™hroni™ re—lities of some
sort —s wellF
    ellig r—rris9s position on deriv—tions in phonology is nu—n™edD —nd must ˜e
understood in the ™ontext of his underst—nding of the go—l of linguisti™ —n—lyE
  10 I   recall one phonologist describing rule ordering then as a way to lie with phonological

  11 Rodney     Huddleston (1972) addressed this question to some degree:

         The problem raised by process models is thus to provide a non-temporal inter-

         pretation for the dynamic terminology of the meta-language. It is a problem that

         has long been recognized: one nds writers speaking apologetically of the `loose-

         ness' of the terminology, admitting that it involves `ctions' or `artifacts'; others

         take the position that it is simply invalid and should be avoidedLamb, for ex-

         ample, bases his main criticism of transformational grammar on such grounds.


sisD whi™h w—s pr—gm—ti™D in — ‚ortyE—n sort of w—y @‚orty IWWIAX —n —n—lyti™
te™hnique w—s —ppropri—te if it led to —n —n—lysis th—t w—s usefulF r—rrisD howE
everD seems to h—ve ˜een du˜ious th—t rule ordering provided enough ˜ene(t
to p—y for its ™on™eptu—l ™ostsF es er—vind toshi @person—l ™ommuni™—tionD
PHHTA h—s pointed outD r—rris —ppe—red to put ™onsider—˜le v—lue on the f—™t
th—t in — system with rulesD ˜ut no extrinsi™ orderingD there is — n—tur—l w—y to
rel—te deriv—tions to o˜je™ts with —n —lge˜r—i™ stru™tureY the imposition of rule
ordering deprives one of the possi˜ility of seeing — n—tur—l rel—tionship ˜etween
™l—sses of deriv—tions —nd — semiEgroup gener—ted ˜y the rulesF prom th—t point
of viewD it is n—tur—l th—t r—rris would look for w—ys to —void rule ordering
in —n—lysisF sn r—rris @IWRRD PHIESAD he suggests how phenomen— des™ri˜ed ˜y
xewm—n in dyn—mi™ terms ™ould ˜e h—ndled

     without ˜ringing in the time or motion —n—logy impli™it in –pro™ess9Y
     —nd without employing —ny prim—™y of the ˜—seF

˜ut —t the s—me timeD r—rris9s m—jor work re™ognizes the v—lidity of su™h —n
—ppro—™hF sn Structural Linguistics @IWSIAD he wroteX

           st is sometimes ™onvenient to ™onsider one of the mem˜ers to ˜e
     the sym˜ol of the new ™l—ss ‘phonemeD morphophoneme or other“Y
     th—t mem˜er is then s—id to ˜e prim—ry @or the ˜—seA while the other
     mem˜ers —re derived from it ˜y — set of environment—lly @or otherE
     wiseA ™onditioned –rules9 or oper—tionsF por ex—mpleD we m—y s—y
     th—t the phoneme GtG is the mem˜er segment ‘t“ plus v—rious ™h—nges
     in v—rious positionsF yr we m—y s—y th—t the morphophoneme GpG is
     the phoneme GfG plus the ™h—nge to voi™ing ˜efore {−s} –plur—l9FF F F sn
     —ll these ™—sesD we would ™onsider one mem˜er a —s prim—ry if we ™—n
     st—te the ™onditions in whi™h the other elements b, c repl—™e it @—re
     derived from itAF „he ™hoi™e of a is ™le—rer if we ™—n not reversi˜ly
     derive a from b or cY iFeF if we ™—n not st—te the ex—™t ™onditions
     in whi™h b is repl—™ed ˜y aF ‡hen no mem˜er of — ™l—ss ™—n ˜e set
     up —s prim—ryD it m—y ˜e possi˜le to set up — theoreti™—l ˜—se form
     from whi™h e—™h mem˜er ™—n ˜e derived @™fF in morphophonemi™sAF

   end — few p—ges l—terD in the ˜ook9s ™on™lusions @pF QUQAD r—rris ™onsiders
w—ys of —n—lyzing d—t— whi™h

     h—ve depended ultim—tely on movingEp—rts models su™h —s m—™hines
     or histori™—l s™ien™esF sn using su™h modelsD the linguisti™ present—E
     tion would spe—kD for ex—mpleD of ˜—se forms @eFgF in morphophoneE
     mi™sD where the o˜served forms —re o˜t—ined from the ˜—se form ˜y


     —pplying — phonemi™ su˜stitutionAD of derived forms @eFgF stems plus
     those —0xes whi™h —re —dded (rst in the des™riptive order migth
     ˜e ™—lled derived stemsAD or pro™esses whi™h yield one form out of
     —notherF sn —ll these types of present—tionD the elements —re seen
     —s h—ving historiesD so th—t the rel—tion of —n element to sequen™es
     whi™h ™ont—in it ˜e™omes the history of the element —s it is su˜je™ted
     to v—rious pro™esses —nd extensionsF pxX sn su™h present—tionsD —
     rel—tion ˜etween two elements a —nd b is essenti—lly the di'eren™e
     ˜etween two histori™—l or otherwise deriv—tion—l p—thsX th—t from A
     to a —nd th—t from A to bF A is set up —s — ˜—se from whi™h ˜oth a
     —nd b h—veD ˜y di'erent p—thsD ˜een derivedF

   gh—rles ro™kett9s re™olle™tionD in the e—rly IWWHsD w—s th—t rule ordering
w—s — hot topi™ in the l—te IWRHsF sn — letter to the present —uthor @pe˜ru—ry UD
IWWIAD he wrote

