J. Linguistics 22 (I986), 443-454. Printed in Great Britain
Dialects and diachronic syntax: free relatives in
Old Spanish'
MARIA-LUISA
RIVERO
University of Ottawa
(Received 14 March I986)
I.
INTRODUCTION
In Rivero (1984), I argued that Old Spanish wh-items in non-interrogative
constructions appear in A(rgument) position in surface structureas quantifierlike expressions, and may constitute the antecedent of a restrictive relative,
as the quantos in (i); they can also be elements in non-A position (COMP)
in headed, or free relatives, as in (2 a-b).
(I)
[NP[NP quantos] [S'COMP qei] [S ti la udieron]]]
All-those
(2)
(a)
NP[NP
that
it they-heard
todos omnes] [SCOMPquantosi] [s ti
All men
all-who
esta carta vieren]]
this letter theywouldsee
(b) [NP [NPe] [S [COMPquantosi] [s ti aqui sedemos]]]]
all-who
here we-are
Then Old Spanish wh-words belong to the list of relative and question items,2
and, in addition, are included within the paradigm of quantifiers such as
muchos 'many Although dialects may differ as to the number of specific
wh-items listed in their lexicons, I could see no variation in the lexical double
categorization (i.e. relative and quantifier), once an element was listed. I
'.3
This work has been partially subsidized by Research Grant 4I0-84-0370 from S.S.H.R.C.C.
I thank Montserrat Morales for her enthusiastic help in gathering the data.
[2] I omit questions from the discussion. These are never of the wh-item+ que type, and can
be 'non-matching' when heading a subcategorized complement. Hybrid constructions such
as Recuerda con los problemas (con) que se enfrenta 'Remember which problems he faces'
are not part of the medieval system (see Hirschbuihler& Rivero, 1983, for discussion). New
compounds of the wh-type pattern syntactically as (i) and (2 b) in the I3th century, as we
see in section 2, but fail to appear in questions. During this period questions seem to have
the wh-item in COMP as in (5).
[1] Quantificational uses of ya + wh-item sequences, as in (i), survive until the i6th century:
(i)
Diol
Dios
man e
mano ya quanta
memoria
Alex goa (o)
Gave-him God
hand and hand
enough
memory
'God gave him enough memory directly.'
For additional examples see Cuervo (1953:338), among others.
[i]
443
15-2
MARIA-LUISA
RIVERO
proposedthat Old Spanishrestrictiverelativesare uniformlyof the type in
(1-2), as in all presentvarieties:[ NP s,[COMPS]]. Under this analysis,
differencesbetween the medieval and later periods follow first from the
specializationof wh-wordsto therelative(andquestion)paradigm,eliminating
(i), and second from the emergenceof principlesprohibitingrelativesin
COMP,if theircontentis recoverable,eliminating(2 a). Thereareno changes
in phrase-structure;
devicesexcludingemptyand doubly-filledCOMPsare
in existencein all dialectsand periods.
RebeccaPosner(I985) had made interestingcommentson my proposal,
indicatingthe need for furtherresearch.HereI will concentrateon only one
of the important questions she raises, namely, dialect variation and
replacement,becauseit has crucialconsequencesfor the model of historical
evolutionI adopted.I willmaintainmyoriginalproposalprovidingadditional
motivationfor it, and outline the line of researchwhich, in my view, can
contributefruitfullyto the debate.
Posnerpoints out that a considerableamount of evidencein supportof
(1-2b) in my papercomes from Aragonese,or texts arguablyinfluencedby
this dialect.However,it is the Castiliandialectthat survivesin laterliterary
materials.If thesetwo medievaldialectsdifferin the relevantrespects,namely
(a) phrase-structure,
(b) conditionson the content of COMP, and (c) the
double lexical classificationof wh-items,then Old Castilianremainsunanalysed.If this is correct,in contrastingmedievalAragonesewith later
Castilian,one could simplybe studyingthe (literary)death of a dialect,as
Posnerindicates.Under the model of diachronicevolutionwith a psychologicalperspectiveI adopted,this conclusionwouldeliminatethe possibility
to hypothesizeANY historicalchangefor Castilian,the survivingdialect,on
the basis of my work. If I did not establishthe output of the grammarof
early generationsof Castilianspeakers,nothingcan be proposedabout the
constructionof a new systemby laterCastilians,and the ensuingdifferences.
