Académique Documents
Professionnel Documents
Culture Documents
Pages43I~4^4
THE LINGUISTIC
CASE FOR THE
AIOLIAN MIGRATION
RECONSIDERED
ABSTRACT
thepresenceofspeakersofLesbianin thenortheast
AegeandurAscribing
from
mainland
times
to
the
of
Aiolian
tribes
Greece
historical
migration
ing
is
not
or
even
from
the
receives
nosupport
Migration
only
primary
linguistics.
wayin whichlanguagesand dialectsmayspread.Moreover,on reexamination,theideaofanAiolicdialectgroupfallsapart.Boiotian,separated
bythe
and
from
Lesbian
FirstCompensatory
Thessalian,appearsas
Lengthening
a conservative
dialect,mostcloselyrelatedto WestGreek.In turn,Lesbian
ofGreekthatsharenodemonstrable
andThessalianarebotharchaicbranches
best
viewed
as twoseparaterelicareasofa
are
commoninnovations.
They
Greek.
unaltered
early
relatively
world- an aggregateof streamsdisTo pass fromthislegendary
comeintoconfluwhichdo notwillingly
tinctand heterogeneous,
intermix
into
the
visionafforded
forced
to
cannot
be
and
ence,
in
the
500s
B.C.thewhole
byHerodotus,we learnfromhimthat
of
fromDardanussouthwardto thepromontory
coast-region
Lektum(includingthetownofIlium),and fromLektumeastward
had beenAeolised,orwas occupiedbyAeolic
to Adramyttium,
Greeks likewisetheinlandtownsof Skepsisand Kebren.1
This papergrewout of theresearchof BrianRose, as setout in theprecedingarticlein thisissueofHesperia?As head of thepost-BronzeAge
excavationsat Troy,Rose had long acceptedthe scholarlyconsensusin
1. Grote 1888, vol. 1, p. 305, referringto Hdt. 1.149-151.
2. Rose 2008. My thanksare due to
Brian Rose, Don Ringe, Ronald Kim,
and the anonymousreviewersforHesperia fordiscussionand clarifications.
Certain conventionsand abbreviationscommon in historicallinguistics
*
are used in this article.An asterisk
marksan unattestedor reconstructed
form.A daggert marksa formthatwe
The American School of Classical
432
HOLT N. PARKER
THE TRIPOD
The theoryof an Aiolian migrationrestson threelegs: archaeological,
historical,and linguistic(Fig. 1). For the first,Rose concludes:"At no
timeduringthe early1st millenniumdo we have evidenceforattacks,
forthe arrivalof a new populationgroup,or foranysubstantive
change
in ceramicproduction."3
Otherscholarshavenoted,ifnotbeen bothered
by,the lack of archaeologicalevidence.Gschnitzer,forexample,writes:
"The migration
toAsia Minor,whichwe couldassumewas a consequence
of the driveforcolonization,has apparentlynot yet been successfully
dated archaeologically;
it musthave occurredbeforethe corresponding,
but equallyundated,migrationof the Ionians."4Coldstreamexpressesa
similarview:"These Aeolians,accordingto literary
sources,had migrated
fromtheirformerhomesin Boeotia and Thessalyat leastas earlyas the
ofIonians;yetthearchaeological
recordcastsverylittle
parallelmovement
on
them
before
the
late
...
At
light
eighthcentury.
presentwe have no
reliablearchaeological
evidenceconcerning
thecomingofthefirst
Aeolians
to Lesbos."5
Nordo theconflicting
accountsofthecolonizationofAiolis
legendary
such
accountshavebeenacceptedas sosupplyconvincing
support.
Though
berhistory
accounts
bysome,Rosehasshownhowtheseandothermythical
developedand wereadaptedin the courseof the 6th-5thcenturiesB.C.
fora varietyofculturalpurposes.6
In particular,
thegenealogicalaccounts
are merelyattemptsto connectlocal aristocracies
to the royalfamilyof
or
a
father
Mycenae
putativeancestorAiolos,who servesas a convenient
Kw > T/_ e readsas "a labioexample,
velarbecomesthecorresponding
dental
in theenvironment
beforee."Language
abbreviations:
Ark.= Arkadian;
Att.Ion. = Attic-Ionic;
Boiot.= Boiotian;
Cret.= Cretan;Cyp.= Cyprian;Dor. =
Doric;Eng.= English;Germ.= German;Horn.= Homeric;Lak. = Lakonian;Lesb.= Lesbian;Myc.= MycePGrk=
naean;Pamph.= Pamphylian;
PIE = Proto-Indo-EuroProto-Greek;
pean;SGrk= SouthGreek;Skt.=
Thes. = Thessalian.Common
Sanskrit;
abbreviations
ofgrammatical
terms
are
alsoused.
nom.,
ace,
(e.g.,
sing.,pl.)
3. Rose2008,p. 420.
4. Gschnitzer
2002,cols.227-228.
I failto followtheauthors use of"must
have"here.He giveshisreasons:"as
theirterritory
to the
theyextended
northto thecoastoftheA[eolians]
Aeolian;Phocaea
(Smyrnapreviously
on theedgeoftheAeolianhinterland;
Aeoliansubstrate
in thenorthern
Ioniandialects)."
I am notsureI see
howanyoftheseindicatespriority
of
settlement.
The linguistic
evidenceis
uselesssincemostscholarsexplainvariousfeatures
ofLesbianbyrecourse
to
Ionicinfluence
on it (see below).
5. Coldstream
2003,pp.262-263.
6. Rose2008,pp.401-404.Hammond(1975) usesThucydides'
account
to producedatessupposedly
accurate
towithina decade:TrojanWar,ca.
1200 B.C.;Thessalianinvasion,
ca. 1140;
Aiolicmigration
to Lesbos,ca. 1130;
Dorianinvasion,
ca. 112O.Thessalians
andBoiotiansareimaginedas a partof
an invading
Greeks,
groupofNorthwest
in
Greek
originating Epiros.Northwest
andDoric aretreatedcorrectly
as subfamilies
ofWestGreek,which,however,Hammond(1975,p. 703) holdsto
be closelyrelatedtoMycenaean.
THE LINGUISTIC
Figure1. Distributionoftraditional
dialectgroupsin theeasternMedi-
terranean. AfterHainsworth
1982,p. 859,
map28
7. RE 1,1893,col. 1031,s.v.Aioles
(2002,col.228)
(E. Meyer).Gschnitzer
of
the
out,
points "Regarding statement
thattheA[eoofantiquity
thescholars
lians]hadoncesettledin certainareas
ofGreece,thereis,in contrast
[toThuc.
3.102.5],notmuchto go on.This is
becauseon theone handtheyareconofthe
nectedwiththeappearance
433
fortheunaffiliated,
andcannotbe usedto inferactualtribal,
gefigure
affiliations.
summed
or
netic, linguistic
Meyerin 1893rightly
up:"Oneis
toconclude
thatallGreektribes
thatwerenotDoricorIonicwere
forced
AioHc.'"7
designated
is gently
corrected
Unduepietytoward
theclassicalsources
byCook:
whichalonegivesa precise
carries
no
"Theconnexion
withOrestes,
dating,
.
.
.
As hefurther
"the
Greek
had
a
conviction."
horror
notes,
antiquarians
intoa chronological
likethisweredulytranslated
vacui Stories
system."
of
"Theschematic
traditions
the
concludes:
He sensibly
migrations
prose
to theEastAegeanaftertheTrojanWarseemin generalto havebeen
B.C."8
Underthatclearlight,
ofthefifth
manypretty
century
compilations
thepathsandeventheexactdatesofa complex
butfanciful
mapsshowing
mustdisappear.9
andsackings
seriesofmigrations,
invasions,
ofthis
us in thepassagequotedat thebeginning
As Grotereminds
there
is
and
such
after mythological
article,
history reconstructions, only
a singlefact.As attested
bySapphoandAlkaios,ataround600 B.C. we
mythological
personalnameAeolus,
theearliestbearerofwhichwasprobas theproablysecondarily
interpreted
genitoroftheA[eolians],andon the
otherhandwiththetheories
concerning
thedivisionoftheGreeksintoa few
thenameA[eolilargetribeswhereby
thatcouldnotbe
an] coveredeverything
ascribedto theDoriansortheIonians."
HOLT
434
N. PARKER
can pushGrotes dateback a bit- peopleon Lesbos werespeakinga distinctiveGreekdialect,one thatmodernlinguistsconsiderto be relatedto
thedialectsofBoiotiaandThessaly(Fig. l).The onlysurviving
leg ofthe
tripod,then,is theidea thatinThessalianand Boiotianwe havea case of
a unifieddialectarea splitbylatecomers,
in thiscase the Doric speakers
of NorthwestGreek,withLesbian as an outlyingprovince.10
That is the
of
this
subject
paper.
THE LINGUISTIC
435
HOLT N. PARKER
436
oftheDark Ages as
withtheRomanEmpireand the Volkerwanderungen
Historical
of
underlies
most
reconstructions
implicitmodels,
prehistory.
and his
Max
Miiller
has longdisplayeda prejudice,evenbefore
linguistics
invasionas itsprimary
model.19
Aryaninvasion,formilitary
In an importantreview,Nichols distinguishes
threemechanismsby
whichlanguagesspread:languageshift,demographicexpansion,and miin thenumberofmobilepeople
The mechanismsdiffer
gration.
primarily
involvedin thelanguagespread:
in responseto thepresenceof at leasta
Languageshiftis normally
fewinfluential
immigrants;
demographic
expansioninvolvessome
and
of
rather
than
extermination;
absorption previouspopulation
leads
to
shift
to
or
from
the
(either
migration
language
immigrants'
and
language).The termslanguageshifty
demographic
expansion,
that
refer
to
the
contributor
with
no
claim
predominant
migration
it is exclusive.Almostall literature
on languagespreadsassumes,at
leastimplicitly,
eitherdemographic
as [the]
expansionor migration
basicmechanism,
butin factlanguageshiftis themostconservativeassumptionand shouldbe thedefaultassumption.