     uite —p—rt from pu˜li™—tionsD — num˜er of us @flo™hD „r—gerD r—rE
     risD †oegelinD ƒmithD toosA were in —™tive ™orresponden™e in the l—te
     IWRHs —nd e—rly IWSHsF s h—ve @or did h—veY some of them —re lostA letE
     ters from ellig r—rris th—t mention — young student n—med ghomE
     skyF yne of them spe—ks enthusi—sti™—lly of ghomsky9s work on
     re˜rew morphophonemi™sD s—ying th—t ghomsky h—d found — w—y
     to put the ordering of morphophonemi™ rules on — logi™—l ˜—sisF

   ro™kett himselfD on the other h—ndD w—s quite skepti™—l —t the time of the
usefulness of the —ppro—™hF re wrote @IWSRX PIIAX

     sf it ˜e s—id th—t the inglish p—stEtense form baked is –formed9 from
     bake ˜y — –pro™ess9 of –su0x—tion9D then no m—tter wh—t dis™l—imer
     of histori™ity is m—deD it is impossi˜le not to ™on™lude th—t some kind
     of priority is ˜eing —ssigned to bakeD —s —g—inst either baked or the
     su0xF end if this priority is not histori™—lD wh—t is itc

   ƒydney v—m˜D —lso in ™orresponden™e with the present —uthor @pe˜ru—ry PQD
PHHTAD h—s more re™ently noted th—t during his gr—du—te study —t ferkeley in
the (rst h—lf of the IWSHsD

     this method ‘of ordered rules“ w—s t—ught ˜y wurr—y imene—u in
     his morphology ™ourse —t …g ferkeley @whi™h s tookAF end —lso in
     his –ƒ—nskrit s—ndhi —nd exer™ises9 ˜ooklet whi™h s —nd others of his
     ƒ—nskrit students usedF

                                                  6     SEPARATION OF LEVELS

    „he weight of the eviden™e thus seems to ˜e th—t using — sequen™e of ordered
rules in — phonologi™—l —n—lysis w—s — re™ognizedD though not espe™i—lly popul—rD
theoreti™—l option ˜y the l—te IWRHs or e—rly IWSHsF ‡ells9s p—per w—s just wh—t
it —ppe—red to ˜eX — propos—lD ˜—sed on sever—l empiri™—l ™—sesD to t—ke more
seriously —n ide— th—t h—d ˜een dis™ussed ˜ut w—s not widely —™™eptedF

6     Separation of levels

e reviewer for this journ—l @who w—s himself — student of gh—rles ro™kettA
h—s rightly r—ised the question of the ™on™eptu—l ™onne™tion ˜etween ‡ells9s
work on phonologi™—l rules —nd the stru™tur—lists9 ™ommitment to — prin™iple
of sep—r—tion of levelsF „he role pl—yed ˜y ™on™erns of sep—r—tion of levels
deserves — long study ˜y itselfD perh—ps one simil—r in spirit to the present
p—perF „he gre—test p—rt of the eviden™e supporting sequenti—l rule —ppli™—tion
of the sort th—t we —re ™onsidering in this p—per is eviden™e only when we t—ke
into ™onsider—tion knowledge —˜out the w—y —n identi(—˜le morpheme @stem or
—0xA is re—lized di'erently in di'erent phonologi™—l ™ontextsF sf the prin™iple of
sep—r—tion of levels ruled out the use of th—t sort of eviden™eD then phonologi™—l
deriv—tions ™ould only r—rely ˜e justi(edD if —t —llF
    „he ˜ottom lineD in this writer9s opinionD is th—t there w—s — spe™trum of
opinion r—nging from those on one endD su™h —s toos —nd ro™kettD who strongly
supported the prin™iple of sep—r—tion of levelsD to those —t the other endD su™h
—s uenneth €ike —nd ellig r—rrisD who thought su™h — prin™iple w—s in no w—y
˜indingF €ike9s views on this —re wellEknownD ˜ut r—rris9s views seem to h—ve
˜een misrepresented in the liter—tureD so s will present some of r—rris9s thoughts
on this in some det—ilF „he upshot is th—t ‡ells9s —™™ount of rule —ppli™—tion is
s—fe within — r—rrisi—n view of the inter—™tion of morphology —nd phonologyF
    s o'er the following —n—logyF „he linguist who —n—lyzes — l—ngu—ge —nd proE
du™es — p—rti™ul—r gr—mm—r is like — fellow who needs to push — he—vy h—nd™—rt
from the tr—in st—tion to — do™k on the riversideD PHH feet downhill —nd — mile
—w—yD —nd he needs to do it without ever h—ving to p—ss through — stret™h where
the route goes uphillD —nd without ever going the wrong w—y down — oneEw—y
streetF „he linguists whoD like toosD supported — strong thesis of the sep—r—tion
of levels im—gined th—t this pro˜lem h—d to ˜e solv—˜le the very (rst time the
fellow —rrives —t the tr—in st—tionD —nd solv—˜le just ˜y looking —roundD seeing
whi™h streets seem to go uphill —nd whi™h downhill —t e—™h ™orner he p—ssesF
„he more li˜er—l linguistsD like ellig r—rris @—s we sh—ll seeAD thought th—t — rigE
orous solution w—s perfe™tly v—lid even if it me—nt th—t the fellow w—s permitted
to w—lk —round town —he—d of time —nd t—ke notesD —nd m—ke himself — m—pF es
long —s the eventu—l route o˜eyed the tr—0™ l—wsD why should the linguist not