Withinthis approach,my analysismay be validas a synchronicgrammarof
medLieval
Aragonese,and contributesto the study of the propertiesof UG.
However,an Aragonesesystemwhich dies cannot be the basis for further
diachronicresearchon laterCastiliangrammarsremovedboth in spaceand
time from contact with Aragonese.If there is dialect death, there are no
changesto discussfor Castilianper se.
Suppose, on the contrary, that Old Aragoneseand Old Castilianare
parallelin theirtreatmentof relatives,even thoughthey may differin other
areasof grammar.Then, texts from the two dialectsrepresentthe outputof
a commonpartialsystem,and providepositiveevidencealong similarlines
for the developmentof latergrammars,suchas the one reflectedin Castilian
literarytexts exclusively.Underthis perspective,thereis no harmin mixing
earlyCastilianand Aragoneseevidencein this area, when proposinga first
system.Also, the death of Aragoneseis immaterialfor the changesseen in
444
FREE RELATIVES IN OLD SPANISH
the later Castilian grammar;those differencescannot be interpretedas
alreadyexistingCastiliandeviceswhichremainconstantthroughtime.
In my earlierpaper,I providedevidencethat the two dialectsweresimilar
in analysingrelativesas in (i) and (2). Withoutrecapitulating,here I will
considertwo new areas of the grammarof the medievalvernacularswhich
motivatemy proposalalong independentlines: (i) the positionof non-tonic
pronouns,and (2) the syntax of wh-compounds.I will mention additional
Castilianand Aragoneseprimarysources,but willnot considerthe necessary
and importantcompilationtask an end in itself. In my view, questionsof
dialectdifferencesand replacementcannotbe answeredby providinglists of
examples arranged by geographical origin, but by postulating partial
grammarsfor the dialect(s)in question.
2.
FREE RELATIVES AND THE POSITION OF NON-TONIC
PRONOUNS
In Rivero (1984), I proposedthat free relativessuch as qui alli se morasse
'who(ever)would live there' (B. Milg. I2 d) should be analyseduniformly
along the lines of (2b), withina structurethat has remainedconstantup to
thepresent.Forthispattern,Posner(I985) outlinesananalysiswherewh-items
appearin the head position followedby an emptyCOMP:
e] S]]].
[ NP [NP quantos [ SIICOMP
In my view, Old Castilianand Aragonesecannotdifferin this respect,since
the latterstructureis not viable in eitherdialect;rather,in both cases this
typeof freerelativeis as in (2b), anddisallowstheemptyCOMP.Theposition
of non-tonicor cliticpronounsduringthemedievalperiodleadsto a common
treatment of all relatives, eliminatingthe possibility of a null COMP
systematically.It is well-knownthatnon-tonicpronounsneverstandin initial
positionin earlyRomance.This manifestationof Wackernagel'slaw (i892)
has been much debatedin Romancelinguistics(see Ramsden(I963: I-24)
for a summary of interpretationsgoing back to Friedrich Diez). For
13th-century
Spanish,I proposea restrictioncheckingthelocationof theclitic
in the syntacticsurfacestructurein view of its phonologicalcontent:
(3) A non-tonicpronominalcannotbe theinitialsyntacticconstituentof
the minimalS' that containsit.
As a result,thereis no obligatoryrequirementthat the clitic be in second
position(i.e. the secondsyntacticconstituentwithinS'). Also, as far as Old
Spanishis concerned,the phenomenonappearsunrelatedto a verb-second
type constraint.I have studied the propertiesof clitics in more detail in
Rivero(I985). Here,I concentrateon theeffectof principle(3) on the analysis
of relativeclauses.