There is no
reasonto believethatthemechanismof spreadhas anyimpacton
thelinguistic
ofthespread.20
geography
This lastremarkcan be turnedaround:equally,linguisticgeographycan
offerno information
on themechanismoflanguagespread.
Though a nucleusof people speakinga formof Greek presumably
cameto Lesbos fromsomewhere,
at sometime,thenumbersneednothave
beenlargeand theformsoflanguagespreadaremorevariousthansimply
one populationdisplacinganother.We can use Nichols'soutlineto make
a fewimportant
points.
Languageshift.Languages and dialectscan spreadinto a new area
withoutanymajorchangein thepopulation.Languagesmaybe replaced
in thecourseofonlya fewgenerations
ofprestigedialects
bytheinfluence
andlanguages,
interritorial
boundaries(so theslowlossofBalinese
byshifts
to Bhasa Indonesia),byexchangenetworks
(so therecentspreadofHausa
or theancientspreadofAramaic),orbyextensive
bilingualism in short,
byculturalratherthanphysicalimperialism.21
19. See ChapmanandHamerow
1997andChapman1997forgeneral
accountsoftherise,fall,andriseof
andinvasionas preferred
migration
modelsin archaeology.
Forlinguistic
see especially
Olender1992.
history,
Forthetroubled
oftheAryan
history
invasionhypothesis,
see Bryant
2001.
Warriors
benton conqueststillseem
thebestanswerto Schlerath
(1981,
(2003,p. 68).
p. 199) andMeier-Briigger
20. Nichols1997,p. 372. Bellwood
(2005,p. 191) claimsthat"historical
dataindicatethatlanguageshiftalone,
without
movement
orsome
population
degreeofdispersal
bythepopulation
thetargetlanguage,
hasnever
carrying
createdanything
remotely
equaling
thosevastintercontinental
genetic
oflanguageswithwhichwe
groupings
arehereconcerned.
. . . Imperialconwithout
and
itself,
questby
large-scale
settlement
of
permanent
bymembers
theconquering
population,
generally
imposeslittleapartfromloanwordsin
thelongterm.Tradealsois generally
of
littlesignificance
as a factor
behind
Whether
large-scale
languagespread."
theseobservations
areapplicableto the
spreadofGreekora dialectofGreek
acrosstheAegeanis uncertain.
His own
data,in anycase,do notsupportthis
He citesthelossofLatin
statement.
"in
those
except
regionscloseto the
heartoftheempirefavored
forintensivesettlement
byLatinspeakers"
was not"lost,"
(p. 192). Latin,however,
itsimplychanged,andthe"heartofthe
fromPortugal
to
extended
empire"
Romania.
21. See Renfrew
1987,pp. 120-144,
foran accountdirected
at archaeologists.Essentialreadingarethetwoliterature
reviews
ofNichols1997
(thoughI putno trustin glottochronology)andBellwood2001 (whois skepticalofall factors
see
exceptmigration;
n.20, above).Foran excellent
account
THE LINGUISTIC
437
Changesin theactualpopulationsofspeakers
Demographic
expansion.
occur
without
mass
Dixon pointsto a numberof known
migration.
may
reasonsforone groupof speakersto fadeor flourish:naturalcauses,such
as drought,flood,earthquake,disease;materialinnovations,
such as new
in
tools,weapons,meansof transportation,
techniques productionor in
social
such
as
state
causes,
formation,
agriculture;
changesin hierarchy,
and especiallyreligion;and linguistic
revolution,
factors,such as prestige
and complexity.22
Migration.Evenwhenlanguagechangeis associatedwithpopulation
change,invasionis nottheonlyoption:a new groupof speakersmayaror slowinfiltration,
and each processmightor
rivebypeacefulmigration
in
record.
thenewarrivals
not
leave
traces
the
Further,
archaeological
might
in
or
drive
out
former
live
contained
settlements,
inhabitants,
any
might
In
invasions
can
and
do
sometimes
and
occur,
short,
migrations
intermarry.23
sometimes
withlittle
effects
on thepreexisting
withcatastrophic
languages,
orinvasionin theabsenceofclear
effect
at all,butto presupposemigration
error.24
historicalor archaeologicaldata is a methodological
about
the spreadof
we
have
been
Furthermore,
talkingprimarily
language.Dialectswithina languageare muchmorelikelyto developby
thespreadofinnovations(isoglosses)thanbypopulationchange.Dialect
contactcan,ofcourse,occurbymigration(witness,forexample,thesudden arrivalofAnatolianGreekin themainlandfollowingthepopulation
exchangesofthe1920s),butchangecomesmoreoftenin theformofdifin speechfromlargerpopulationcentersto smaller
fusionofan innovation
onesand thenceto moreruralareas(theso-calledgravity
model),withthe
"the
extentand rateof changeheavilyinfluenced
by
phenomenonitself,
communication
distance,time,and socialstructure/'25
networks,
the presenceof a feature
In earlierperiodsof historicallinguistics,
of one dialectin anotherwas usuallyexplainedby
deemedcharacteristic
some40
However,as Cowgillwroteoptimistically
populationmovements.
dialect
canspreadacrossexisting
thatinnovations
yearsago,"therealization
boundarieshas led to sobererviewsofprehistoric
migrations."26
distinoflanguagespreadinAfrica,
trade(whichcanbe accomguishing
"trade
plishedbypidginsandsimplified
a
cultural
vs.
"generalized
languages")
see Nettle1996,esp.p. 412.
exchange,"
ofprestige
Fortheinfluence
dialects,
see Dixon1997,pp.22-25,79-80,
104-114,145-148.
22. Dixon 1997,pp.22-24,75-83.
ForthespreadofArabic,seeVersteegh
1997,esp.pp. 71-72,93-98,102-113.
a good
23.The Normansoffer
exampleofa groupwhocompletely
replacedtheirlanguage(fromOld
withinperNorseto Gallo-Romance)
with
two
generations, no discernhaps
record.
ibletracesin thearchaeological
24. Rhodes(2006) studiedtheinteractionsofCreeandOjibwe-Potawatomiin theGreatLakesregionover
500 years,notingthat"every
logically
possibletypeoflanguagespreadis
in thisarea,allbutone inattested
In this
volvingonlyhunter-gatherers.
intounoccupied
areawe findmigration
withone population
migration
territory,
and
thena number
another,
replacing
ofspreadswithintermingling
populato
tionsbothwithmigrators
switching
andwithmigrators'
thelocallanguage,
thelocallanlanguagesupplanting
of
thereareinstances
guage,andfinally
languagesmovingwithoutmigration,
i.e.,languageshift."
Modernconsensusseemsto favora
model,explicequilibrium"
"punctuated
itlybasedon Goulds modelofevolutionandspeciation
(e.g.,Dixon 1997,
andDixon2001,
Aikenvald
67-86;
pp.
and
See
9-11).
Janda
Joseph2003,
pp.
history.
pp.50-58,fortheintellectual
Watkins(2001,pp.48-49),however,
to Indoquestionsitsapplicability
European:"To speakofpunctuation
theissuerather
by'invasion'
prejudges
of
the
severely; Indo-Europeanization
Italyandmanyotherareasseemsto
andin
havetakenplacebothgradually
a woefullabChinaprovides
driblets."
andinvaofmassmigrations
oratory
times:thelinguistic
sionsin historical
effects
aresurveyed
byLaPolla2001.
and Schilling-Estes
25. Wolfram
2003,p. 727; see theentirearticlefor
Hock andJoseph
overview.
an excellent
(1996,pp.346-365) setoutthedetails
ofdialectgeography
andproblems
"Aiolic"as one oftheir
and
use
clearly,
examples.
26. Cowgill1966,p. 78.
HOLT
438
N. PARKER
THE LINGUISTIC
439
30.Chadwick
1973,p.255.
440
HOLT N. PARKER
See thestrictures
ofRix
proximity.
(1994,pp. 18-19). Fora detailedcriticismoftheexplanatory
powerofsubstratain thiscontext,
see Hock and
Joseph1996,pp.382-387;a pithy
critiquebyTrask(2000,p. 329); anda
casestudyofCastilianinTrask1997,
pp.415-429.
32.This is a persistent
bad habit.
See,e.g.,Risch1955,p. 75,witha chart
of20 randomisoglosses;
Wyatt1970,
witha different
20 plus29 othersubrulesto arriveat a final25; Finkelberg
setof20; and
1994,fora different
2005,pp. 115-117,witha
Finkelberg
revisedlistof21. Coleman(1963) tops
thelistwith51 features
to a
subjected
correlation
coefficient
Each
analysis.
authorcombinesphonological,
derivaandlexicaldiffertional,inflectional,
ences,andfailsto distinguish
rigorously
betweeninnovations,
selecretentions,
tions,andindependent
parallelchanges.
33. See,e.g.,Wyatt1970,p. 627;
MendezDosuna 2007,p. 460.
34. Sihler,2000,p. 173:"Thereis
aboutthis:innovanothing
mystical
tionsin theisolatedgroupwillbe limitedto thosethatariselocally.Bycona groupin contactwithothersimtrast,
ilarformsofspeechwillbothmakeits
owninnovations
andbe influenced
by
elsewhere."