                                               6   SEPARATION OF LEVELS

t—ke —dv—nt—ge of some free time the d—y ˜eforeD —nd exploit the more glo˜—l
knowledge he might o˜t—inc es we will seeD toos thought th—t phonologists h—d
—greed th—t su™h knowledge w—s illi™it for phonologistsD while r—rrisD like €ikeD
simply did not —greeF
   w—rtin toos @IWTRA expressed his point of view in IWTRD the extreme one th—t
is remem˜ered tod—yX

     uite — few of us —re old enough to remem˜er some of the linguisti™
     qu—rrels of the IWQHsF yne th—t p—rti™ul—rly sti™ks in my mind h—d
     to do with the ˜eginnings of phonemi™ theoryF st seems th—t there
     were two words spelled candied —nd candidD —nd the pro˜lem w—s
     to prove th—t they were di'erent wordsF floom(eld tr—ns™ri˜ed the
     (rst of them with GijG —nd the se™ond with pl—in GiGF xow th—t w—s
     — su0™ient solutionY ˜ut w—s it —lso — ne™ess—ry onec floom(eld
     himself s—id th—t it w—s not — ne™ess—ry solutionX eny tr—ns™ription
     th—t works is — good oneD he s—idY provided th—t it works through the
     whole lexi™on —nd is not w—steful of sym˜olsF h—niel tones pointed
     out th—t the two words ™ould ˜e tr—ns™ri˜ed with the s—me vowel
     sym˜olsD —nd candied with — hyphen ˜efore the (n—l ™onson—ntD —nd
     th—t then ˜oth the fritish st—nd—rd identity of the two words —nd the
     emeri™—n di'eren™e ˜etween them emerge from the single rule th—t
     ˜efore — hyphen one pronoun™es the s—me —s —t the end of the word
     candyF„he floom(eld theory of the d—y h—d no defense —g—inst th—t
     propos—lY in f—™tD floom(eld himself w—s not —˜ove using hyphens
     on o™™—sionF
     row did we get out of th—t j—mc fy instituting the prin™iple of
     the ƒep—r—tion of vevelsF „h—t hyphen w—s — gr—mm—ti™—l sym˜olD
     —nd it h—d no ˜usiness intruding into — phonologi™—l des™riptionY —nd
     ˜y the s—me tokenD nothing spe™i(™—lly sem—nti™ should ˜e —llowed
     to intrude into — gr—mm—ti™—l des™riptionF st w—s —n e—sy prin™iple
     to defendD for we ™ould simply rem—rk th—t every mixing of levels
     —mounted to ˜egging the questionF pirstD we s—idD the ™omplete
     phonemi™ des™ription without gr—mm—ti™—l ™ont—min—tions w—s —
     prerequisite to ˜eginning to des™ri˜e the gr—mm—r of — l—ngu—geY —nd
     then the ™omplete gr—mm—ti™—l des™ription would ˜e — prerequisite
     to ˜eginning the study of me—ningF
     fy the e—rly p—rt of the se™ond world w—r we were ere™ting defenses
     on ˜oth )—nks of this prin™iple of the sep—r—tion of levelsF yn one
     )—nk were r—nged the liter—ry ™riti™s"remem˜er veo ƒpitzer twenty
     ye—rs —go in Language c"who kept sniping —t us for denying th—t

                                               6   SEPARATION OF LEVELS

     me—ning existsY th—t for™ed us to s—y th—t we me—nt to use me—nE
     ing only di'erenti—llyX we promised to ™on(ne ourselves to —sking
     whether two things h—d the s—me or — di'erent me—ningF yne ˜old
     defenderD ellig r—rrisD undertook to show th—t even th—t employE
     ment of me—ning w—s unne™ess—ry for phonology —nd gr—mm—rY ˜ut
     most of us ™on™eded th—t life would ˜e too ™ompli™—ted on those
     termsF yn the other )—nk were those who pointed out th—t even the
     di'erenti—l use of me—ning w—s not enough for — pr—™ti™—l dis™overyE
     pro™edureX thereD the le—der w—s uenneth €ike with his qr—mm—ti™—l
     €rerequisites to phonologi™—l —n—lysisY —nd we ™overed th—t )—nk ˜y
     distinguishing ˜etween pr—™ti™—l —n—lysis with no holds ˜—rredD on
     the one h—ndD —nd on the other h—nd — pu˜lish—˜le des™ription for
     whi™h we would m—int—in the sep—r—tion of levelsF w—ny of us still
     m—int—in th—t th—t ™—n ˜e done with the help of — long spoonD ˜ut
     we ™—n9t deny th—t it is di0™ultF @SWETHAF

   toos ™ert—inly ™ould not h—ve ˜een ™le—rerD —nd he w—s right th—t €ike w—s
not —t —ll in —greementF fut he w—s quite wrong —˜out r—rrisD —s we ™—n see if
we —™tu—lly re—d wh—t he h—d to s—yF „he ™ru™i—l pointD —s the re—der will seeD
is th—t the phonemi™ —n—lysis w—s not completed —fter studying the distri˜ution
of soundsF uite to the ™ontr—ryY — tent—tive phonology h—d ˜een set upD ˜ut it
would ˜e modi(ed —s we le—rned more —˜out the morphologyF
   „his r—rris m—kes ™le—r in gh—pter V of Methods in Structural Linguistics
@IWSIAD whi™h presents — dis™ussion whi™h is quite surprising to — re—der tod—yD
˜e™—use while it des™ri˜es itself —s ˜eing ™on™erned with ˜ound—ry elements @—s
we would s—y tod—yY r—rris ™—lls them jun™turesAD it is re—lly —˜out —˜str—™t
—n—lysesF „he point of the dis™ussion is to show th—t — gre—t de—l of form—l simE
pli(™—tion ™—n ˜e —™hieved with the —ddition of — sm—ll —mount of —˜str—™tionF
r—rris9 ex—mple is the rel—tionship of G—yG@minusA —ndGeyG@slyD slynessAF
qiven the p—ir minus/slynessD the two phones seem to ˜e in ™ontr—stD —nd r—rris
™ites — simil—r ™—se for GeyG—nd GiyG @whi™h —ppe—rs to no longer exist in emerE
i™—n inglishAF fut GeyG9s environment is so restri™ted"it —ppe—rs prim—rily —t
the end of utter—n™esD plus in — few other words"th—t it seems in—ppropri—te
to set it up —s — sep—r—te phonemeD even though the le—ding prin™iples l—id out
so f—r dem—nd th—t this ˜e doneF
   r—rris proposes th—t this is — (ne pl—™e to posit —n —˜str—™t element @whi™h
he indi™—tes this w—yX GEGAD —nd it will —ppe—r in words su™h —s GslyEnessGF
„his —˜str—™t element is motiv—ted ˜y three ™onsider—tionsX we m—y ˜e —˜le to
redu™e the set of phonemes ˜y doing soF sn the ™—se —t h—ndD while we introdu™e
GEGD r—rris suggests we ™—n get rid of two phonemesD GeG —nd GiGD where—s