Under my proposal,the V in (4) is the firstconstituentin S' containing
phonologicalmaterial,and the structureis well formed.In (5), the question
445
MARIA-LUISA
RIVERO
constituent located in COMP licenses the adjacent clitic as second constituent
in S'.
(4) (a) Alzaronlo de tierra B. Milg. 148a
'They-raised-him from (the) ground'
(b) [ ,[COMPe][salzaron lo. ]]
Por qe me non recudes B. Milg. 293 a
Why me not answer 'Why don't you answer me?'
(b) ISI[ COMPPor qe] [sme non recudes]]
(5) (a)
In Left-dislocations, the overwhelming preference is for a postverbal clitic,
if no material intervenes between the topic constituent and the V. The pattern
appears in early materials, (6), and survives until the I5th century, (7).
mj alma acomjendolo a ti
(6) Mi cuerpo i
My body and my soul I-entrust-it to you
(7) En amar onbres de poca manera
In to-love men of little manner
fazen lo esto por una de dos maneras
they-do it this by one of two ways
'They love men little in two different ways.
SME
1325
COR 8i
Recent analyses of dislocations (Chomsky, 1977; Rivero, I980, for Modern
Spanish) place the topic constituent outside of S'. Applying the same
treatment to the Old Spanish sentences in (6-7), as in (8), I conclude that V is
the first constituent with phonological content in the S'-structure. A
preverbal clitic in this type of structure would be S'-initial and in violation
of the proposed restriction.
(8) (a) [ I TOP mi cuerpo i mj alma] [ S [COMPe] [sacomjendo lo...]]]
(b)
[S,,[TOPen amar onbres de poca manera][SCOMPe][fazen lo. ]]]
A few dislocated constructions parallel to (6-7) in the relevant aspects have
preverbal clitics, as in (9). I assume that en is the resumptive element
coindexed with prior por la montanya.
SME 1414
metio
la montanya sen
(9) Por
she-went
Through the mountain herself-there
'She went through the mountain.'
If it is assumed that Old Spanish has preposed phrases in Focus position (i.e.
adjoined to S), in addition to topic nodes under S, this less frequent order
complies with the system as well, as in (IO):
(io?)
[ [FOCpor la montanya][ssen metio]]]
[ [COMPe]
446
FREE RELATIVES IN OLD SPANISH
In brief,I correlatethe fluctuationin the position of the clitic in dislocated
constructionswith the differentmode of attachmentof the initialconstituent
(topic vs. focus).
Let us now turn to restrictiverelativesfrom this perspective.As often
noticed, an exceptionlessword order patternemerges:the non-tonicpronominalis preverbalin headedor freerelativesalike.Then,thepatternsin (I I)
and (I2) are extremelyfrequentin all dialects.
(I I) (a) Testigos qui lo oyeron...
DLE 50 (1228)
Witnesses whoit heard
F. Ar. I8I
(b) Aquel qui la conpro...
The-one whoit bought
(I2)
(a) Qui
laquisiereuenderal monasterio... DLE 50 (1228)
Who (ever)it wanted to-sell to-the monastery
(b) Non sea
qui te ajude...
F. Ar. 70
Not there-be whoyou help
'Let therebe no one to help you.'
The order seen in (1-12) follows from the hypothesisthat the wh-itemis
within S', i.e. in the COMP position, not from the assumptionthat it is
outsideof S', i.e. in the NP in head position. The analysisfor (12) must be
as in (I3a), and not as in (13b), in all dialects(tracesare omitted):
( 3) (a) [[
e] [s[ compqui] [ te ajude]]]
[NP NP
(b)* [
NP qui]
SIOPS
[SJ COMPe] [te ajude]]]
For our purposes, the structurein (i3b) is parallel to (8a-b) and the
preverbalclitic would be S'-initial if free relatives had the wh-itemin
antecedentposition. Had one dialect analysedfree relativesas in (13b),
lackingthe empty COMP restriction,the word orderpatternwould be qui
ajudete, underthe proposalsI have given. Suchorderis unattested.If these
assumptionsare correct,the position of non-tonicpronominalsin relative
clauses is the type of positive evidence easily available to the medieval
languagelearnerto reachthe conclusionthat the COMPmust be lexically
filled,barringa possibledeparturefroma commongrammarof Castilianand
Aragoneseup to the Renaissance.