Retenchangesoriginating
tionis not,ofcourse,a necessary
feature
ofcolonies;thepointis merely
that
neither
is "progress."
The phrase"coloniallag"coinedbyMarckwardt
(1958,
beenmisapplied;
p. 80) hassometimes
see Gorlach1987;Trudgill
1999.
35. See Rodrfgues
Adrados1952for
a clearexplanation;
also Karali2007,
389.
p.
THE
LINGUISTIC
CASE
FOR
THE
AIOLIAN
MIGRATION
441
Shared archaismscannotpoint to
innovationsshow any relationship.36
selections
froma commonstock.The point
nor
can
different
groupings
can be made clearby a biologicalanalogy.Both fishand catsretaintheir
humansdo not.That is not,however,
a
tailsfromtheancestralcreatures;
ratherthancatsand
reasonto classfishand cats(who sharetails)together
thatdifofbeingmammals).Further,
humans(who sharetheinnovations
to
is
cannot
a
reliable
selections
guide grouping perhapsone
provide
fering
To takea linguistic
ofdialectology.
ofthemostoftenoverlookedprinciples
words
for
'one':
theregular
has
two
*sem->
example,Proto-Indo-European
ordinaland *oy-probablymeaning'alone' or the like.Greek,Tocharian,
*sem-(Greekeiq,TocharianB seyArmenianmek;
and Armenianinherited
>
Sanskrit*dy-kos
see below).All otherschose *dy-withvarioussuffixes:
> Old Irishden,
eka;*6y-wos> Avestanaeuuo,Old Persianaiva; *dy-nos
Latin unus,Gothic ainsyEnglishone.But Greek and Tocharianare not
sisterlanguagesanymorethanareLatinand English.Choice fromamong
tellsus nothing.37
alternatives
Second,fora sharedinnovationto provideanyevidence,theinnovationmustbe trulyshared,thatis, it mustbe genetic.Again,a biological
examplecan be provided:both birdsand bats have wings.But a bird's
innovationthana bats.Justso in dialects
different
wingsarea completely
similarfeaturesare oftendue notto thespread
or languages,remarkably
innovaofan isogloss,thatis,bycontact,butbyparallelbutindependent
tionswithineach branch.
The Second CompensatoryLengtheningin Greekis a good examLengthple(Fig.2). Final-ns(andso notsubjecttotheFirstCompensatory
ening,see below),forexample,in accusativepluraltons,stayedas xovq
beforevowels,butdevelopedto xo<;beforeconsonants.Older Cretan(the
butthexoqforms
GortynLaw Code, forexample)keepsthisdistribution,
aregeneralizedin Thessalian(but not Lesbian or Boiotian)as well as in
Arkadian,and withinDoric in Theran, Kyrenaian,and Koan, but not
elsewhere.All theotherdialectsgeneralizethetov<;form,whichis either
36. So Wyatt(1970,pp.560-561):
"In thissearchonlysharedlinguistic
innovations
can,thoughtheyneednot,
value.Sharedretenhaveevidential
is notevidence
tionofan archaism
Hainsworth
whatsoever."
(1982,p. 857)
a weak
is willingto grantconservation
as of
value:"Puttogether
evidentiary
of
manner
all
were
equalimportance (a)
ofcertain
innovation,
(b) generalization
at theexpenseofothfeatures
inherited
These must
ers,and(c) conservation.
orderof
as ofdescending
be regarded
innovation
shared
And
importance.
ofgeneticrelation
is indicative
onlyif
of
to thegeneralpattern
itconforms
isoglosses."
37.This problemstillbedevils
ProtoForexample,
Greekdialectology.
ofwaysto say'if:
Greekhad a variety
The first,
ei,is thelocativeof
ei,ai, r\.
*e-/othepronominal
(cf.elxa):so
'when/
'where,'
temporally
spatially
of
'if; ai is thefeminine
conditionally
theold
thesame;r\'where/when
Adrados1952,
instrumental
(Rodrfgues
1992,vol.1,
pp.31-32; Meier-Bnigger
1999,p. 316,
p. 67,E 305; Chantraine
thesemantic
s.v.ei); all haveroughly
do this'-> 'if
'when
you
development
distribution
youdo this.'The differing
ofArt.-Ion.andArk.ei,Aiol.andDor.
ai, andCyp.T|hasbeentakenas an
ifdisturbing,
isogloss
important,
(Rischsno.5 [1955,p. 75,chart];but
whydoesAiolicagreewithDoric?
WhydoesCypriannotagreewith
inherited
Yet,Attic-Ionic
Arkadian?).
all three:ei themostgeneralized;
r\
1939withav > ea"v(rightly
Schwyzer
442
HOLT N. PARKER
Figure2. ChartoftheSecond
CompensatoryLengthening
1982,p. 862:
weight.So Hainsworth
"Aeolicappearsas a medleyofWest
andEast Greek.... It is thusan early
exampleofa bridgedialect."GarciaRamon2002b,col. 1016:"TrotoAeolic/whichsprangup inThessaly
andhad East-and,in particular,
West-Greek
underwent
a
features,
seriesofprobably
post-Mycenaean
changesthereandthenspreadto
Boeotia(around1250) andAeolis
(around1000)."
THE LINGUISTIC
443
40. Ringe (pers. comm.) has performeda similarstudyon ArkadoCyprianand foundno secureshared
innovations.South Greek appears to
into fourdifferent
have split,therefore,
dialects:Myc, Ark.,Cyp., and Att.Ion. Wyatt (1970, p. 627) also doubts
the existenceof an Aiolic family.He,
however,sees the dialectsin purelysociological terms:"Indeed, we nevercan
referto Ae[olic] as a whole, and have
insteadalwaysto thinkof L[esbian]
T[hessalian] B[oiotian] as separateentities,L[esbian] a low-class P[roto-]
G[reek] dialectwhich latermoved into
the innovatingsphereof G [reek];
T[hessalian] a low-class P[roto-]
G[reek] dialectwhich remainedout of
touchwith the restof the G[reek]
world fromthe veryearliesttimes;
B[oiotian], a low-class P[roto-]G[reek]
dialectwhich affectedthe conservatism
of N[orth-]W[est]
characteristic
G[reek]." See also Wyatt 1973, p. 43.
41. Cowgill 1966; Schmitt1977.
HOLT N. PARKER
444
Some standardexamplesare:
> xoaaoq; but SouthGreek
(cf.Lat. tot< tot'i)> *totsos
*tdty-o-s
to-so-de
(xoaov8e).
togo<;,Mycenaean
> hdtsos
(Cretanoo<;,spelling-ts-\lateroxxoq)> Lesbian
*Hydty-os
ooooq; butAtticand Arkadianoaoq.
> *kwdtsos
> Lesbiannooooc,(versusBoiotian6-noixoq,
*kwdty-o-s
Cretano-rcoxxoq);
but SouthGreeknocoq.
> *metyos
>
Lat.
(cf.
medius)> Proto-Greek*methyos
*medh-yo-s
>
*metsosLesbian ueaaoq (versusBoiotianand Cretanjxexxoq);
but SouthGreekueooq,cf.Mycenaeanme-sa-ta'middlequality*
= ueo(o)ccxo<;.
With original*-ts-:*pod-si> Proto-Greek*potsi> noaci but South
GreekkogL
With original-ss-:Proto-Greek*genes-si> yeveaai (theanalogical
sourcethenofthethird-declension
dativepluralin -eooi) but
SouthGreekyeveai.
manner,sucha group
Despite ourtendencyto thinkin a bifurcating
oftheotherdialectsto each other.
impliesnothingabouttherelationships
A SouthGreekdialectdoes notimplya unitary
"NorthGreek"dialector
other
situation.46
any
AlOLIC
WITHIN
THE GREEK
DlALECTS
1. labialreflexes
ofkwevenbeforee
2.*r>po/op
Morphological
3. dativepluralin -eaai
4. perfectparticiplein -cov,-ovxLexical/derivational
5. i'oc= uia
6. patronymic
adjectivein -loq
THE LINGUISTIC
445
fromthe glosses.
53. Lejeune 1972, pp. 50-52, 3840.
54. Later Arkadian shows the dental
in all formsof ogtk;,as well as Ttevxe,
eo-diXX-ovzec,
(pdMxo), odekoq. See
Schmitt1977, p. 86. This is due to
influencefromotherdialects(see, e.g.,
Buck 1955, pp. 174-175), thoughit
may also representa regulardevelopmentof the new *ts> t>*dz > d.
HOLT
446
N. PARKER
otuPpoToq)
4. euuopuevov(Alk. 39a.7) < *se-smr-men-on
(cf.Att.eiuocpuevov)
5. Pp68ov,and compounds(Sappho 2.6, 53.1, 55.2, 94.13) a borrowingfromIndo-Iranian*wrd6. oxpoxov(Sappho 16.1; Alk. 372.1, 382.2), and oxpoxocyoi
(7GXII.2 5, etc.);propernames<*str-toThe firstfiveall showa labialenvironment,
butaxpoxovis good evidence
fora regulardevelopment
of*r> po/op.57
The evidenceforBoiotianand Thessalianis verythin.Boiotianhas
andIxpox-in propernames,whichseems
onlytwoforms:e-aoxpoxet>-a9r|
tobe good evidencefora regularchangeof *r> po.The otherformis found
in thepropernameswiththerootBpo%-x>XXo<;>
etc.,thoughpropernames
arealwaysuncertain.
There arealso onlytwoformscitedforThessalian,and onlyforEast
Thessalian at that,which shows Bpo^ix;as a propername and rcexpoand so may
exeipi8a,'periodoffouryears/Botharein a labialenvironment
indicatea conditionedreflex,
thoughwe cannotbe sure,and it is doubtful
thatTcexporeallycontinuesan unalteredzero-grade.58
55. The changes in Arkadian and
Cyprianpoint to the latter.