                                                   6   SEPARATION OF LEVELS

emeri™—n inglish of PHHT m—y h—ve only one soundD GeGD whi™h ™—n ˜e gotten
rid of hereF ƒe™ondD the —˜str—™t element m—y —™™ount for otherD quite sep—r—te
phenomen—F r—rris identi(es this ˜ound—ry element with one th—t would ˜e
posited in ™ompound nounsD su™h —s night-rate @whose pronun™i—tion is quite
di'erent from th—t of nitrate AY the —llophones of the (n—l sound of night in night-
rate —re di'erent from the single phone possi˜le in nitrateD —nd this di'eren™e ™—n
˜e des™ri˜ed ˜y positing — GEG jun™ture in night-rateF r—rris9s third —rgument
is th—t the GEG jun™ture ™oin™ides with — position of possi˜le p—useF r—rris —lso
suggests @pF VPA th—t this jun™ture element ™—n ˜e used to repl—™e the notion of
syll—˜leY inste—d of s—ying th—t — segment is the (rst segment of — syll—˜leD we
™—n s—y th—t it is pre™eded ˜y — GEGjun™tureF
      st is ™le—r th—t r—rris re—lizes he is on to — new methodD with this postul—tion
of ˜ound—riesF re writesD ˜y the setting up of the jun™turesD segments whi™h h—d
previously ™ontr—sted m—y now ˜e —sso™i—ted together into one phonemeD sin™e
they —re ™omplement—ry in respe™t to the jun™tureF yf ™ourse this )ies in the
f—™e of phonemi™ist methodologyY of ™ourse —ny ™ontr—st ™—n now ˜e —™™ounted
for without positing — new phoneme ˜y positing — jun™ture th—t triggers —
™ondition th—t the phoneme is re—lized in — spe™i—l w—y when in the ™ontext of
this ˜ound—ryF ƒo wh—t does r—rris doc yf ™ourse he tells his re—der th—t he is
doing nothing new3

        elthough the expli™it use of jun™tures is rel—tively re™entD the fund—E
        ment—l te™hnique is involved in su™h tr—dition—l linguisti™ ™onsiderE
        —tions —s –wordE(n—l9D –syll—˜i(™—tion9D —nd the use of sp—™e ˜etween
        written wordsF

      w—y˜e soY ˜ut m—ny phonologists were unh—ppy —˜out —llowing phoneme
re—liz—tion rules to ˜e sensitive to ™ontexts like wordE(n—l for pre™isely this
re—sonF xow r—rris does something ™uriousX he tells us wh—t — linguist doesX

        ‡hen — linguist sets up the phonemes of — l—ngu—geD he does not
        stop —t the ™omplement—ry elements of gh—pter U ‘th—t isD tr—dition—l
        phonemi™ —n—lysisD tq“D ˜ut ™o—les™es sets of these ™omplement—ry
        elements ˜y using ™onsider—tions of jun™tureF

      „hey do"if they —re ellig r—rrisF sn f—™tD r—rris goes on to point out th—t
in order to get inglish rightD we need to postul—te two @—˜str—™tA ˜ound—ry
elementsD one whi™h he notes —s 5 —nd the other —s EF „he (rst —ppe—rs
˜etween wordsD while the se™ond —ppe—rs inside ™ert—in wordsD like slynessD where
the stem sly is longer th—n — syll—˜le in — monomorphemi™ word like minus would

                                               6   SEPARATION OF LEVELS

   ‡h—t9s going on herec st ™ert—inly looks like r—rris is en™our—ging phoE
nologists to postul—te ˜ound—ry sym˜ols in order to simplify the phonology"
˜ound—ry sym˜ols th—t —re essenti—lly the re)e™tion of morphologi™—l stru™tureF
@‡h—t h—s h—ppened to morphologyEless phonologycA ‡h—t does r—rris s—yc
That's rightD he s—ysX