After 1450, the positionof cliticsbeginsto be morphologicallyratherthan
syntacticallydefined.This changecoincideswith a new treatmentof clitics.
In the MiddleAges they are syntacticconstituentsof the NP or PP type (see
Rivero(I985) for discussion),and this is why a principlesuch as (3) can be
formulatedin termsof a phrase-marker.Lateron, clitics becomemorphologicaldependentson the verb.Thus,beforeI450, it is impossibleto interpret
thepatternsin (I 2) alongthe linesof (I 3 b). Then,if laterCastilianis analysed
as in (I3 a) too, we can concludethat the death of Aragonesefor literary
447
MARfA-LUISA
RIVERO
purposes has no effect on our hypothesis that no syntactic change has
occurred.
As usual, a principle such as (3) leads to select specific analyses in more
complex structures. The clitics in (14a) must be located as in (04b).
(14) (a) El que ladezir quiere
He who it to-say wants
(b) [SIICOMP
que]
Por 47
la [
Iss5dezir]
quiere]]
Assuming that dezir is the verb of the full sentential complement of quiere, if
no restructuring or clause union has applied, la is not inside the infinitival
clause, since it would be S'-initial. Rather, the clitic is in the matrix S, and
its position is licensed by the material in COMP, as indicated.
Left dislocations containing free relatives show the complex word order
effects following from the simple principles I have assumed, as in (i6), which
is the analysis proposed for (I5) (traces are omitted):
(I5) quanta renda
se
levantare
dista
hered[at]
whateverinterest itself would-accrue from-this possession
dolo
a...
DLE 158 (1207)
(I)-give-it to...
'Whatever interests accrue from this property, I will give them to...'
(i 6)
S"
TOP
S'
I
~
NP
COMP
S
e
A
NP
COMP
quantarenda
do lo a....
S
se levantaredista hered[at]
Se is not initial in its minimal S', nor is lo.
To summarize, clitics are sensitive to Wackernagel's law in Old Spanish,
forrnulatedas in (3). In free relatives, the initial wh-item behaves as the first
synitacticconstituent of the 'sentence' (i.e. S' or lower). If, as often assumed,
448
FREE RELATIVES IN OLD SPANISH
the COMPis reservedfor phraseswith a wh-feature,thenin Old Spanishthe
relativeitem is in COMP:the structureis as in (2b), regardlessof dialect.
3.
THE SYNTAX OF WH-COMPOUNDS
The new wh-compoundsin all dialectsof Old Spanishacquirethe syntactic
propertiesof the olderwh-items,as they enterthe language.Theyinterrelate
with clitic positionin a way that indicatesthat they can fill the COMP.
In the 13th century,the paradigmof wh-compoundssuch as qualquier(e)
'whoever,whatever',quantoquier(e)
'whatever',etc. coexists with the syntactic sequences qual... quier(e) 'what one-may-want' at their historical
origin, as in the contrast in (17) (see also Gessner, I895; Cuervo, 1953;
Palomo, 1934; Lombard, 1937, 38; Lombard, 1947, 1948; Meier, 1950, in
particular).I will use the symbol * to indicate word boundarieswhere
relevant.
(17) (a)
Sea
significador daquel rey
Let-it-be indicative of-that king
por qual otra manera
quier
Cruz8oa
in what other manner
(one)-may-want
(b) Por fazer su
prouechoen qual quier otra manera
To do
his benefit in whatever other manner
Cruz 149B
Documentsfrom the periodfluctuatein the frequencyof use of the two
patterns. For instance, the Libro Conplido and the Libro de las Cruzes come
fromthe courtof AlphonseX, but the firstprefersthe olderpatternin (I 7a),
whichI analyseas [NP [e
NP ] [ SII COMP* qual * otramanera][squier]]], while
the second uses the innovativesequencein (07b) more often, revealingthe
Aragonese origin of its translator (see later):[, NPQp
qualquier #] otra
manera]. Therefore, I3th-century speakers have added qualquier(e) to the
medievallexicon,but retainclearindicationsthat it is compoundformedon
a wh-itemalso listedin the lexiconof relatives.