56. For basic accounts,see Lejeune
1972, pp. 195-198, 199-202; Sihler
1995, pp. 92-96, 95-98. The original
conditionsdetermining*r> po/op
probablydepended on root shape and
accent,but the patternhas been so
disturbedby paradigmaticand analogical levelingin both directionsthatwe
cannot recoverthe rules.Further,there
appear to be no good cases of */in
Lesbian, Boiotian, or Thessalian.
57. The othercases cited by Hamm
etc.). The
ypdrcxa,
epigraphicypdqniv,
formscited fromthe grammariansare
withoutidentification,
even if their
etymologieswere certain.So Hsch.
k 3669 (Latte): Kopxepd-Kpaxepd,
ia%upd;\i.1679 uopvd|ievo<;uaxouevo<;.The sometimes-cited8poaeco<;
and 7tTopuo<;
are foundonlyin the third
anonymousCompendiumKepiSiaXeKxcovattributedto JohannesGrammaticus (ed. Hoffmann1891-1898, vol. 2,
pp. 215, 221).
58. Further,althoughJiexpomight
continuedirectlya *kwetwr-,
the
THE LINGUISTIC
447
63. Finkelberg
(1994,p. 19; 2005,
pp. 129-130),themostrecentto revive
ofPamtheidea,makesno mention
orWyatt's1973
Kyrenaian,
phylian,
the
demolishes
whicheffectively
article,
notion.
64. See MorpurgoDavies 1976for
themechanism.
Wyatt1973,p. 39:
"Thereis no needto assumean Aeolic
in orderto explainthespread
substrate
ofthe
ofthishighlyusefulallomorph
dativeplural."Ruijgh(1996,pp.486from
487) seesitas a simpleborrowing
withCorintheneighboring
dialects,
itsreplacement.
thianlaterreplacing
65. Lesb.:masc.taA,d6-cov,
neq>x>yyetc.;fern.
cov,yevov-ovxa,
jcap-eaxaKThes.:masc.
oiaav; inf.xeOvaK-riv.
erc-eoxdic-ovxa,
etc.;inf.eaA,oi)K-e-uev.
dn-eiA,8e{Boiot.:masc.dv-xe-9e(-ovxa,
inf.
attested.
etc.
No
ovxec;,
pf.
66. Chantraine
1961,pp. 184-185,
211,p. 278, 329.
HOLT
448
N. PARKER
in variousformsofDoric
We findsimilarperfectthematicinfinitives
East
Island
Doric:
(Crete,
Kos, Nisyros).There is an
Argolis,Phokis,
at
isolatedthematicperfect
participle Kyrene,and Delphi showsa similar
butconfined
formsintheinfinitive,
andin theparticiple
spreadofthematic
tothefeminine.68
but
This is thestrongest
evidencefora sharedinnovation,
thefactthatotherdialectssuccumbto thetemptation
of thematicforms
fortheperfect
weakensthecase somewhat.69
m = jiia. What thehandbooksusuallypresentas thesole lexicalitem
Aiolic,ia intheplaceofuiocfoundin otherdialects,is theresultof
defining
paradigmatic
leveling.There arethreepointsto bearin mindabout'one':
1. Though it is sometimesnotclearfromthehandbooks,we havea
fullparadigmof elq,10c,
ev (to givetheLesbian forms).That is,
thefeatureis notreallya lexicalitemat all,buta morphological
one.70
2. A modelfromwithina paradigmof 1110c,
etc.,is hardto
jLLiaq,
GarciaRamon
and
for
justify.
Ruijgh, example,thoughtthat
theu- of uiochad beenlostsomehowto bringtheparadigm
intoalignment
withthevowelinitialeiq and ev.But thereis no
proportional
analogicalmodel;Atticand otherdialectshaveno
with
theirregular
to see
problem
paradigm;and it is difficult
whyan attemptto regularizethepatternwouldstopthereand
notproduce,forexample,elq,tea, ev.71
- thatwe havethefeminineof
3. The othercommonexplanation
an ioq meaning'thatone/foundin theGortynLaw Code and
a fewotherplaces- willnotdo.72The formis foundonlyin the
feminine(no fioq attestedforLesbian,etc.),and it is clearlynot
thematic(no fia, ti&v).73
The answer,therefore,
is phonologicaland thebasicoutlinewas given
Schmidt.74
We
are
longagoby
dealingwitha paradigmthathasbeenexten> 6\ioq)
*som-d-s
sivelyremodeled.The PIE root*sem(as in Latin semely
had an originalrootnounparadigmwithmasculine*sem(extendedgrade)
recharacterized
as *sems,75
and a proterokinetic
devi feminine,
withfull67. Hodot 1990, p. 159: "Au parfait,
de la finale
Tadoptionpour l'infinitif
'thematique'-tivest correlativede remploi du suffixe-ovx-pour le participe."
68. Thumb and Kieckers 1932,
pp. 166, 181, 202, 275-276; Buck 1955,
p. 199, 147a; Schmitt1977, p. 48.
One needs to be preciseabout the dialects in which finiteformsof the perfectare inflectedlike the present,and
those in which the infiniteformsare
inflectedlike thematics(presentsand
aorists).
69. There is a similarspread of the
THE LINGUISTIC
449
Proto-Greek
>
*sm-ih2
*sm-yeh2s
>*syeh2s
*sm-yeh2ey
>*syeh2ey
>
*sm-ih2-m
*smiya
> *iyds
*syds
> *iydy
*sydy
*smiyan
Pan-Greek
uia
> Horn.if|q
i<x<;
ia > ip
jiiav
These
and a zero-gradegenitive*sm-yeh2s.
grade nominative*sem-ih2
wereremodeled(withthe zero-gradeof the rootgeneralized)as *sm-ih2
vocalizedto *smiya(the usual deThe nominative*sm-ih2
and *sm-yeh2s.
*hmiato *(m)miatobecome
velopmentin Greek),and thispassedthrough
The obliquecases,however,
the familiaru(oc(see belowfor*sm> mm).76
was simplified
ranintotrouble.The initialconsonantclusterof *smyeh2s
not
to
Siever
s Law),77
so
PIE
sound
to
subject
change *syehs(and
by a
The intervocalic
whichthenbecameGreek*syds.
developmentof -sy-is
>
We expectan initial
to -yy-(*nas-yo-> *nayyo,
xoTo,
etc.).
vocico,
*tosyo
the
vocalize
which
would
to
(following usualright-to-left
*syds give*yydsy
of PIE *sem-ih2
The
etc.78
iterative
rule)as theattestedi&<;,
development
is summarizedinTable 1.
Each dialectthenregularizedthe paradigmof uia, iaq, ia (and so
in
on) itsownway.Most generalizedtheformofthenominative,
creating
jiia, uxaq,butLesbian,Thessalian,and Boiotiangeneralizedtheoblique,
The
Lengthening).
Compensatory
masc.andneut.,too,haveundergone
Myc.showstheorileveling.
paradigm
the
butall otherdia-mof
root,
ginal
lectshavespreadthe-n-ofthenom.
> *he-> *sem-ds
andneut.So gen.*sme's
mos(seeninMyc.e-me= hemeidat.)->
bends(evoq);neut.nom.ace.*sem> ev.
afterCRy
76. Forthedevelopment
cf.*trih2>
(Skt.patnt)>
xpia;*potn-ih2
> e-Tipia-xo
(cf.Skt.
Tcoxvia,
*e-kwih2n. 80;
See
Peters
132,
1980,
kri-td-).
p.
is
best
This
162.
2004,
p.
Gippert
viewedas thenormaldevelopment
to Siever'sLaw; see Schindaccording
ler1977,p. 57; Peters1980,pp. 127132,esp.p. 132,n. 80; Ringe2006,
showsthesame
p. 16.Armenian
> mi'one'(Darms
changein *smiya
Peters
1980,p. 132,n. 80).
1976,p. 13;
77. ForSiever'sLaw,see n. 76,
above.A similarlossofnasalis seenin
'stone'(collective
'sharp')but
*h2e'km6
>
dsmanVedic
(stem),
gen.*h2k-mn-es
with
ds'nas
161,
2004,
p.
(gen.)(Gippert
in
versusthetreatment
references),
> Semxoq.
> 8eica,*dekmt-o-s
*dekmt
This lawhelpsaccountfora puzzling
ofthepronouns:
detailoftheinflection
butdat.*tosm-ey/
nom.*so/*seh2/*tod>
As Ringe(2006,
*tosy-eh2-ey/*t6sm-ey.
has
beensuspected
notes:
"It
55)
long
p.
thatthe-sm-ofthemasc.andneut.sg.
is a reducedformof 'one.'. . . If thatis
true,itshouldfollowthatthesyeh2thecorreofthefern.
sg.formsreflects
thefact
ofthenumeral;
spondingfern,
*-m-hasbeen
thattheroot-final
thansyllabified
might
droppedrather
an earlierpre-PIEphonothenreflect
(inwhichcasethisinlogicalsystem
wouldbe veryarchaic),orthe
flection
clustermightsimplyhavebeenreduced
Schmidt(1898,
byallegrophonology."
tietheformant
p. 399) did notdirectly
demto thewordfor'one,'butcorrectly
See also Szethephonology.
onstrated
1996,p. 206; Gippert2004,
merenyi
esp.pp. 156,161,nn.6, 22,25; Hackstein2005,p. 178.