     „he gre—t import—n™e of jun™tures lies in the f—™t th—t they ™—n ˜e
     so pl—™ed —s to indi™—te v—rious morphologi™—l ˜ound—riesF @VUA

   sf — l—ngu—ge h—s predi™t—˜le penultim—te stressD for ex—mple"like ƒw—hili"
then we ™—n elimin—te stress —s —n element of the phonemi™ represent—tion just
—s long —s we in™lude word ˜ound—ries ˜etween the wordsF @VUA sn f—™tD r—rris
goes on to point out th—t the phonologist would ˜e wise to restri™t his use of
˜ound—ry sym˜ols to ™—ses where they re—lly do m—rk morpheme ˜ound—riesF
qerm—n presents —n interesting ™—seX the phonologist knows th—t wordE(n—l
o˜struents —re devoi™ed in qerm—nD —nd so he might w—nt to remove the voi™eless
o˜struents from the phonemi™ inventoryD repl—™ing everywhere GtGD for ex—mpleD
˜y Gd5GD ˜ut this would h—ve the unfortun—te ™onsequen™e of requiring us to
pl—™e 59s in —ll sorts of pl—™es th—t —re not —t —ll morpheme ˜ound—riesD like
right —fter — wordEiniti—l ™onson—ntX Teil –p—rt9 would ˜e Gd5—ylGD —nd this
would not ™orrel—te with morphologi™—l ˜ound—riesF
   sn inglishD r—rris notesD every ™—se where — GEG ˜ound—ry is neededD it ™orreE
sponds to — morpheme ˜ound—ry @—s in slynessD for ex—mpleAD ˜ut the ™onverse
does not holdX not every morpheme ˜ound—ry ™orresponds phonemi™—lly to —
GEGX r—rris s—ys playful does not h—ve — GEG ˜ound—ryD ˜ut trayfull does"purely
on des™riptive groundsF
   r—rris is quite ™le—r @pF VVA th—t phonemi™ —n—lysis should t—ke morphemi™
—n—lysis into —™™ount when the d—t— of the l—ngu—ge suggest th—t this ˜e doneX
phonologi™—l —n—lysis ™—n ˜e simpli(ed ˜y positing phonologi™—l ˜ound—ry eleE
ments whi™h typi™—lly ™orrespond to morphologi™—l ˜ound—riesX

     „he —greement ‘˜etween the needs of the phonemi™ —n—lysis —nd the
     ˜ound—ries motiv—ted ˜y morphology“ isD furthermoreD due in p—rt
     to the p—rti—l dependen™e ˜etween phonemes —nd morphemesF

   end the phonologists just m—y h—ve to guess where the morpheme ˜ound—ries
—re to ˜e pl—™edD ˜y seeing how this simpli(es the phonologi™—l —n—lysisX

     sn mu™h linguisti™ pr—™ti™eD where phonemes —re tent—tively set up
     while prelimin—ry guesses —re ˜eing m—de —s to morphemesD tent—tive
     jun™tures m—y ˜e de(ned not on the ˜—sis of —ny knowledge th—t
     p—rti™ul—r morphemes —re worth unitingF F F ˜ut only on the ˜—sis of

                                               6   SEPARATION OF LEVELS

     suspi™ions —s to where morpheme ˜ound—ries lie in given utter—n™esF
     @pF VWA

   „o summ—rizeD thenD in ™ontempor—ry termsX the phonologists m—y posit
—˜str—™t ˜ound—ry sym˜ols"—ny num˜er of them"in his phonologyD if he susE
pe™ts th—t — morphologi™—l —n—lysis will (nd motiv—tion for themF xo one ™ould
re—d this ™—refully —nd interpret this method —s one in whi™h phonemi™ —n—lysis
pre™edes morphologi™—l —n—lysis 3
   r—rris turns next to —nother w—y to le—d to — simpler phonologi™—l —n—lysis"
not ˜y positing —˜str—™t elements of — sort th—t —re never pronoun™edD ˜ut ˜y
—n—lyzing — phoneti™ sound —s ˜eing the re—liz—tion of two distin™t phonemesD
one pre™eding the other @though r—rris points out th—t re—lly this is — gener—lE
iz—tion of the —˜str—™t ˜ound—ry ™—seX pF WTAF r—rris refers to this —s rephone-
micizationD —nd its purposeD —nd its go—lD is twoEfoldX it —llows us to redu™e
the size of the underlying phoneme inventoryD —nd it elimin—tes @or simpli(esA
™onditions on wh—t sequen™es of phonemes —re permitting in phonemi™ repreE
sent—tionsF r—rris o'ers the ex—mple of the n—s—lized )—p in some emeri™—n
pronun™i—tions of paintingD whi™h ™—n ˜e re—n—lyzed phonemi™—lly —s — n—s—l
followed ˜y — GtGF
   „rue to his methodologi™—l prin™iplesD he does not insist th—t one must perE
form this kind of —n—lysisF w—ny linguists —re doing thisD r—rris knowsX „he
™urrent development of linguisti™ work is in p—rt in this dire™tion @WRAF fut
don9t feel o˜liged to do soX —ny degree of redu™tion —nd —ny type of simpli(E
™—tion merely yields — di'erentD —nd in the l—st —n—lysis equiv—lentD phonemi™
represent—tion whi™h m—y ˜e more or less suited to p—rti™ul—r purposesF @WRAF
ƒtillD this method ™—n ˜e very usefulY one of r—rris9s ex—mples ™on™erns re—n—lE
ysis of GˇG —s GsyGD whi™h does not le—d to — simpli(™—tion of the phonemi™
—n—lysisD ˜ut de(nitely simpli(es the morphologi™—l —n—lysisD sin™e it —llows us
to h—ve — single represent—tion for the morpheme admiss in admissible @with
GsGA —nd admission @with GˇGAF
   ‡here do things st—ndD thenD with —llowing morphologi™—l ™onsider—tions to
in)uen™e the phonemi™ —n—lysisc „he —nswer is essenti—lly thisX use morphoE
logi™—l inform—tion in developing — phonemi™ —n—lysisD unless th—t would h—ve —
™le—rly undesir—˜le e'e™t on the phonologyF sn r—rris9s words @pF IIIAX

     sf two segments h—ving di'erent environments @iFeFD nonE™ontr—stingA
     o™™urs in two morphemi™ segments whi™h we would l—ter wish to ™onE
     sider —re v—ri—nts of the s—me morpheme in di'erent environmentsD
     we will group the two segments into one phonemeF F F