Older wh-itemssuch as qual have a double lexical classificationin this
period. When a lexical innovation enters the language, it acquires the
lexico-syntacticpropertiespresentin the paradigmitjoins. Qualquier
in ( 7b)
is identicalto the quantifierwhich survivesup to the present;in the 13th
century,it also functionsas a 'relative'much like qualdoes, as in (i8-I9):
(I8) (a) Quienquier llore o riya, el non a ninguncuydado
Whoever
cries or laughs,he not has any
worry
Apo 339
(b) E qualquierecasa d'ellas fuereflaca
Andwhatever house of them was weak
449
MARIA-LUISA
(i9)
RIVERO
muestra dannamiento e
infortunio
and bad-luck
shows damage
en las cosas que son d'aquella casa
Cruz 17 B
in the things that are of that
house
(a) E qualquier logar d'estos
sobredichos
And whatever place of these
above-mentioned
fallares
Cruz 53 B
fortunado, iudga...
you-would-find lucky,
judge...
(b) Pierda [el matador] quanto quier oviere
Lose
[the killer] whatever
he-would-have
(cited by (Cuervo I953:663) from the Fuero de Escalona 1223)
The previous examples show wh-compounds relating to the subject (i8), or
the object (I9), of a relative clause which appears in A-position (i8b-igb)
or non-A position (i8a-iga).4 In Modern Spanish the pattern is ungrammatical because it lacks a que 'that' next to the right bracket of the constituent containing the compound. The present-day version of (i8a) is
Quienquieraque llore o ria. Within the medieval system, (I8-I9) show the
expected relative use of wh-items, with quienquier in COMP: [NP [NPe]
S[ 1COMP
quien quieri#] [ti llore o riya]]].
is not unusual in 13th-century materials. In the Libro
I42A, i99A, 20I A, 256A).
In Aragonese documents it abounds too, as we shall later see. The modern
The pattern in (I8-I9)
Conplido,it is foundrepeatedly(I7B, 53B, 87A,
[4] The examples in (i8a-iga) are completely equivalent to the 'conditional' free relatives
mentioned by Posner (I985:197). Their interpretation should perhaps correlate with a
syntactic analysis where they are S'-structures with a - wh (i.e. non-question) rather than
a + wh (question) in COMP, and no antecedent NP-structure, as in (i); the conditional free
relative is in TOP.
S
(i)
TOP
SI
SI\
iudga....
COMP
s
[qualquier logar d'estos sobredichos]i
450
fallares t,
FREE RELATIVES IN OLD SPANISH
reader is inclined to think that the copyists were adamant in omitting que's,
but it seems to me that we are observing a syntactic construction which has
not survived, and not a series of systematic scribal errors. Cuervo (1953) has
noticed that the omissions of que in relative constructions are extremely
infrequent in the history of Spanish. This observation remains valid for the
13th century under the present analysis for (I8-I9), since these relatives
require no que. In later periods, the pattern in (i8-I9) disappears, but this
is not due to a syntactic change. Rather, when wh-items lose their double role
in the lexicon, wh-compounds survive as quantifiers exclusively, while
non-compounds are restricted to a relative use. Then the former must appear
in the NP-head, not the COMP, of a relative, a que or another wh-item must
necessarily follow them, and the pattern in (I8i9)
becomes ungrammatical
at that point. Thus, Cuervo's observation is valid for all periods, even though
older materials would seem to contradict it at first sight. The position of
clitics in examples such as (20) indicates that the wh-compound is the first
constituent within the S'-structure, as before. In other words, (i8) and (I9)
should not be analysed as having an empty COMP.
(20)
(a) E assi d'esta manera segundo
esta orden
And so of this manner according-to this order
iudgaras
en qual quiere
logar
you-will-judge in whatever
place
lo fallares
de las otras casas
it you-would-find of the other houses
Lib con 87A
(b) [s I COMPen # qual quiere # logar] [s lo fallares.. .]]