78.That is,thereis no needto invokeLesbianpsilosisto explainthe
see Lejeune1972,
form.For-sy-> -yy-,
of
pp. 132-133,127. A development
>
ruled
out
seems
to
*hiby
*yy- *yithefactthatitviolatestheusualvoca-
lizationrule(clearlystatedbyRinge
a hypotheti2006,pp. 15-16). Further,
wouldhaveto be createdfairly
cal *yids
latewithinGreekitselfto escapethe
of*y-> C,-(theunusualdevelopment
conditioned
outcome),sinceh-develis
from
ops only
*Hy-.This difference
in
orincorporated
notfullyunderstood
Rix1992,one ofthe
mosttreatments.
unfortumostup-to-date
handbooks,
reversed
the
situation
has
nately
(pp.60, 70, 68, 80); see Peters1976
fordetails.PIE contains(as farI can
find)onlyone lexemewith*sy-:*syuH'sewn,'syU-ma
'sew,'Skt.syu-td'strap.'
withrough
In Greekwe havex>\ri\v
butall Greekwordswith
breathing,
initiald haveroughbreathing
(thereabutitmaybe a regular
sonis unclear,
change;Lejeune1972,
phonological
pp.280-281,320; Sihler1995,
p. 173). HereI suspecta lossofy simiandLat.
larto thatseenin Vedicsutray
subula'awl',thatis,a regular
suo>sutusy
(loss)beforeu in Greek.
development
Schmidt(1898) explainedtheinitial
he
vowelbythetypeofepenthesis
> *iyds.
>
so
for
io9i,
posited
*syds *isyds
HOLT
45O
N. PARKER
Aiolic:
The First
Compensatory
We can go further.
Ratherthana unity,
Aiolic is splitbyone oftheearliestGreeksoundchanges,theFirstCompensatory
Lengthening(1CL).81
This is Hainsworth's
no. 4 (see above,p. 444), withthetellingpoint"not
a poorcandidatefora definingqualityofAiolic.
Boiotian,"and therefore
A properformulation
of the FirstCompensatory
Lengtheningis crucial
to understanding
thedevelopmentof the Greekdialects,and so requires
a certainamountof space. Those whose eyes glaze over at linguistics
(but have stuckwithme so far)maywish to skipthe followingdetailed
The importantfactto be bornein mindis thatthe forms
presentation.
attestedin Lesbian and Thessalianwithdoubleresonants(e.g.,Lesbian)
theoriginal
Pan-Greek
Boiotian,
represent
stage;all otherdialects,including
haveundergonea laterchange,theFirstCompensatory
Lengthening:so
Boiotianand Doric oeA,otva,
and Attic-IonicaeA,f|VTi.That
is,once again,
LesbianandThessalianaredistinguished
notbyan innovation
butmerely
bya failureto undergoa changeseenelsewhere.
The presentations
in mosthandbooksare necessarilyscatteredand
failto captureseveralgeneralrules.The basic sequenceof eventsin the
FirstCompensatory
was thata resonant+ sys + resonant,
or
Lengthening
resonant+ y all becamedoubleresonants
in all dialects,butthenin certain
dialectsVRR > VR, whichone can viewas a compensatory
lengthening
or a simpleshiftof mora/assimilation.
More precisely,
a vowelis lengthenedas a resultof thesimplification
of a followingnonpalataldoubleresonantcluster.82
In brief,VRR > VRy
79. Forthisoriginalparadigm
with
nom.u(ot,gen.ia<;,see Schmidt1898;
Peters1980,p. 132,n. 80 (crediting
Eichner);Meier-Briigger
1992,vol.1,
Hackstein
60;
2005,
p.
pp. 178-179,
whoshowsthesamedevelopment
in
TocharianA si andTocharianB sana.
This analysisdiffers
somewhat
from
thatofMeier-Briigger
(1992,vol.1,
p. 60) andGippert(2004,pp. 162163),whostartwithfreeLindemanns
Law variants
withintheparadigm
I *sm-iyeh2s,
which Meier*sm-yeh2s
Briiggersees as developingto h(i)yds
(with*smy> *sy)I *(m)midsy
respec-
The resulting
ia<;(withLesbian
tively.
thenformanalogical
psilosis)and uux<;
nominatives
ictand uia. Lindemanns
Law variants
withinGreekaredoubtandwe expectthePGrk
ful,however,
>
to followthepattern
*syds
o*dyiws
Zevq, *dyem> Zfiv(a).
here.Unfortunately,
someofhisrules
forthesoundchangesareproneto
as a resultoftheuseofa
error,
partially
framework,
synchronic,
generative
whichdoesnotdistinguish
between
andlater,dialectical,
Proto-Greek
soundchanges.He failsto capturethe
oftheFirstCompengeneralization
andhiscomsatoryLengthening
(pp.78-79,95-96,
plexexplanations
108-109) callforseveraldifferent
rules,nonewithouttheirdifficulties,
whichin factproduceincorrect
forms.
82.This formulation
of1CL is based
on thatofmyteacher
WarrenCowgill
1967).Though
(modifying
Kiparsky
THE
LINGUISTIC
CASE
FOR THE
AIOLIAN
MIGRATION
451
Figure3. Chartofthedepalatalizaclusters
tionofresonant
in cases of mm> m>nn > nyrr> r,//> /,and ww > w, but notin cases of
*yyor *IT (whichcomesfrom*/yand laterdepalatalizesto XXonlyafter
This changeis foundin all GreekdialectsexceptLesbian
1CL is over).83
andThessalian,whichmerelyretainthePan-Greekstage.84
There arethreesourcesforthesenew doubleresonantclusters:
> RR (eitherbymetathesis
1. *Rs/sR> *hR/RA
or simpleassimilation).
2. A depalatalized-RR- from-Ry-.ForRy > R'R',whathappens
nextdependson theprecedingvowel.If thevowelis non-high
and back (i.e., as unpalatalas possible),namelya or o, then
a new diphthongis created:*ann > ayn,*oriri> oyny
i.e.,*ri
merelyloses itsnasalityand turnsintothehomorganicresonant
*n is depalatalized.Similarly*arY > ayr>
y, and theremaining
*orY> oyr.But ifthevowelis highor frontft,eyu)>thenR'R!
depalatalizes> RR and eitherstaysinThessalianand Lesbian
or elsewherefeedstheFirstCompensatory
Lengthening(see
two
treatto
note
that
these
It
is
3).
differing
Fig.
important
mentsof R'R' arepresentin all dialectsand so arePan-Greek.85
Lesbian andThessalianalone of Greekdialectssimplymaintain
theresulting
-RR-,butBoiotianshowsthecompensatory
long
vowels.86
an outlinecanbe
neverfully
published,
foundin Cowgill1969 andin Crist
counter2001,pp.76-77.The putative
ofRR,
clusters
examplesaresurface
mostofwhichareeasilyexplained.
ofhR
Manyarefromthetreatment
in quasi-initial
position:#sR-> hRspelledFH, PH, etc.,whencetherough
>
> *hpofo<;,
on p-.So *srdwos
breathing
Horn,pooq(povq).In quasi-initial
posilateenoughto
tion,hR-is stillpresent
(at morundergoa secondary-change
to -RR-,butonly
phemeboundaries)
>
afterthe1CL is over.So: *srew-o> peco,
etc.Note
butmxappeco,
*hrewothatseveralsuchformsin Homerare
corthephonologically
likelyto reflect
> eppeov;
so:*e-srew-on
rectPGrkreflex;
> uei5-idco
but(piA,oun.ei8r|<;
*smeydfondofsmiles'(Sihler1995,pp. 170171,171). The otherclassofputawouldbe derivativecounterexamples
> ypa\L\iaf
etc.
tivessuchas *Ypd(p-n.<x
that
Thereis goodevidence,
however,
unassimilated
thesegroupscontinued
untilsurprisingly
late;so,e.g.,Ion.
from*GT<p-ua;
dissimilated
eoxe6uevoi
cf.Hsch.axe9|iaxa axe^jxaxa(I owe
thisexampleto Don Ringe;see Scherer
1959,p. 263). ewea, ofwhatever
remodelsource,showsconsiderable
ingfromPIE *newn.
HOLT N. PARKER
452
outcomesoftheFirstCompensatory
Examplesofthediffering
Lengthare
for
the
first
source
(sR):
ening
-sm-
(Bartonek
2003,p. 146).Myc.doesnot
indicatedoubleconsonants.
normally
Much dependson theinterpretation
of
thero2and ra2signs:ro2spellsetymo=
in thecomp.a-ro2-alogical*ryo
aryoa< *aroha< *ar-yos-a(cf.ap-iato<;),
etc.,andsecondary
(syncopated)
-ryo= 7cop(p/6p(i)o(;,
=
mpo-pu-ro2
tu-ro2
rop(i)ov,dim.'littlecheese,'as wellas
=
secondary
-lyo-inpi-ti-ro2-we-sa
> *stalnd> Lesb. aiotMxx,
so *stl-neh2
of
(no example primary
ptilyo-we(s)sa
Dor.axdXa,Att.(5rr\kr\.