   „h—t9s ™le—rX design your phonology in order to simplify the morphologyF
fut the senten™e ™ontinesX

                                                               7   DISCUSSION

      F F F provided this does not otherwise ™ompli™—te our gener—l phonemi™


      yur —ssignments of segments to phonemes shouldD if possi˜leD ˜e
      m—de on the ˜—sis of ‘purely phonemi™ ™riteri—“D sin™e ‘this p—rti™uE
      l—r prin™iple“ introdu™es ™onsider—tions dr—wn from — l—ter level of

    ƒo we ™—nnot w—lk —w—y from this s—ying th—t things —re ™ryst—l ™le—rD from
— methodologi™—l point of viewF „his seems like — f—ir summ—ryX use purely
phonologi™—l ™riteri— to ™ome up with the sm—llest inventory of phonemes —nd
the fewest ™onstr—ints on distri˜ution of phonemesF ‡hen it is possi˜le to
simplify the morphology ˜y rel—tively modest modi(™—tions of the phonemi™
inventoryD feel free to do so"r—rris ™ert—inly will do so himselfF fut don9t feel
o˜liged toD if you don9t w—nt toF ƒep—r—tion of levels is — ™onsider—tionD ˜ut one
whose signi(™—n™e is less th—n th—t of —ny insight th—t ™—n ˜e o˜t—ined on the
morphologi™—l level of —n—lysisF
    s urge the re—der to ™omp—re toos9s rem—rks —nd r—rris9sF r—rris w—s the
™onsumm—te theoreti™i—n who followed every hypothesis through to its ultim—te
™on™lusionD —nd he h—d no pl—™e for sep—r—tion of levels —s — fund—ment—l prinE
™ipleF toos dis—greedF „here w—s — r—nge of opinion during the IWRHs —nd IWSHs
in emeri™—n phonologi™—l theoryD —nd gener—tive phonology ™ontinued the tr—E
dition th—t w—s ™le—rly enun™i—ted in r—rris9 Methods X sep—r—tion of levels w—s
not — fund—ment—l prin™iple of linguisti™ —n—lysisF

7    Discussion

elthough no phonologist is o˜liged to ˜e interested in the history of his or her
dis™iplineD the w—y we view our p—st inevit—˜ly ˜rings — ˜i—s to some of our
workD if only with reg—rd to wh—t we ™onsider to ˜e new —nd ™it—tionEworthyF
fut most students of the history of s™ien™e give more weight to the import—n™e
of how — dis™ipline views its p—stD —nd the dis™rep—n™ies ˜etween the history
of — dis™iplineD —s do™umented in its —nn—lsD —nd its present—tion to — l—ter
gener—tion h—s ˜een of interest to more th—n one s™hol—rF „hom—s uuhnD in
his widely in)uenti—l Structure of Scientic Revolutions @IWTPAD rem—rked th—t
text˜ooks in the s™ienti(™ dis™iplines th—t he —n—lyzed tended to ˜e in—™™ur—te in
th—t they ex—gger—ted the extent to whi™h rese—r™hers in the p—st were gr—ppling
with the kinds of issues th—t we ™—re —˜out tod—yF re wroteD

                                                                   7   DISCUSSION

      FFFtext˜ooks of s™ien™e ™ont—in just — ˜it of historyD either in —n introE
      du™tory ™h—pter orD more oftenD in s™—ttered referen™es to the gre—t
      heroes of —n e—rlierF prom su™h referen™es ˜oth students —nd proE
      fession—ls ™ome to feel like p—rti™ip—nts in — longEst—nding histori™—l
      tr—ditionFFFF€—rtly ˜y sele™tion —nd p—rtly ˜y distortionD the s™ientists
      of e—rlier —ges —re impli™itly represented —s h—ving worked upon the
      s—me set of (xed pro˜lems —nd in —™™ord—n™e with the s—me set of
      (xed ™—nons th—t the most re™ent revolution in s™ienti(™ theory —nd
      method h—s m—de seem s™ienti(™F xo wonder th—t text˜ooks —nd
      the histori™—l tr—dition they imply h—ve to ˜e rewritten —fter e—™h
      s™ienti(™ revolutionF end no wonder th—tD —s they —re rewrittenD
      s™ien™e on™e —g—in ™omes to seem l—rgely ™umul—tiveF uuhn @IWTPD

   „he ™—se we h—ve looked —t in this p—per le—ns in the opposite dire™tionX the
pi™ture th—t we h—ve of phonology in the IWRHs is one th—t is overEsimpli(edF
„he existen™e of —n—lyses with deriv—tions —nd sequenti—l rule —ppli™—tion w—s —n
issueD even in the IWRHsD —˜out whi™h phonologists ™ould dis—greeD —nd pu˜lish
—rti™les @see the referen™es ™ited in footnote IAF
   „here h—s ˜een some dis™ussion in the liter—ture in re™ent ye—rs —˜out how
the e'e™ts of ordered rules were understood ˜y the emeri™—n linguisti™ ™ommuE
nity during the IWRHs —nd IWSHs @from˜erger 8 r—lle IWVWAD —nd the present
p—per is indeed motiv—ted ˜y — desire to shed further light on this questionF „he
p—per ˜y from˜erger —nd r—lle stimul—ted — reply ˜y in™reve @IWWUAD repu˜E
lished in somewh—t modi(ed form —s in™revé @PHHHAF from˜erger —nd r—lle9s
position is th—t —n—lyses with ordered rules in syn™hroni™ —™™ounts were widely
reje™ted in the IWQHs @they ™—ll this reje™tion the prev—iling wisdomAD ˜ut they
note th—t — solution with ordered rules w—s seriously proposed ˜y floom(eld
in IWQW in wenomini morphophonemi™sY this p—perD they suggest w—s so unE
known in emeri™— th—t ghomsky tells us th—t he h—d not re—d –wenomini morE
phophonemi™s9 until his —ttention w—s dr—wn to it ˜y r—lle in the l—te IWSHsF
in™revéD —s well —s uoerner @PHHQD PHHRA12 h—ve ˜oth —ddressed the question —s
to whetherD in retrospe™tD the view th—t ghomsky w—s un—w—re of floom(eld9s
work on ordered rules is pl—usi˜leF ‡h—tever the —nswer m—y ˜e th—t questionD
it seems to the present —uthor th—t the question is of little intelle™tu—l interestF
‡hy would we ™—re whether — student —t the …niversity of €ennsylv—ni— h—d
or h—d not re—d — tenEye—rEold pu˜lished p—per ˜y the most prominent linguist
of th—t time @iFeFD veon—rd floom(eldAc ƒtillD given the —mount of ink th—t h—s
˜een devoted to the questionD it is ™le—r th—t some people have ™—redF ‡hy soc
 12 See   also Koerner (2002, Chapter 9).