The wh-compound qualsequier(e) often found in Aragonese documents
mirrors the properties of the syntax still at work, providing additional
evidence for my proposals. Two patterns fluctuate in the 13th century, with
(21 a) the older one, and (2I b) the innovative sequence.
(21)
(a) E dierone
a don Fernan Pedrez, la meatad de
Andgave-to-him to Mr Fernan Pedrez, the half
of
casas,...
por qual guisa sequier
que sea
houses,.
for what manner itself-wants that it be
DLE 266 (1206)
(b) qual se quiere color que tiene de gerca
whatever
colour that he-has of near
tornase
ella en tal
turn-herself she in such
'She acquires whatever colour she has near.'
Alex
147
ib-d
The quantifier qualsequierin (2I n) originates in a syntactic structure with a
wh-item followed by the non-tonic reflexive and the verb quier(e). For the
reanalysis to have been possible, it must be the case that earlier periods
45'
MARIA-LUISA
analysed
Y the relevant sequenceas [s,[
SI f
NP INP
RIVERO
qual][ 5 se quier]], and not as
~~~COMP
[
e] [ se quier]]],in viewof the positionof se withinthe
qual][S,COMP
5
compound.If the latter analysis had been postulated, the expected result
whichis unattested.
would be qualquierese,
to adopt the syntacticpropertiesof wh-items
Again,we expectqualsequier
and show a 'relative'use. In documentswhereit appearsfrequently,this is
indeedthe case, as seen in (22):
(22)
(a) Alguna otra
iusticia, qual se quiere sea
F. de Ar. IIO
Some other justice, whateverit-is
(b) [NP [NPel [SI[COMP qual se quiere* ] [s seal]]
The compound qualqueoffers similarcharacteristics.As I pointed out in
Rivero (1984), it is often assumed that this lexical item is Occitan or
Aragonesein origin in its later uses as a quantifier.Within the medieval
system, there is reason to believe that it had a syntactic(internal)source
NP[qual][compque]S ] as othercompoundsdo; this would accountfor its
appearancein early Castiliandocumentswith no literaryvalue, and geographicallyremotefrom Aragonese.However,even if it was borrowed,in
medievaltimesit positionsitselfin COMPandlicencesa perverbalclitic.Thus
it entersthe system,and does not changeit:
(23)
F. de Ar. 126
(a) Et en qualque manera los ganare
Andin whatevermannerthem he-would-win
(b) [, IcoMp en * qualque* maneral][ los ganare]]
Wtithina common syntacticsystem, and a sharedmechanismof lexical
classification,dialectsmaydifferby developinga largeror smallerlexiconfor
a given paradigm.This, I submit, is what separatesOld Castilian and
Aragonesein relationto freerelatives.Aragoneseis richerin wh-compounds,
employsthem earlier,prefersinnovativepatternsof type (07b), compound
paradigmsalong the lines of qualsequierin (2ib), and uses qualqueto a
formed
extentthan Castilian.Considerthe compoundqualquequiere
greaLter
on the reanalysisof qualplus queplus quiere,or alternatively,qualqueplus
quiere,withinthe proposedsyntacticstructures.This sequenceis not found
in Castiliandocumentsin a way that would unambiguouslymotivatethe
assumptionthatit has enteredthe lexiconas a word.However,the samedoes
not apply to Aragonese in view of examples such as (24) where qualquequier
is a prenominal quantifier in a non-relative use.5
[5] The examples with this type of compound as an unequivocal word in the Fuero Juzgo cited
by Gessner (I895) and Cuervo (1953) must then be attributed to Aragonese influence.
452
FREE RELATIVES IN OLD SPANISH
(24)...
casas
dreytos y pertinen9ias que alas dictas
rights and possessions that to-the mentioned.houses
portaynen.. por qual que quieremanera
in whatever
manner
DLA 26 (1276)
pertain
If qualquequierand the related paradigmare Aragonese words, they
neverthelessacquirethe propertiesof the commonsystemsharedby the two
dialects,as we see in the contrastbetween(24) and (25):
(25)
(a) quantaque quier sea la demanda,
F. de Ar. 134
is the demand
whatever
* ] [s sea la demanda]].
(b) [ [ COMP* quantaquequier
If the Aragonesedialecthad continuedwhileCastiliandisappearedfromthe
literarydocuments(and if my proposalsare correct),we would observea
greatervarietyof wh-compoundssurvivingas quantifiersin laterperiods,but
the same syntaxwe see in laterCastilian.
4.
CONCLUSION
In conclusion,the qualitativeand the quantitativeevidencefor the proposed
system of relatives in the medieval period grows very rich as a larger
perspectiveof the overallgrammaris developed.Differentprinciplesconspire
to lead the learnerto selectone analysisover severalalternatives,and to add
apparentlyunrelateddata to the stock of positive informationrelevantto
relatives.Here I have concentratedon the contributionof clitics, and the
lexicalinnovationsof the wh-compoundtype.As this additionalinformation
is studied, the regular pattern of alternation quanto vs. quanto que first
discussedin Rivero (1984) becomes pervasive,and affects Castilianand
Aragonesein parallelways.
REFERENCES
A Primary sources and abbreviations
See Rivero (1984) in addition to the following:
Cor, page: Alfonso Martinez de Toledo. Arcipreste de Talavera o Corbacho (1438). Gonzilez
Muela, J. (ed.). Madrid: Clasicos Castalia (1970).
Cruz, page and column: Libro de las cruzes (I259). Kasten, L. A. & Kiddle, L. B. (eds.). Madrid:
CSIC (196I).
F. de Ar., section: Los Fueros de Aragon (c. 1300). Tilander, G. (ed). Lund: C. W. K. Gleerup
(1937).
Lib con, page and column: Aly Aben Ragel. El libro conplido en los ludizios de las Estrellas (c.
1254). Hilthy, G. (ed.). Madrid: Real Academia Espaniola(1954).
SME, line: La vida de Santa Maria Egipciaca (c. 12I5). Andres Castellanos, M. (ed.). Madrid:
Real Academia Espafiola (1964).
B. Secondarv sources
Chomsky, N. (1977). On wh-movement. In Culicover, P. W., Wasow, T. & Akmajian, A. (eds),
Formal syntax. New York: Academic Press. 7 I-I 32.
453
MARIA-LUISA
RIVERO
Cuervo, R. (1953). Diccionario de construccion y regimen de la lengua castellana. Bogota:
Instituto Caro y Cuervo.
Gessnier,E. (1895). Das spanischeindefinitePronomen.ZRPh 19. 153-159.
Hirschbhfiler, P. & Rivero, M. L. (1983). Non-matching concealed questions in Catalan and
Spanish and the Projection Principle. The Linguistic Revieu 2. 33I-363.
Lombard, A. (1938, 1939). Une classe speciale de termes indefinis dans les langues romanes.
SNPL ii. I86-209.
Lombard, A. (I947, 48). A propos de quienquiera.SNPL 20. 21-36.
Meie:r,H. (1950). Indefinita von Typus Sp. cualquiera,it. qualsivoglia. RF 62. 385-401.
Paloino, J. R. (I934). The relative combined with 'querer' in Old Spanish. HR 2. 5I-64.
Posner, R. (1985). Diachronic syntax - free relatives in Romance. JL 2I. I8I-189.
Ramrsden,H. (I963). Weak pronoun position in the early Romance languages. Manchester:
Manchester University Press.
Rivero, M.-L. (I980). On left dislocation and topicalization in Spanish. LIn II. 363-393.
Rivero, M.-L. (I984). Diachronic syntax and learnability: free relatives in I 3th-century Spanish.
JL 20. 81-129.
Rivero, M.-L. (I985). Parameters in the typology of clitics in Romance, and Old Spanish. To
appear in Language 62.4. (December I986).
Wackernagel, J. (1892) Uber ein gesetz der indogermanischen Wortstellung. IF I. 333-436.
454