Risch(1974,
in
lyo-).ra2spellsetymological
*-ryamostcases,butspellsetymological
*-rsp. 110,39f) had alreadyderivedit
= *ager-safrom*sth2sleh2
in thefirst
aor.a-ske-ra2-te
'stand'),withthe
(*stf>2-slasuffix,
seenin Lat.
nt-es> *agerhantesl*
infrequent
(cf.3.
agerrantes
*
> scdla.However,
=
*skand-sla
to
a-ke-re
cf.
pointing
sing.pres.
ageryeil*
agerrei;
*stl-nais Old Saxonstollo,
Lesb.dyeppco,
butAtt.dyeipco).
showing
Again,
boththezero-grade
andan /z-stem.
an *agersantes
wouldbe spelledfa-kea
connection
with
the
sa-nt-es
the
Further,
The
though
by usualconventions.
rootfor'stand*
is appealing,
the
has
been
taken
as
spellingboth
ra2sign
semantics
arelessattractive:
-slarY andrr,butthisis unlikely,
sincenot
formant
appearsto be an instrumental
onlyis a separatesignfordouble
consonants
in LinearB,
('thatbywhichone climbs').
unparalleled
- 6(peMxov/6(peAxov, buta signforrra(Ila,rro,116)necessar89. So o-pe-ro
sincean *6<peA,vcov
wouldbe spelled
overa syllable
break
ilyextending
violatestheprinciples
ofthesyllabic
bytheusualconventions
fo-pe-no
The likeliest
writing
system.
explanationis thatoriginalsignsforrya/Iya,
andryo/lyOy
whether
fromprimary
or
continued
(<
etc.,
secondary
*rya *riya)>
tobe usedevenwhen*ryahadbecome
-rra-,etc.,andso wasusedfornewrra
from*rha< *rsa,in a waysimilarto the
use oftheAttic"spurious
diphthong."
See Bartonek
2003,pp. 105-106,146147,fortheevidence.
90. A goodexampleof1CL preceding(feeding)theAtt.-Ion.changeofa
> t|.The newroughbreathing
seenin
Art.,Boiot.,andDoriccomesfrom2.
is recharacpl.\)ue-.In Att.this*f|ueterizedwiththeace.pl. endingfiue-cx<;
> fjuaq(Sihler1995,p. 380,369).
91. NotethatGrassmann's
Law
ofaspirates)
doesnot
(dissimilation
*hR>
apply,thatis,theassimilation
RR precedesGrassmann's
Law,which
againshowsthatwe arenotdealing
withthemerecompensatory
lossofh
butthesimplification
ofdouble
resonants.
THE LINGUISTIC
453
-sn-
> *arjgelTd
but
(in all dialectsand so not 1CL) > dyye^Xco
*arjgel-yo
>
Boiot.
but
firstaor.*ayyeX-aa Lesb.,Thes. dyYeAAd-uevoq,
Art.inf.dy/eTAm
(1CL withhighvowel).
dyyei^auevcoq,
-nsbutBoiot. ueivdxco,
*men-sa-> Thes. part.gen.pl. cruu-UwdvT-oi)v,
Art.eueiv-e.
Thes. ueiw-oq;in the 1CL
*mens-os:
gen.'month'> Lesb. UT^vv-oq,
dialectsthelengthening
appliesvacuouslyto thelongvowel,but
the
show
RRyso: Boiot. ueiv-oq,Art.uriv-oq.93
simplified
they
The -nn-formsshowthatRs musthavepassedat leastto Rh
(> RR) beforePan-GreekOstoffs Law applied.
It is especiallyimportantto note herethatthe FirstCompensatory
Lengtheningis cruciallyorderedafterthe changeof sonant*r> pa. So
>
> Tpotpcov
> *trarron
> *trahron
*tfs-ro-Hon'possessingfear'> *tfs-ron
TpT|pCOV.94
therearemanyexamplesin Lesbian(though
Forthesecondsource,Ryy
fewerattestedin
aftera/o:
Boiotian).For thetreatment
> *gwann6>
> *gwany6
(Lat. venio)> PGrk *gwamy6
*gwm-y6
Pan-Greekpaivco.
> *hmorYa> Pan-Greekuoipot.95
*smor-ya
*
But aftere:
> *awerYo> Pan-Greek*awerro,
whichremainsin Lesb.
*awer-yobut > Boiot.,Art.deipco(1CL).
deppco,
92. The initiala- is an old and unsolved problem,possiblyinfluencedby
linguistictaboo.
93. Att. back-formsnom. uf|v.Ion.
shows originalnom. *mens> *mens
(OstofFs Law) > ml2s= |iei<;by the
Second CompensatoryLengthening.
94. A furtherexample is suggested
by Lejeune^ derivation(1972, p. 122,
('head,' 'in
115) PGrk *krs-ro-s
> Kepva)
*kers-na
full
cf.
grade
charge,'
> Att. va\>-Kpapo<;
> *krarros
> *krahros
> v<xi>-kA,tipo<;.
However,see Nussbaum
1986, pp. 24, 167, 221, 244-245, forthe
*
>
preformof PIE krh2s-r-6-> *krdsro
>
with 1CL
*krdrro -Kpapo^ > -icA,T|po<;,
applyingvacuously.
95. The comparative*ar-yos-a
(cf.dp-iaxoq),Myc. a-ro-a (*aryosa>
*aryoha> *aryoa)[see above, n. 89],
seems to show thatMyc. is stillat the
before
stage of*aryo-/*arYo-/*arro-y
the diphthonghas formed(i.e., we have
a-ro2-a aryoaand not the expected
'
fayro-spelled presumablyt a-ro-a or
the like). The problemis thatthe expected faipcovdoes not show up anywhere.Instead we have Att. dpeicov
with some typeof extensiveremodeling
(Sihler 1995, p. 362, 354.4b). A similar recombinationmightalso account
forthe Myc. form.
96. Again cruciallyorderedafter
*r> ap but before1CL, which it
bleeds.
HOLT N. PARKER
454
Figure4. ChartoftheFirstCompensatoryLengthening
*kten-yo> *kteririo> Pan-Greek ktenno,remains in Lesb. Kievvo),but
ktt|vcoDor., KiewtflAtt. (1CL; not attestedforBoiot.).
Compare afteri:
*krin-yo> *kriririo> Pan-Greek krinno,remains in Lesb. Kpiwco >
KpivcoAtt. (1CL; not attestedforBoiot.).
For the thirdsource, -In- clusters,there are excellent examples from
inheritedand remodeled nasal presents:
> Lesb.
PIE *h3bhe'l'owe' formsa Greek nasal present*opbe/-nobut
Doric
and
and
Boiot.
Att.-Ion.
ocpeMxo,
6(pt|Xco,
ocpeiAxo
(1CL).97
> *gwole-(a regularmetathesis) in
PIE *gwelh3'will,wish/ *gwe/oArk., Cyp., Pamph., and West Ion. p6A,oum;with a nasal infix
> *gw/-no->
remodeled to a full-grade
pres. *gwl-ne-h3-/*gwl-n-h3on the thematic root present.98This in turn
*gwe/no-/*gwolnoshows the various outcomes of*-In- and labiovelars before e:
Lesb. p6M,oum,Thes. peAAouai, with retained -XX-,but Att.Ion. Pot>A,ouai,Dor. 8r|A,o|iou,
and Boiot. peiA,our|(1CL).99
97. See theexcellent
summary
by
Slings1975.Fortheinitiallaryngeal,
> <p^-ov
cf.aor.*e-h3bhl-e/o(Myc.
and
the
o-po-ro)y
neg.vco(peA,T|<;
(Myc.
Beekes1969,pp.56,225no-pe-re)\
256.Theseverbscannotreflect
-Resincethiswouldresultin
presents,
-XXin all dialects,
andtheonlyother
phonologically
possiblesourcewouldbe
in *-se/so-,
a suffix
whichis impossible
tojustify
orsemantimorphologically
see Slings
cally(i.e.,a desiderative);
1975,pp.3-4: "I failto see howanywouldbe
bodyinwakingcondition
ableto terma verbmeaning
'towish'a
'desiderative.'"
98. Fortheshapeoftheroot,see
Sihler1995,pp.498-500,453-454;
Rixet al. 2001,p. 17.
99. NoticethatBoiotiansharesthe
as well;pcuA.fi,
Doric -grade
etc.,are
builtdirectly
to theverb.See Rixet al.
2001,pp.208-209.
THE LINGUISTIC
455
in Greekwithsecondary^-grade
*wel-'roll/PIE *w/-new/nu-y
butDor. fr\ke.(oy
Att.ei^eco(1CL);100
*wel-new-> Lesb. eXXeco,
cf.thezero-grade*sm-wl-nes'rolledtogether'> Lesb. aoXfoy;
( with*/> ol and Lesbian psilosis),and thezero-gradeadv.in
(Hsch. a 2761 [Latte]).The ^-gradeform&AAt|<;
ot^avecoq
> *hawelles>
is attestedat //.3.13 and underlies*sm-wel-nes
to
Att.
(with1CL) contracting
aki\q.
*aeikr\q
The sourcesand outcomesof the FirstCompensatoryLengthening
aresummarizedin Figure4.
The Position
100.Rixetal.2001,p.675.
101. See Garcia-Ramon
1975,p. 70,
6.2.1,whoputsitdownto a Boiotian
ofgeminates.
Although
simplification
he saysthechangecannotbe dated,he
nonetheless
placesitafterca. 1125,the
datehe proposesforBoiotiansplitting
offfrom"proto-thessalien"
(p. 110).
102.1 canfindno examplesofthe
FirstCompensatory
Lengthening
in Boiotian(but
to *r> po/op
applying
formsareveryfew).
thepo/op
103.Hainsworth
(1982,p. 862)
explainsitbysayingthat"thegeminationofliquidsandnasalswas toolate
thatis,he takesthe
to affect
Boeotian,"
>
of
*sn
-nn-,
etc.,to be a late
changes
Lesbian-Thessalian
affair,
failingto see
thatitis in factProto-Greek.
for"West
104.The usualmarker
Greek"is the1. pl.endingin -ue<;.It is
statedthatBoiotianshared
sometimes
-uevwithLesbian,buttheendingis
beforethespreadofkoine
notattested
(Scherer1959,p. 18).Thessalianhas
onlya singleexample(-uev),fromLarissa (DialectorumGraecarumExempla
of Boiotian
as a laterand independentchangeof
difference
Explainingthisstriking
VRRto longvowelplusR withinBoiotianafterit splitofffroma ProtoThe examplesoftheFirstCompensatory
Aiolic cannotwork.101
Lengthin
come
onlyfromProto-Greek-RR- clustersand there
ening Boiotian
are apparentlyno cases of any secondarysourcesfordouble resonants
That is, a laterBoiotian rule should apply
undergoingthis treatment.
forexample,shouldgivefa^oq; the new
withoutexception,and dtXkoq,
> Uekonow^Goq,shouldhavegiven
as
such IlekoKoc-vx\oo<;
assimilations,
etc.
trietamcbvTiGoq,
Further,since the First CompensatoryLengtheningcuts Boiotian
all the otherfeaturesthatare said to
offfroma putativeAiolic family,
Aiolic thephonologicalchangesof*Kwe> Pin all positions,
characterize
ofdativepluralin -eaai, perfect
innovations
*r> po/op;themorphological
- musthavebeen
=
in
ioc
uia
-cov,
-ovx-,
earlyand
exceptionally
participle
This is not a conhave precededthe FirstCompensatoryLengthening.
anditseemstobe ruledoutbythefactthattheFirstComvincingscenario,
pensatoryLengtheningis cruciallyorderedafterthechangeof*f> po/op
in otherdialects.102
The conclusionseemsto be clearthatBoiotianbelongsto a different
whichdo notundergo
groupofGreekdialectsthanLesbianandThessalian,
No earlyisoglossesseparateit
the FirstCompensatoryLengthening.103
fromWest Greekuntilthe innovationof labiovelarsto dentalsbeforee
branchof
We mayviewBoiotianthenas a conservative
marksoffDoric.104
a
from
Doric, althoughnothingpreventsBoiotian
being separatebranch
of Greekaltogether.
Other
Claimed
Subgroupings
of Aiolic
The departureof Boiotianleavesjust Thessalianand Lesbian. Is it possible thatat least thisgroupsomehowformsa family?Again,thereare
and I see no evidenceforThessalianand
no securelysharedinnovations,
Lesbian havinganycommonancestorotherthanProto-Greekitself.
ofthequestion,holds
Garcia-Ramon,in themostdetailedtreatment
thatBoiotiansplitofffroma Proto-Aiolic,whichhe rathertendentiously
laterLesbian migrated
and thatabouta century
labels"proto-thessalien,"
There are problemswith thisview,however,
from"thessalo-lesbien."105
and he himselfrightly
pointsout thatthelackoftheFirstCompensatory
andThessalianis not a reasonto groupthetwo
in
Lesbian
Lengthening
HOLT
456
N. PARKER
THE LINGUISTIC
457
HOLT N. PARKER
458
Thereareevenmoresourcesfor-ecoverbs:
1. athematicstativesin *-eh2made to zero-gradeCaland roots,such
-> ep-oGe-co
> y\to\\i\
->
as *h1rudh-eh2-mi
'be red'; *bhil-eh2-mi
cpiXeco;
in -e-yo-y
2. thematicdenominatives
builtto thematicnominals,
> oiiceco;
suchas *woik-e-ydin -eyo->
3. thematiccausative/frequentatives
withtherootoftenin
> (popeco;
the0-grade,suchas *bhor-eyo4. plainthematicsbuiltto stemsendinginy, wyor s (whichthen
disappearsbetweenvowels),suchas *srew-o-> peco;
5. thematicdenominatives
in -yo-builtto s-stems,suchas *teles-yo> teleyyo> xe^eico,laterxe^eco.124
In short,all thedialectshaveamplematerials
tobuildanalogicalforms
and to regularizethevaryingparadigms.Most dialects(includingThesbutArkadian,Cyprian,
saliotis)havechosenthemorenumerous
thematics,
of
athematic
forms.Lesbian
Lesbian,andPelasgiotisshowvarying
spreads
showsinherited
stativeslike<p(A,Tiiu,
besidedenominatives
likeoikt|ui,and
iterative
causativeslike<popf|ue0<x;
butforthelasttwocategoriestheforms
remainthematic,e.g., peovxoc,
and numerousattestations
of
7U7i^eovTa,
formsof xeX,eico.125
That is, at the time(s)the old thematicswererecast
as athematics(at leastin some of theirforms),pecoand xeXeico
werestill
in theformof*pep-co
and *xetayy-co.
The Thessaliandata showonlynine
formswithclearathematicinflection,
but Sie-aoccpei-uevoc
seemsto point
to thespreadof athematicformsto *-es-yoverbsas well.126
Equally,any attemptto link the mainlanddialects,Thessalian and
Boiotian,as a subgroupdoes notwork,eventhoughtheyaretheonlypair
to whicha versionof a dialectgeographyapplies.127
Not onlyareThessalianand Boiotianseparatedby the FirstCompensatoryLengthening,
Of thethreefeatures
theyshareno commoninnovation.
commonlylisted
in handbooks,128
ThessalianandBoiotianti(mediallyandfinally)
is merely
retainedfromProto-Greek,
whileLesbianundergoesa latersoundchange
to si.129
The spreadof theathematicinfinitive
endingin -jievto thematic
stems(e.g.,(pep-e-uev)is foundlatein Boiotian,butinThessalianapparwhiletheotherpartsofThessaly
entlyonlyin Pelasgiotisand Perrhaibia,
show*-e-en> -en.mIn short,the further
awayone is fromBoiotia the
morethethematic-e-uevseemsto flourish.
That thecreationofsuchnew
formsdoes notindicatesharedancestry
is shownbyCretannpofeut-e-uev.
Greek
inherited
a
wide varietyof infinitive
formants
and created
Rather,
stillmore;whichof themwerechosenor generalizedvariesgreatlyfrom
dialectto dialectand providesno firmbasisforgrouping.131
124.Possibleexplanations
ofKaAico,
etc.,andthepurelyGreek-6-classare
outsidethescopeofthispaper.
125.Forthedata,see Blumel1982,
pp. 172-178,187-191,pp.222-223,
236; Hodot 1990,pp. 192-198.
126.Lesbianevepyei-e-VT-eoai
besideThes. e\)epYex-e-<;
in
the
(< *-nt-s)
derivedcompoundis unlikely
to representa directly
inherited
-es-yo->
butsimply
AtticevepyeTTn;,
euep-eyyoto thenewpatterns.
Yeteco
subjected
127. See,e.g.,Rodrigues
Adrados
1956.
128.Buck1955,p. 148,204,who
notesthatonlyone,thethematic
inf.in
"whichis Homeric,belongsto
-men>
theAeolicelements
ofthesedialects";
Schmitt1977,pp. 75-78.
129.Cowgill1966,p. 80; Schmitt
THE LINGUISTIC
459
CONCLUSIONS
thedialectsalonga
databyarranging
We mightsummarizethelinguistic
on theleftandthosewitha greaternumber
linewiththemostconservative
of significant
changeson theright:
Thessalian| Lesbian | Boiotian | Doric | SouthGreek
SouthGreek(Mycenaean,Arkadian,Cyprian,Attic-Ionic)sharesthe
> ss > sywhileBoiotian,Doric,and the
earlyassibilationofti > si and *t(b)y
historicaldialectsof SouthGreeksharetheFirstCompensatory
Lengthto be
and
Lesbian
Thessalian
consider
we
In
this
lineup, might
ening.
us
to
that
view.
but
relatedconservative
Rather,
dialects, nothingcompels
to their
itmightbe betterto arrangethemin a linecorresponding
roughly
geographicpositionin historicaltimes:
Thessalian| Boiotian | Doric | SouthGreek | Lesbian
dialectsappearnot
the two mostconservative
In thisarrangement,
and thefar
as relateddialectsbutratheras relicareasin thefarnorthwest
can be used withanyconfidenceto
northeast.
However,no arrangement
dialectgeography.
reconstruct
prehistoric
the
Attic-Ionic,
p. 83. So, forexample,
Lokrian
of
northern
part Thessalian,
Greek),Corinthian,
(Northwest
Megarian,andRhodian(Doric) all sharethe
inf.in -e-en,whileArkadian
thematic
andmostoftherestofDoric show-en
to thestem.See Sihler
addeddirectly
1995,p.608,552A.l.a.
132.Blumel(1982,pp. 155-158,
theevidence.
171-173) surveys
133. Garcia-Ramon
(1975,pp.65en
"On hesiterait
66, 6.1.7) writes:
principea rangerce traitparmiles traits
notingitsabsencein Lespaneoliens,"
ofPhokianforms.
bianandthepresence
134. Schulze1933,p. 399; followed
byScherer1959,p. 39, 237.14;Blumel1982,p. 156,n. 148; Schmitt1977,
(1975,
p. 71,no. 15. Garcia-Ramon
to see an analogyfrom
p. 65) prefers
-ueBa,-o0e,butno proportional
analogycanbe made.
135.Fordetailsandfurther
complications,seeSihler1995,pp.548-549,492.
136.Blumel1982,pp. 155-158,
beside
171-173. Cf.MeA,dve-io<;
MeA,dvT-ac.
137. Schwyzer
1939-1953,vol.1,
Garcia-Ramon
1975,
p. 353,A.18,43;
p. 66.
460
HOLT N. PARKER
of
138.Cf.thesensibleremarks
citedon p. 433
MeyerandGschnitzer
andin n. 7, above.One consequence
is thatthewholetopic
ofthisresearch
in Homerneedstobe
of"Aiolisms"
reexamined.
THE
LINGUISTIC
CASE
FOR
THE
AIOLIAN
MIGRATION
461
REFERENCES
Adams, D. 1981. "The Pre-Historyof
Tocharian PreteriteParticiples,"
in Bono Homini Donum: Essaysin
HistoricalLinguisticsin Memoryof
J.AlexanderKerns,ed. Y. L. Arbeitman and A. R. Bomhard,Amsterdam, pp. 17-24.
Ahrens,H. 1839-1843. Degraecae linguae dialectis,2 vols., Gottingen.
Aikhenvald,A. Y., and R. M. W.
Dixon, eds. 2001. Areal Diffusion
Problemsin
and GeneticInheritance:
ComparativeLinguistics,Oxford.
Bartonek,A. 2003. Handbuchdes
Griechisch,
Heidelberg.
mykenischen
Beekes, R. S. P. 1969. TheDevelopment
LarynoftheProto-Indo-European
geals in Greek,trans.T. S. Preston,
The Hague.
Bellwood, P. 2001. "Early Agriculturalist Population Diasporas, Farming,
Languages, and Genes," Annual
Review ofAnthropology
30, pp. 181207.
. 2005. FirstFarmers:The OriSocieties,Oxgins ofAgricultural
ford.
Blumel,W. 1982. Die aiolischenDialekte:PhonologieundMorphologieder
Texteaus generativer
inschriftlichen
Sicht(Zeitschriftfurvergleichende
Erganzungshefte
Sprachforschung,
30), Gottingen.
Bryant,E. 2001. The Questfor theOrigins ofVedicCulture:TheIndo-Aryan
MigrationDebate, Oxford.
Buck, C. D. 1955. The GreekDialects:
GlosGrammar,SelectedInscriptions,
sary,2nd ed., Chicago.
Cartledge,P., ed. 2002. The Cambridge
IllustratedHistoryofAncientGreece,
Cambridge.
Chadwick,J. 1956. "The Greek Dialects and Greek Pre-History,"GaR
3, 2nd ser.,pp. 38-50 (repr.in The
Language and BackgroundofHomer,
ed. G. S. Kirk,Cambridge 1964,
pp. 106-118).
. 1973. "Discussion" to V. I.
Georgiev,"The Arrivalof the
Greeks in Greece: The Linguistic
Evidence,"in BronzeAge Migrations
and Linin theAegean:Archaeological
in
Greek
Problems
Prehistory,
guistic
ed. R. A. Crossland and A. Birchall,
London, pp. 254-255.
pp.58-126.
HOLT
462
Desborough, V. 1972. The GreekDark
Ages,London.
Dixon, R. M. W. 1997. The Rise and
Fall ofLanguages,Cambridge.
M. 1994."The Dialect
Finkelberg,
ofAncientGreek,"
Continuum
//SCP96,pp.l-36.
8,pp.41-60.
N. PARKER
Ridge,NJ.
THE LINGUISTIC
LateAntiquity
yed.A.-F. Christidis,
Cambridge,
pp.387-394.
P. 1967."SonorantClusters
Kiparsky,
in Greek,"Language43,pp.619635.
P. 1909."ZurGeschichte
Kretschmer,
dergriechischen
Dialekte,"Glotta1,
pp. 1-59.
LaPolla,R. 2001."The RoleofMigrationandLanguageContactin the
oftheSino-Tibetan
Development
in Aikhenvald
LanguageFamily,"
andDixon2001,pp.224-254.
. 1997."The Homelandsofthe
inArchaeology
Indo-Europeans,"
and
and Language 1: Theoretical
ed.
Orientations,
Methodological
R. BlenchandM. Spriggs,
London,
pp.93-121.
J.P.,andD. Q^ Adams.2006.
Mallory,
to ProtoThe OxfordIntroduction
Indo-Europeanand theProto-IndoEuropean World,Oxford.
A. H. 195S.American
Marckwardt,
NewYork.
English,
M. 1992.Griechische
Meier-Briigger,
2 vols., Berlin.
Sprachwissenschaft,
. 2003. Indo-EuropeanLin-
trans.C. Gertmenian,
guistics,
Berlin.
MendezDosuna,J.2007."The Aeolic
Dialects," in A HistoryofAncient
Greek:FromtheBeginningstoLate
ed.A.-F. Christidis,
Antiquity,
Cambridge,
pp.460-474.
of
A. 1968."TheTreatment
Morpurgo,
rand/inMycenaeanandArcadoCyprian,"in Atti e memoriedel T
internazionalede micenoCongresso
logia(Incunabulagraeca25), Rome,
pp.791-814.
MorpurgoDavies,A. 1976."The -eaai
Datives,Aeolic-ss-,andtheLesbian
BeitragezurSprachwissenschaft
16),ed.A. MorpurgoDaviesand
W. Meid,Innsbruck,
pp. 181-197.
Nettle,D. 1996."LanguageDiversity
inWestAfrica:An EcologicalAp15,pp.403-438.
JAnthArch
proach,"
Nichols,J.1997."ModelingAncient
andMovePopulationStructures
mentsin Linguistics,"
AnnualReviewofAnthropology
26,pp.359384.
. 1998."The EurasianSpread
Zone andtheIndo-European
and LanDispersal," in Archaeology
2:
guage Correcting
Archaeological
and LinguisticHypotheses,
ed.
R. BlenchandM. Spriggs,
London,
pp.220-266.
A. J.1986.HeadandHorn
Nussbaum,
in Indo-European,Berlin.
A. Goldhammer,
Cambridge,
Mass.
bePejros,1. 1997."AreCorrelations
tweenArchaeological
andLinguisin
ticReconstructions
Possible?"
and Language 1: TheoArchaeology
reticaland MethodologicalOrienta-
Vienna.
463
. 1974. Wortbildung
derhomerischen
2nd ed.,Berlin.
Sprache,
Grammatikdes
Rix, H. 1992. Historische
Laut- undFormenlehre,
Griechischen:
2nd ed.,Darmstadt.
. 1994."LateinundSabellisch,"
Wiesbaden.
Adrados,F. 1952.La diaRodrigues
lectologia
griegacomofuentepara el
estudiode las migraciones
indoeuropeas
enGrecia,
Salamanca.
. 1956."Achaisch,
und
Jonisch,
IGForsch
62,pp.240Mykenisch,"
248.
Fact
Rose,C. B. 2008."Separating
fromFictionin theAiolianMigration,"
Hesperia11,pp.399430.
minora
Ruijgh,C. J.1991-1996.Scripta
ad linguamgraecampertinentia,
2 vols.,Amsterdam.
. 1995."D'Homereauxorigines
de la tradition
proto-myceniennes
du
epique:Analysedialectologique
avec
un
excursus
langagehomerique,
surla creation
de l'alphabetgrec,"
in HomericQuestions:Essaysin PhiAncientHistory,and Archaeollology,
Amsterdam,
ogy,ed.J.P. Crielaard,
pp. 1-96.
B. 1981."1stein Raum/ZeitSchlerath,
Modellfureinerekonstruierte
Cambridge.
Origins,
Rhodes,R. 2006."OjibweLanguage
Sprachemoglich?"
Zeitschrift
fiir
Shift:1600-Present,"
95,
Sprachforschung
http://lingweb. vergleichende
eva.mpg.de/HunterGathererWork- pp. 175-202.
Schmidt,
(acJ.1898."Das Zahlwortuia,
shop2006/participants.html
cessedJune24,2007);
1'a,"Zeitschrift
fur vergleichende
Ringe, D. 2006. A LinguisticHistoryof
English1: FromProto-Indo-EuroOxford.
pean to Proto-Germanic,
DialectRisch,E. 1949."Altgriechische
MusHelv6, pp. 9-28.
geographie?"
der
. 1955."Die Gliederung
Dialektein neuer
griechischen
Sicht,"MusHelv12,pp.61-76
(repr.in TheLanguage and BackgroundofHomer,ed. G. S. Kirk,
Cambridge1964,pp.90-105).
R. 1977.Einfuhrung
indie
Schmitt,
Dialekte,Darmstadt.
griechischen
Schulze,W. 1933. KleineSchriften,
Gottingen.
E. 1939-1953.Griechische
Schwyzer,
4 vols.,Munich.
Grammatik,
464
of
Slings,R. S. 1975."The Etymology
BOYAOMAIandOOEIAQ,"Mnemosyne28, pp. 1-16.
0. 1996.Introduction
to
Szemerenyi,
Oxford.
Indo-European
Linguistics,
1932.
Thumb,A., andE. Kieckers.
Dialekte 1,
Handbuchdergriechischen
2nded.,rev.E. Kieckers,
Heidelberg.
Trask,R. L. 1997. TheHistory
of
Basque,London.
. 2000. TheDictionary
ofHistorical
andComparative
Linguistics,
Edinburgh.
P. 1999."AWindowon the
Trudgill,
HoltN.Parker
University of Cincinnati
department of classics
4IO BLEGEN LIBRARY
CINCINNATI, OHIO 4522I-O226
holt.parker@uc.edu
HOLT N. PARKER
Past:'ColonialLag' andNew ZealandEvidenceforthePhonology
ofNineteenth-Century
English,"
American
74,pp.227-239.
Speech
K. 1997. TheArabicLanVersteegh,
New
York.
guage,
C. 2001."AnIndo-European
Watkins,
AreaandIts CharacterisLinguistic
tics:AncientAnatolia.ArealDiffusionas a Challengeto theComMethod?"inAikhenvald
parative
andDixon2001,pp.44-63.
Weiss,M. 1998."Erotica:On thePreofGreekDesire,"HSCP 98,
history
31-61.
pp.