                                                               7    DISCUSSION

   „he —nswer surely is thisX there is — legitim—te interest in —s™ert—ining the
degree to whi™h the rise of gener—tive phonology in the IWSHs formed — ™ontinE
uous development with the intelle™tu—l themes th—t were in pl—y —lre—dy in the
IWRHsF ho™umenting th—t is the ™entr—l point of this p—perF yne might think
th—t eviden™e showing th—t this development w—s in f—™t continuous would ˜e
—˜out —s interesting —s the o˜serv—tion th—t — rooster ™rowed —t d—wnX why
shouldn't the (eld show ™ontinuity in its developmentD —fter —llc
   en —dequ—te response to th—t question would require — mu™h fuller —nd
more det—iled —™™ount th—n sp—™e permits in this p—perF ƒu™h —n —™™ount would
de—l not simply with the histori™—l development of ide—s in phonologyD whi™h
is the topi™ to whi™h s h—ve limited myself hereD ˜ut —lso the w—ys in whi™h
so™i—l units th—t —re l—rger th—n the individu—lD ˜ut ™onsider—˜ly sm—ller th—n
the dis™ipline"— group su™h —s e—rly gener—tive gr—mm—ri—ns"undert—ke to
form —n intelle™tu—l v—ngu—rdD —nd de(ne themselves in opposition to wh—t —re
per™eived to ˜e the domin—nt viewsF ƒu™h so™i—l form—tions —re ˜oth n—tur—l
—nd numerousD —nd in phonologyD they —re to ˜e found ˜oth ˜efore —nd —fter
the rise of gener—tive phonologyF sn order to strengthen the so™i—l ™h—r—™ter of
the form—tionD work must ˜e undert—ken to ™h—r—™terizeD in retrospe™tD wh—t the
domin—nt views wereD —nd this work is for o˜vious re—sons ˜i—sed in — ™ert—in
dire™tionF „he existen™e in the l—te IWRHs in m—instre—m journ—ls of ide—s th—t
were to ˜e™ome ™entr—l to gener—tive phonology is thus — thre—t to the su™™ess
of su™h — ™h—r—™teriz—tionF
   sn o˜serving thisD s h—ve in mind st—tements like the one in ghomsky @IWVT
pF IQD fnFQA where he notes th—t — modern ™ounterp—rt ‘to €—nini9s gr—mm—r“ is
@floom(eld IWQWAD whi™h w—s r—di™—lly di'erent in ™h—r—™ter from the work of
the period —nd in™onsistent with his own theories of l—ngu—geD —nd rem—ined virE
tu—lly without in)uen™e or even —w—reness despite floom(eld9s gre—t prestigeF
„he pu˜lished liter—tureD howeverD does not support this ™l—imD —nd we h—ve
seen in this ˜rief p—per th—t ‡ells goes out of his w—y to for™efully —nd did—™tiE
™—lly ™h—r—™terize the stepE˜yEstep —n—lysis th—t gener—tive phonology ™—me to
™h—r—™terize —s — deriv—tionF
   sndeedD the IWRW —rti™le ˜y ‡ells whi™h is the fo™us of our —ttention here
™ites floom(eld9s wenomini morphophonemi™s on the (rst footnote on the
(rst p—geD —nd it is the third of PR items ™ited"the list re—ds for the most p—rt
like — list of the most import—nt p—pers in phonology over the pre™eding two
de™—desF e s™hol—rly look —t the m—jor pu˜li™—tions of the IWRHs —nd IWSHs @to
whi™h this p—per is intended to ˜e — ™ontri˜utionA would h—ve to dr—w two ™onE
™lusionsX (rstD th—t floom(eld9s —n—lysis in wenomini morphophonemi™s w—s
˜oth ™ited —nd in)uenti—lD —nd se™ondD th—t the line of in)uen™e w—s ™ontinuous
from floom(eldD through r—rrisD ‡ellsD —nd ghomskyF „o ˜elieve otherwiseD it

REFERENCES                                                         REFERENCES

seems to meD is to turn — ˜lind eye to the do™ument—ry re™ordF
   s do not dou˜t for — moment th—t someone in ghomsky9s position might
even tod—y look ˜—™k —t ‡ells9s —rti™le from IWRW —nd see it —s pure des™riptionD
l—™king —t le—st some of wh—t emerged in gener—tive phonology ten to (fteen
ye—rs l—terF fut the ™l—im th—t intelle™tu—l ™ontinuity is the norm would expe™t
just th—tX r—lle —nd ghomsky9s work in the l—te IWSHs should indeed ˜e —n
—dv—n™e when judged —g—inst ‡ells9s work in IWRTF fut it ill ˜ehooves us to
dismiss e—rlier work ˜e™—use it f—ils to surp—ss work th—t still lies in the futureF
   xeedless to s—yD s en™our—ge the re—der to re—d ‡ells9s p—per for himselfD —nd
to judge whether it is not — ™—utious —nd ™—reful exegesis of the ˜ene(ts th—t
™—n ˜e re—ped from deriv—tion—l —n—lysisD —imed —t —n —udien™e th—t w—s le—ry of
™onfusing syn™hroni™ —nd di—™hroni™ —n—lysisF es — phonologist working —t the
˜eginning of the PIst ™enturyD s would —rgue th—t we should not ™h—r—™terize the
work of linguists su™h —s ‡ellsD r—rrisD —nd ro™kett —s the l—st g—sp of — dying
stru™tur—lismD ˜ut —s — ˜ody of s™hol—rship out of whi™h gener—tive phonology
w—s — n—tur—l developmentF
   ƒurely this ™on™lusion is re—son—˜le —ndD ultim—telyD not —t —ll surprisingF wy
—dmir—tion for gener—tive phonology is in no w—y diminished ˜y the re—liz—tion
th—t its key ide—s were ˜eing ™onsidered —nd developed ˜y the mid IWRHsF st isD
—fter —llD the ide—s th—t m—tter to us nowF


Anderson, Stephen R.        IWVSF Phonology in the twentieth century : theories
    of rules and theories of representations F ghi™—goX …niversity of ghi™—go

Bloomfield, LeonardF        IWPTF e set of postul—tes for the s™ien™e of l—ngu—geF
    Language PFISQ!ITRF

""F IWQWF wenomini morphophonemi™sF Travaux du Cercle Linguistique de
    Prague VFIHS!IISF

Bromberger, SylvainD       8 Morris    HalleF   IWVWF ‡hy phonology is di'erentF
    Linguistic Inquiry PHFSI!UHF

Chomsky, NoamF       IWVTF Knowledge of language : its nature, origin, and use F
    xew ‰orkX €r—egerF

Encrevé, PierreF      IWWUF v9—n™ien et le nouve—uX quelques rem—rques sur l—
    phonologie et son histoireF Langages IPSF

REFERENCES                                                     REFERENCES

""F PHHHF „he old —nd the new X ƒome rem—rks on phonology —nd its historyF
    Folia Linguistica ˆˆˆs†FST!VRF

Fischer-Jörgensen, E.       IWUSF Trends in Phonological Theory F gopenh—genX
    ek—demisk porl—gF

Goldsmith, John A.       IWWQF r—rmoni™ phonologyF sn The Last Phonological
    Rule D edF ˜y tohn qoldsmithD PPI!PTWF ghi™—goX …niversity of ghi™—go

" IWWWF Phonological Theory: The Essential Readings F yxfordX fl—™kwell

Harris, Zellig S.       IWRRF ‰okuts stru™ture —nd —nd xewm—n9s gr—mm—rF
    International Journal of American Linguistics IHFIWT!PIIF

" IWSIF Methods in Structural Linguistics F ghi™—goX …niversity of ghi™—go

Huddleston, RodneyF         IWUPF „he development of — nonEpro™ess model in
    emeri™—n stru™tur—l lingusiti™sF Lingua QHFQQQ!QVRF

Hymes, DelD   8    John FoughtF    IWVIF American Structuralism F „he r—gueX

Joos, MartinF     IWTRF e ™h—pter of semology in the inglish ver˜F sn Monograph
    Series on Languages and Linguistics D edF ˜y gF sF tF wF ƒtu—rtD SW!UPF

Kilbury, JamesF     IWUTF The Development of Morphophonemic Theory F tohn

Kiparsky, PaulF     IWTVF row —˜str—™t is phonologyc sn Three Dimensions of
    Linguistic Theory D edF ˜y yF pujimur—D I!IQTF „okyoX „—ikush—F

Koerner, E. F. K.       PHHPF Toward a History of American Linguistics F xew
    ‰orkX ‚outledgeF

" PHHQF ‚em—rks on the origins of morphophonemi™s in emeri™—n stru™tur—lE
    ist linguisti™sF Language and Communication PIFI!RQF

"" PHHRF yn –in)uen™e9 in linguisti™ historiogr—phyX worphophonemi™s in
    emeri™—n stru™tur—lismF sn Essays in the History of Linguistics D ™h—pter RF
    tohn fenj—minsF

Kuhn, Thomas S.       IWTPF The Structure of Scientic Revolutions F …niversity
    yf ghi™—go €ressF

REFERENCES                                                  REFERENCES

McCawley, JamesF     IWUWF „he phonologi™—l theory ˜ehind ‡hitney9s ƒ—nskrit
   gr—mm—rF sn Adverbs, Vowels, and Other Objects of Wonder D edF ˜y t—mes
   w™g—wleyD IH!IWF ghi™—goX …niversity of ghi™—go €ressF

Rorty, RichardF   IWWIF Objectivity, Relativism and Truth F g—m˜ridgeX g—mE
   ˜ridge …niversity €ressF

Sommerstein, AlanF     IWURF yn phonot—™ti™—lly motiv—ted rulesF Journal of
   Linguistics IHFUI!WRF

Voegelin, Carl F.D   8   Morris SwadeshF    IWQSF e pro˜lem in phonologi™—l
   —ltern—tionF Language ISFI!IHF

Wells, Rulon S.   IWRWF eutom—ti™ —ltern—tionsF Language PSFWW!IITF


Shared By: