You are on page 1of 82

THE LATINITY OF IBERO-ROMANCE

Hugh E. Wilkinson

For the general Romance philologist like myself, who has not
specialised deeply in any one particular branch, there is a special
fascination in Ibero-Romance (IR), by which I mean the various dialects of
Spanish and Portuguese, limiting the sense of the word to exclude Catalan.
In their modern forms these two languages present features which are
markedly different from the other Romance languages, including Catalan,
and this on points where the other languages agree in the main. For
example, there are only three conjugations, and these are all highly
regularised with hardly any irregular verbs and with only the auxiliary
haber (Ptg. ter) used to form the perfect; on the other hand they have
retained the Latin pluperfect and future perfect (now only vestigial in
Spanish). Then there are no -ir verbs with -esc- infix, but -ecer verbs with
-ec- throughout. The interrogative pronoun shows forms derived from
QUEM and CUIUS but not CUI. And there are many differences in vocabulary.
The questions that naturally spring to my mind are, did IR undergo a
separate development from earliest times, or at what stage or stages did it
start to diverge? The problem was made all the more interesting for me by
the fact that certain of the works I have read assume, apparently
unquestioningly, that certain features, such as the -ecer verbs and the
accented perfect ending -ieron, go straight back to classical Latin forms. I
felt inclined to question this assumption straightaway as there seemed to be

1
no reason for supposing IR to have been so much isolated from the koine
of the Empire, especially when the dividing line between IR and the rest
does not come at the natural barrier of the Pyrenees, since Catalan and the
mountain dialects of Navarre and Aragon do not share in the changes I am
concerned with, while Gascon in southern France has many Iberian
features.

Furthermore, it is established beyond dispute that IR belongs with


Gallo-Romance (GR) to the Western Romance (WR) group, and as such
had a long period of common development with GR, which must also
make us at once suspect any superficial derivation of Iberian forms from a
variety of Latin which was not the common WR. At the same time, we are
free to admit that this latter was not one unified language and that each
area would have its own local peculiarities, so that we have also to
establish to what extent we can be justified in attributing IR phenomena to
local retention of an earlier idiom, and how far we should seek an
explanation in the common tongue. Points of common development are,
for example, articles derived from UNUS and ILLE (traces of IPSE), reduction
of the case system, comparison expressed with the help of MAGIS or PLUS in
all but a few cases, and loss of a separate superlative form, adverbs formed
(especially in the West) with the use of MENTE, new compound
prepositions; in the verb the points of common development are as listed
below in the section dealing with the verbs. Besides these points of
universal agreement there are other special WR points; the tendency to
voice intervocalic voiceless consonants between vowels; the change of ē

2
and ō to ī and ū before an ī in the next syllable (metaphony); the early
development of a one-case system in the -a declension; the loss of the
neuter gender and the conversion of neuter plurals to feminine singulars
(this tendency is also found in Italian and Rumanian, where the neuter
plurals are, at the same time, preserved in their plural function).

This broad field of agreement in innovations points to a basic unity in


the spoken Latin of the Empire over a long period of time; just how long
can be seen from the fact that there are really no morphological
developments common to all the Romance languages of Western Europe
which are not also found in Rumanian. That is, linguistic unity coincided
with political unity. After that unity was broken, the amount of further
common development would depend on the closeness of contact.

The question of tracing the development of IR is naturally bound to


be a speculative one in view of the lack of records showing the transition
from Latin to the various vernaculars, but we can go back to the earliest
vernacular records we have and see what they show. This means not only
examining the earliest recorded IR forms, but, most important, comparing
them with the forms found in the other Romance languages; very often a
derivation which suits one particular language will not work when applied
to the cognate forms in other languages. Here the non-specialist is
sometimes at an advantage, in being able to step back and take the broad
view; it is the conclusions reached from such a viewpoint that I wish to
present in this paper.

3
Before proceeding to a detailed discussion, let us briefly look at the
history of IR and its relation with its neighbours.

History
Latin arrived in Spain with Scipio’s legions in 207 B.C., but for
several centuries it was only the language of an urban aristocracy, who
would naturally be conservative in their speech habits. Certain words, such
as cuyo and cueva, clearly derive, like Sardinian cuiu, from pre-classical
Latin, but unlike Sardinia Spain was not isolated from the rest of Romania
with the break-up of the Empire. Isolation came with the Arab-Berber
conquest of A.D. 711, and we may thus be justified in expecting IR to
share in the changes that took place commonly in Romance prior to that
date, especially if those changes are also found in Rumanian. We may
perhaps liken the position of IR in Romania to that of American English in
the modern English-speaking world; many features survive from the
earliest stratum, the English of the colonial period in America, which will
not be found in the English of Australia, for example. At the same time
America has been subjected to constant waves of immigrants from the
mother country (linguistically speaking), so that American English has not
developed as a separate language in the way that Afrikaans, for example,
has.

Spain, then, was subjected to waves of different varieties of Latin up


to the 8th century, when we can imagine isolation setting in. From then on,

4
on the one hand, the common Latin of that time would tend to be
preserved; on the other, local peculiarities would develop unchecked by
outside influences. What we are presented with in our earliest literary
records is the isolated language of that corner of Spain from which the
Reconquest started, which would be likely to be more differentiated even
than the language of those people in the part of Spain bordering the
Spanish March. Unfortunately our knowledge of the Mozarabic dialects is
slight. If it were greater, it is possible that we might find in them greater
affinities with the rest of Romania (we do indeed find loss of final -o as in
GR and Catalan); however, from what little we know of them they seem
already to have undergone the typical IR changes. For example, the word
vivireyu shows the -ir infinitive ending, and corajon and queris are typical
examples of IR vocabulary.1
We are thus faced with a language which can only be said with
certainty to have started on an isolated development in the last two
centuries before the emergence of the earliest written documents in the
vernacular, the Glosses of San Millan2, in the 10th century. It is, however,
highly probable that it started becoming a distinct language well before
then, developing the traits that were to mark IR off not only from its
nearest neighbours, but from the rest of Romania. Let us first examine its
relationship with its closest neighbour.

In contrast with Spanish and Portuguese, Catalan and the dialects


spoken in the Pyrenean areas of Navarre and Aragon which were not

5
occupied by the Moors (e.g. Jaca)3, while containing certain IR features,
adhere in general to the common Romance type on points where Spanish
and Portuguese differ. Thus we can see IR features in words like cova, nu,
uytubre > octubre (cf. Sp. cueva, nudo, ochubre > octubre) compared with
Prov. cava, notz, ochoure, and in the accentuation of the plural of the
imperfect indicative and subjunctive, and also in the early loss of the
distinction between nominative and accusative. But against this there is the
retention of the four infinitive types, the existence of an -ir conjugation
with -esc- infix, and in common with GR the loss of final -e and -o, and
together with all this a vocabulary that is closer to French than Spanish;
compare Cat. voler, pregar, trobar, parlar, menjar, formatge, buit, petit
with Fr. vouloir, prier, trouver, parler, manger, fromage, vide, petit and Sp.
querer, rogar, hallar, hablar, comer, queso, vacío, pequeño. Also Cat. tot,
regeu and Fr. tout, raide go back to Latin forms *TOTTU, RIGIDU, whereas
Sp. todo and recio come from TOTU, *RICIDU; in terms of sound change
Cat. molt agrees with OFr. mout rather than Sp. mucho.

As we proceed further west and south into Navarre and Aragon we


find in the old texts a meeting of two distinct types, Spanish and Catalan,
rather than a gradual shading from one to the other. “Standard”
Navarro-Aragonese of the Spanish type has diphthongisation of e and o,
infinitive types and -ecer verbs following the Spanish pattern, the 3rd
person singular of the perfect ending in -ó, presence of final -o and partly
of final -e, the numerals for the ‘tens’ ending in -enta, possessive pronouns

6
taken from the Latin strong forms, the infinitive ser and present
subjunctive sea for the verb ‘to be’, confusion of -ir and -er verbs in the
perfect paradigm, retention of the Latin future perfect, and a Spanish
vocabulary typified by the use of querer for voler. But interspersed in
these texts can be found GR features, such as forms without diphthongs,
e.g., poble, tenda, bonos, grossa; words without final -o, especially names,
Montagut, Monclus, Pere, poble, comandament, sagrament, palazet, cens,
quarter, castiell, arcebispe, cavaliers, aucelles, bell, pas (in some cases
the final -o has been replaced by a neutral vowel of support written -e); the
numerals vint(e), trenta, quaranta, cinquanta, xixanta, setanta etc.; the
infinitive esser and subjunctive sia; a marked absence of the future perfect
and its replacement by various other tenses; and words of the GR type such
as trobar, bueyto, pregar. However, certain other features are restricted to
the dialects of the Pyrenees: infinitives of the type creistre, perdre, treire,
prendre, defendre; a clear-cut -ir conjugation with infix; perfect 3rd person
singular in -a, not -ó, as mena, though -ie, as against -ió, as in establie,
murie (presumably from -EDIT), is also found elsewhere; feminine
possessive forms such as ma, sas, from the Latin weak forms; -ir-, -iss- in
the perfect paradigm of the -ir verbs; voler, not querer.

The first thing that strikes one about this language frontier is that it
coincides with the extreme limit of permanent Moorish occupation in
north-east Spain east of the Basque-speaking country; as if the area beyond,
after being cut off from the rest of the Spanish-speaking territory, and
turning towards France politically during the period of the Spanish

7
March, established by the Franks, and the French-orientation of the
unoccupied Pyrenean region, also became Gallicised in its speech. But this
conclusion, while there may be some truth in it, is probably too simple.
Firstly, if that area had originally begun to produce peculiarly Spanish
features during the centuries following the break-up of the Empire they
would have been more apparent in the earliest surviving forms of the
language, and the non-Iberian features would appear superimposed,
instead of integral as they are; secondly, it is hard to suppose that IR
became differentiated from the rest of WR as a result of an isolation that
took place in the mere two centuries preceding the emergence of the
earliest documents. We must therefore suppose a closer affinity between
NE Spain and Gaul dating from an earlier period. This may be found in the
5th century when the Visigoth Athaulf and his successors, after moving
into Spain, were also established in southern Gaul until their eviction by
the Franks in 507. Of equal, and perhaps greater, significance is the fact
that the diocese of eastern Catalonia was under the jurisdiction of the
metropolitan see of Narbonne. We can thus imagine the differentiation of
IR to have started from this time, with no perceptible dialect boundary to
begin with, but a gradual transition; then, with the Moorish occupation and
the subsequent polarisation of the dialects on either side of the boundary
around different centres (with Catalonia now absorbed into the
French-oriented Spanish March), the differences become more marked. La
Rioja set the standard for Navarro-Aragonese as Barcelona did for Catalan;
when the former became Castilianised this also had its effect on the

8
standard Navarro-Aragonese which gradually came to have more and more
Castilian features such as -ch- for -it-, -j- for -ll-, h- for f- etc., as can be
seen from the mediaeval documents.

We come now to the problem of how closely we can date the


appearance of distinct IR features. The Latin of Spain in Imperial times,
while containing certain local words, of which St. Isidore of Seville gives
the most instructive account, in general illustrates features which are
common to all Romania, and often goes against modern Spanish usage;
thus inscriptions contain the words AUNCOLO, SERORI, NEPOTA. This is
only what we should expect in view of the general Romance nature of IR,
which we shall analyse in detail later. As far as the Common Romance
ancestor of the major modern languages itself is concerned, it seems to
have been remarkably homogeneous in its morphology, all the major
changes from classical Latin having taken place before the isolation of
Rumanian. We are only left with phonological differences, in which field
IR shares a common development with GR and northern Italian for a long
time, and differences of vocabulary, which are always a most striking
feature of regional varieties of any language. To take the example of
English again, morphological differences are negligible throughout the
English-speaking world, but pronunciation and vocabulary vary greatly,
being affected by a variety of factors. The difference between the situation
in present-day English and that in the Latin of the Empire (apart from the
existence of communications media which work for standardisation all the
time) is that in English the written language sets the standard for the

9
spoken language, whereas in the Roman Empire all areas seem to have
spoken broadly the same kind of Vulgar Latin (to use this outdated but
convenient term to denote the vernacular Latin in use, and steadily
evolving, during Imperial times) in spite of its being distinct from the
written norm.

As we have no record of the Romance spoken during the time of the


Empire and the centuries following its break-up, our only indicator can be
the extent of common development, and it is this field that I wish to
examine, with regard to vocabulary, phonology and morphology, in
particular the morphology of the verbs. Many scholars, including
Menéndez Pidal, seem to derive certain Spanish verb forms directly from
the classical Latin forms, without apparently finding this inconsistent with
the fact that IR so patently shared in the same Common Romance
development as the other languages, perhaps because they lay too much
stress on the fact that Spanish Latin dates from the year 207 B.C. and that
Spanish in some instances clung to a vocabulary that was superseded
elsewhere. For this reason I wish to set out my own conclusions, for what
they are worth, in the hope that I may thereby stimulate others, who are
better qualified, to pursue the subject further.

Vocabulary

10
Let us see first of all what light the vocabulary sheds on the problem.
In this realm we find that IR very often has a different word from the one
used in Catalan, French and Italian. For example, Ptg. falar, Sp. hablar:
Cat. parlar, Fr. parler, It. parlare; Ptg., Sp. comer: Cat. menjar, Fr.
manger, It. mangiare/manicare; Ptg. achar, Sp. hallar: Cat. trobar, Fr.
trouver, It. trovare; Ptg. vazio, Sp. vacío: Cat. buit, Fr. vide, It. vuoto. Or
else they have a different form of the same word: Ptg. frio, Sp. frío: Cat.
fred, Fr. froid, It. freddo; Sp. recio (Latin *RICIDU): Cat. regeu, Fr. raide,
It. rigido. Navarro-Aragonese is the border area; mediaeval texts from this
region show manducaret (Glos. Sil.)4, trobar, bueyto (evidently with t
from a form *bueyt with the consonant unvoiced in the final position).
Comparison with Rumanian shows that in some cases Spanish has
preserved the word that was probably in general Romance use during the
time of the Empire; to hallar corresponds Rum. afla, and to recio, rece. On
the other hand Rumanian has mânca, and even Sardinian has mandigare,
so that here we must suppose that both words were current (Rumanian has
comând for a ‘funeral repast’, reflecting comedere) and that each area
picked on the one it preferred. In other cases we can see clear examples of
neologisms in French and Italian, such as caille/quaglia, though Italian
still preserves dialectal cotornice. Yet again, the neologisms may be
clearly Spanish, such as querer for voler (only found in certain pronominal
forms in Berceo, such as sivuelque), some of them having developed in
“historical” times, such as tener, tomar, salir for haber (as a full verb),
prender, exir. In a large number of cases it is clear that each area has

11
settled for one of several near synonyms which were all used concurrently
at one stage from the fact that the other word is preserved dialectally or
with a change of meaning, or has died out within historical times. So
against Sp. arena we can find It. (a)rena, OFr. areine; against hablar, OIt.
favolare, OFr. fabler; Sp. río, It. rio; Sp. hermoso, It. formoso, Prov.
formos; Sp. hermano, It. germano, Fr. germain; Sp. corazón (from
*COR-ACEU-ONE), It. coraccio (*COR-ACEU, in a pejorative sense); Sp.
cabeza, It. capezza/cavezza, Sard. capitha; Sp. queso, It. cacio/caso; Sp.
trigo, It. tritico; Sp. pedir, OIt. petire; Sp. rogar, It. rogare; Sp., It. calle.
Conversely, traces of the alternative word are found in Spanish: Galician
najo, It. naso; OSp. pregar, It. pregare; OSp. arribar, It. arrivare; OSp.
tiesta, It. testa; OSp. camba, It. gamba. There are also some words
common to Spanish and French which do not appear (or only have a
limited use) in Italian: llorar/pleurer, cara/chère, acabar/achever (Italian
forms plorare (learned, but cf. OPied. piorer), cara, accapare are recorded
but not general).

Granting the fact that in many cases it was a question of choosing


between two or more words which existed concurrently with a similar
meaning, we are still faced with the fact that in general France, Italy and
Catalonia use one word, where IR uses another (though southern Italian
often agrees with IR; compare cacio/caso, asciare and crai with Sp. queso,
hallar, Ptg. queijo, achar, OSp., OPtg. cras). In fact we are back at the
same boundary as discussed above in the paragraph on History. One
puzzling point here is that, though the French influence was resisted as

12
far as such words were concerned, the Germanic words introduced into the
Western Empire seem to have entered freely. Because of their wide
diffusion, Elcock (in The Romance Languages, p. 208 ff.) dates the
introduction of such words as appear in Spanish in the forms guisa, blanco,
bruno, bastir, guardar, guarnir, guarir, fornir to the period of Vulgar
Latin unity; but others, such as OSp. cosir and fardida, with its f
representing the aspirate h retained in northern France, are clearly Frankish.
With regard to the Germanic -ir verbs, such as bastir, we shall see later
that the special IR development of the -ecer verbs must have taken place
after the introduction of these verbs, replacing the form bastir by bastecer.
Similarly it would appear that these Germanic borrowings preceded the
loss of the two-case system, as shown by the nom. and acc. forms in OSp.
compaño/compañón, Cat. company/companyó (a translation-loan from
Germanic), the existence of two case forms for Germanic names (see
below), and the plural form gasalianes in a Latin document of 804
(Gifford & Hodcroft, No. 1) as compared with Sp. (a)gasajar from a nom.
form *gasalia.

One likely possibility is that the original vocabulary did not coincide
with the present one, but that the French terms were introduced into
Catalan (as they were also partly into Navarro-Aragonese) after Catalonia
became French-orientated.

The general evidence of the vocabulary shows, therefore, that though


there were regional peculiarities in Imperial times, they were not

13
inconsistent with the idea of a basic Latin unity, and we cannot decide on
the evidence of vocabulary alone at what point a distinct Spanish language
emerged. All we can say is that, insofar as there is a language boundary, it
coincides with the morphological boundary, but not entirely, as Catalan
has some IR vocabulary (germà), and Provençal is often the meeting
ground for the two types, having, for example, both cest, cel and aquest,
aquel.

Phonology

Let us first examine some aspects of the phonology of WR as


compared with ER as exemplified in Italian and Rumanian. Firstly the
intervocalic voiceless stops have become voiced, and then undergone
various further changes: Latin VITA, Ptg., Sp., Cat., Prov. vida, Fr. vie, but
It. vita, Rum. vită; Latin RIPA, Ptg., Sp., Cat., Prov. riba, Fr. rive, It. ripa,
and riva (from a dialect in the WR area), Rum. râpă; Latin PACAT, Ptg., Sp.,
Cat., Prov. paga, Fr. paie, It. paca/(WR) paga, Rum. pacă. The cause of
this change is not known, but the same phenomenon is observable in the
Celtic languages (originally spoken in the WR area).5 Another change
which has also affected the Iberian languages and French, and to a lesser
extent Provençal, is also found in the Celtic languages and Basque (see
Entwistle, The Spanish Language, p.17) and in the Germanic languages,
and that is the change of intervocalic voiced stops to fricatives. In French,
Catalan and IR both voiced and voiceless Latin stops were affected (the

14
voiceless stops less consistently so in Portguese), in Provençal only the
voiced. The fricatives coming from D tended to disappear in WR, though
other solutions are also seen, such as Prov. vezer from VIDERE, Cat. peu
from PEDE. Latin B fell in with V, and followed the same development; G

before a back vowel was preserved in some cases. As far as the voiced
sounds coming from the Latin voiceless stops are concerned, we find them
preserved in Iberia and also in Provençal, while in French they were
subjected to further changes. Here the dental was lost (cf. vie above), and
the velar was palatalised after a front vowel (cf. paie above) and
disappeared before and after back vowels (seur > sûr from SECURU, feu
from FOCU, fétu from *FESTUCU). The bilabial was changed to the
labio-dental [v] (a more stable sound) as in rive, or lost before an accented
u, as in seu > su from *SAPUTU. Intervocalic Latin B and V must have
changed to bilabial [β], and this sound is retained in Spanish, some local
variations of Catalan, which evidently follow the Spanish style, and the
south-western dialects of Provençal; elsewhere, including in standard
Portuguese, the fricative has become a labiodental [v]. It is not clear
whether or not the areas with [β] once also changed to [v] and then
reverted after the b coming from Latin P (cf. riba above) changed from a
stop to a bilabial fricative, a process which took place during the 16th
century. But the chances are that [β] remained unchanged, as it is not
unreasonable to suppose that this bilabial fricative was still prevalent
throughout Romania at the time of the isolation of IR, in fact it may help to
account for the confusion in French between Romance v and Germanic w

15
which produced gué, guêpe etc. (Another IR feature which probably
survived till late in France is the Latin future perfect.)

Another WR feature is the change of Latin CL to a palatal [λ] as


opposed to [kj]: Ptg. olho, Sp. (Leonese, Aragonese) uello, Cat. ull, Prov.
uelh, Fr. oeil as against It. occhio, Rum. ochiu, from Latin OCULU. Again,
for DI and Greek Z they show a palatal continuant as opposed to a dental or
palatal affricate: Ptg. raio/raia/raiar/rajar, -eja, Sp. rayo/raya/rayar, -ea
(from -eia), Cat. raig/rajar, -eja, Prov. rai/raj/rajar, -eja, Fr.
rai(s)/raie/rayer, -oie, as against It. raggio/raggiare, -eggia (ER) and
razzo/razza/razzare, -ezza (WR), Rum. rază/răza, -ează, from Latin
RADIU/*RADIA/RADIARE, -IZAT. TI and CI both produce dentals in WR; in
Italian there are many derivatives of TI, but CI has everywhere produced a
palatal affricate, whereas in Rumanian both sounds have produced the
same dental affricate. In IR, as in WR, TI in intervocalic position originally
produced a voiced sound and CI (which became CCI in late Latin) a
voiceless, but in modern Spanish these have fallen together, and there
appears always to have been some confusion between the two, as is
apparent from Portuguese (cf. poço < PUTEU, aguçar < ACUTIARE).

Examples of TI are: Ptg. -eza (beside semilearned -iça), Sp. -eza, Cat. -ea
(beside -esa), Prov. -eza, Fr. -ise (beside irregular -esse pointing to *-ITTIA),
It. -ezza, Rum. -eaŃă from the Latin abstract ending -ITIA. Examples of CI:

Ptg. ameaçar, Sp. amenazar (OSp. amenaçar), Cat. amenaçar, Prov.


menassar, Fr. menacer, It. minacciare, Rum. ameninŃa from Latin
*MINACCIARE. Original IR kept to the WR voicing of intervocalic

16
single consonants; the later unvoicing of Spanish s, z, j was a local
phenomenon which need not concern us here.

One more common WR feature is the change of Latin CT to -it- or


-ch-: Ptg. oito, Sp. ocho/hueyto (Arag.), Cat. vuit, Prov. uech/ueit, Fr. huit
as against It. otto, Rum. opt from Latin OCTO. Another is the retention of
final -S and -T (this latter disappearing in all areas, however, at an early
date). This final -s has become a special WR feature in the formation of
plurals (from the old accusative) and the 2nd persons of verbs; in ER the
nominative forms are generalised throughout the plural, and in the singular
of the verbs new forms have had to be evolved to distinguish the 2nd
person from the 3rd.

Again in WR a long i in the final syllable affected an accented close e


or o in the preceding syllable, changing them into i and u; compare Ptg.
vinte, fiz, pus, OPtg. fuste, Sp. veinte, hice, puse, OSp. conuve, fuste, Cat.
vint, fiu, OCat. fust, Prov. vint, fis, conuc, fust, Fr. vingt, fis, OFr. pus,
conui, Fr. fus, with It. venti, feci, posi, conobbi, fosti (Latin VIGINTI, FECI,

*POSI, *COGNOVUI, FU(I)STI).

Another tendency which operated in the various WR languages at


different periods was the loss of the unaccented post-tonic vowel in
trisyllabic or longer words. That this tendency did not operate at the same
time or with equal effect in all areas can be seen by comparing such words
as Ptg. manga, homem, Sp. manga, hombre, Cat. mánega, home, Prov.

17
manga/marga, ome/omne, Fr. manche, homme, from Latin MANICA,

HOMINE. ER was not affected: It. manica, uomini, Rum. mânecă, oameni.
In connection with the phonology one may also note that Portuguese
shows that there were variations even within the Vulgar Latin of Iberia
itself. Compare Ptg. fome (Rum. foame) with Sp. hambre (Latin FAME,

*FOME, *FAMINE), pêssego with prisco (Latin PERSICU), faça with haga
(Latin FACIAT), chantagem, fuligem, seja, veja with llantén, hollín, sea, vea,
from Latin PLANTAGINE, FULIGINE, SEDEA(M), VIDEA(M) (in these last
examples Portuguese presumably derived from forms with a doubled
consonant, cf. Cat. plantatge, It. piantaggine, fuliggine, seggia, veggia).

It is noteworthy that there are very few features common to all the
Iberian dialects which are peculiar to Iberia alone. The most striking,
perhaps, is the alternation between stops and fricatives in the pronunciation
of b, d, g, though this is not universal. It seems to be significant that this is
also a feature of Basque, as are both the lack of f which is a notable feature
of Castilian (an interesting parallel is found in Japanese where bilabial f,
from an earlier p, has changed to h before every vowel except u, cf. the
Spanish retention of f in fuego, fuerte etc.), and the occurrence of
palatalised l, n, and t, which may also have had an effect on the Spanish of
certain areas. In many respects Portuguese has remained closer than
Castilian to the original IR consonant system, and thus more nearly
resembles French, having [s], [ʒ], [z], [z], [∫], [λ], from original [ts], [dʒ],

[dz], [z], [∫], [λ], where Castilian has [θ] (via [s]), [χ] (via [ʒ], [∫]), [θ] (via

18
[z], [ð]), [s], [χ], [χ] (via [ʒ], [∫]); compare Ptg. cinco, junco, vizinho, rosa,
deixar, olho with Sp. cinco, junco, vecino, rosa, dejar, ojo. Portuguese also
keeps f, cf. falar as opposed to hablar < FABULARI, and also shows the WR
development of Latin J, GE in janeiro < IANUARIU, giesta < GENISTA, jazer
< IACERE, as opposed to Sp. enero, hiniesta, yacer. The conclusion to be
drawn from the phonology is that IR had a long history of common
development with GR.

Morphology

Nouns and adjectives

Some sort of case system survived throughout the Empire till quite a
late date, as is evident from late Latin documents. Moreover, a case system
still exists in Rumanian, mainly dependent, it is true, on the changes in the
form of the article; nevertheless, feminine nouns have genitive-dative
forms distinct from the nominative-accusative in the singular. In the other
languages there are what I suspect to be fragmentary remains of this dative
with adverbial or prepositional use in words like It. vie, or, allor, ca’, Fr.
or, alors, chez, Prov. or, laor, chas/encò, Cat. llavors, cal (< ca el), Sp.
cas/ca, OPtg. cas, reflecting VIAE, HORAE, CASAE. And a fossilised genitive
singular has been preserved everywhere in the names of the days of the
week (except in modern Portuguese, which uses different words), while
the genitive plural in -ORUM, generalised in Vulgar Latin (mesoru,

19
martyrorum in inscriptions, fratrorum in the Cartons des Rois)6, has
survived in place names and a few other sporadic forms. In the Western
Empire the possessive function of the genitive was partly taken over by the
undifferentiated oblique case (representing the Latin dative, cf. the
Rumanian use of the genitive-dative), and this usage is found in Old
Spanish just as in Old French and Old Provençal. (See Orígenes del
Español, p. 376.) The main point of interest for our purposes is to see how
far IR concurred with GR in preserving this nominative and
accusative/oblique two-case system, in which GR evidently reflects the
state of things in the later stages of WR, and also what common WR
features it shows in its plural formations.

To judge from the evidence of the Romance languages, the five


declensions of Latin were reduced to three, the fourth falling in with the
second and the fifth with the third (but generally with substitution of -IA
for -IES, notably in -ITIA for -ITIES). Again, on the evidence of ER and GR
we conclude that the third declension formed a new nominative plural
ending in -i for the masculines and perhaps also for the feminines. (In the
singular of the third declension, we already find in late Latin that the
monosyllabic nominatives have, as a rule, given way to disyllabic ones,
thus pedis, bovis, dentis for pes, bos, dens.) The neuters we will discuss
later. Then ER and WR diverged because of the loss of final -s in the
former; ER merged the nominatives and accusatives, taking the accusative
form in the singular (from which the nominative had become

20
indistinguishable) and the nominative form in the plural (though the
situation in Italian is not clear, as there are dialectal instances of plurals in
-a and it is noteworthy that the velars are not palatalised before the final -e,
compare amiche and amici; could there have been an intermediate stage -ai
between -a and -e ?), whereas WR kept the two separate, at least in the
case of the masculines. The GR position as reflected in OProv. is :

Masc. Sing. Pl.


Nom. -s –
Acc. – -s
Fem.
Nom. -a; -s -as; -s
Acc. -a; – -as; -s

In the masculines the 2nd and 3rd declensions have been merged; the
two feminine types represent the 1st and 3rd declensions respectively. One
significant innovation will be seen, and that is the extension of the
accusative -as ending to the nominative in the plural of the 1st declension.

If the same situation had been found in OSp., we could have expected
the forms:

Masc. Sing. Pl.


Nom. -os; -es -e

21
Acc. -o; -e -os; -es
Fem.
Nom. -a; -es -as; -es
Acc. -a; -e -as; -es

This accords in part with the situation actually found; that is, the -a
declension follows this pattern, and the other feminines differ only in that
the nominative singular has lost its -s. This situation is also found in Old
French, where it is not even clear which is the earlier form, the one with -s
or the one without it. For the rest, it can be imagined that IR would be
quicker than GR in rejecting the -e form from the -o declension, just as
WR was quick to remove the -e forms (from Latin -AE) from the -a
declension, with substitution of forms in -s in both cases. In these two
declensions the vowels -o and -a would be felt to be integral to the stem
and thus inviolable. In the case of GR, where final -o and -e were lost, the
two-vowel contrast did not arise, so there was no reason to interfere with
the inherited situation. As far as the masculines with -es in the nominative
singular (from the Latin third declension) are concerned we can imagine
that they followed either the example of the feminines or that of the
remodelled -o stems. We can thus imagine the next position in IR to have
been:

Masc. Sing. Pl.


Nom. -os; -es (?-e) -os; -es

22
Acc. -o; -e -os; -es

We now have a situation nearer that of the feminines, where -s is


becoming a mark of the plural. All that was needed to produce the
historical situation was to eliminate the -s of the nominative singular
(already absent from the Latin -ER nominatives), and this happened at
some point before the emergence of documents in the Iberian languages. It
is interesting that the same process took place in GR at a later date. The
question that remains for IR is to try and establish how late these changes
took place.

In conjunction with this it is worth while considering one other Latin


type, the imparisyllabic nouns, which also appear in GR. They are in the
main words which refer to persons, all other nouns of this class having
become parisyllabic, as noted above. Their interest in this connection lies
in the fact that where forms derived from the old nominatives appear in IR
it is also in words referring to persons, or in some cases to other living
creatures. It seems as if there is a natural tendency to feel the need to
differentiate subject from object formally in the case of persons; modern
Spanish perpetuates this with the use of the preposition a to indicate the
object, and Rumanian with pe (compare the Russian use of the masculine
genitive in similar circumstances). Vulgar Latin also created new forms for
the oblique cases of words referring to persons (notably masculine nouns
ending in -A) and proper names; a pagan inscription (pre-4th c.) has IOLENI

23
from IOLE, and a Christian inscription FORTUNATANEM. In the modern
languages French has nonnain beside nonne and Rumanian tătâni-su
beside tată. OSp. has Wamba/Wambán and Cintila/Cintillán, and OPtg.
Sintião in Germanic names, but the tendency is evidently earlier than the
introduction of Germanic names, though the latter must have given it a
new boost, as in GR too.7

Nominative forms surviving in Spanish are, for example, Dios (beside


accusative Dio in Jewish Spanish), names like Carlos, Marcos, Oliveros,
and, in the imparisyllabic nouns, fray (as against the accusative seen in old
fradre, confradre > cofrade), sor (with OSp. acc. seror), preste8, sierpe
(with loss of final -s and flanked by acc. serpiente)9, drago (acc. dragón),
buho (Arag. acc. bobón), gorgojo, pavo (acc. pavón), OSp. ladro (acc.
ladrón) etc., as well as the previously quoted compaño/compañón;
Portuguese has Deus, Carlos, Marcos etc., frei ((con)frade is presumably
an oblique form, like Sp. cofrade), sor, preste, serpe/(acc. serpente),
drago/(acc. dragão), bufo, gorgulho, pavo/(acc. pavão), ladro/(acc.
ladrão), demo/(OPtg. pl. demões), tredo/(acc. tredor). Pai, mãe (earlier
mai) also look like the nominatives corresponding to the accusatives padre,
madre, rather than pet names. Italian parallels these with frate (fra before
consonants; old acc. confratre), suora/(old acc. sorore/se-),
draco/drago/(acc. dragone), prete (old preste)/(old acc. prevede),
serpe/(acc. serpente), ladro/(acc. ladrone), OIt. traito/(acc. traitor, now
traditore), pate/(acc. patre/padre), mate/(acc. matre/madre), and has a

24
number of other old nominatives such as sarto, moglie, uomo. Rumanian
has soră/(gen., dat. surori), frate, drac, preot, şarpe (şerpe), ORum. lotru,
and other forms such as împărat, om, while French and Provençal have
similar pairs in their two-case system. When we look to see what has
happened to the case system in Catalan we find this language taking a
middle position between IR and GR in that the earliest texts preserve a few
nominative forms such as Deus still consciously used as nominatives (acc.
Deu); apart from this, there are other surviving nominatives, such as fra
(beside fra(i)re, confrare), sor, serp, old drach (beside dragó), lladre
(beside old lladró), hom ‘one’ (beside home ‘man’) and the above
company/companyó. With this amount of concurrence among the different
languages we may be justified in concluding that a two-case system was
everywhere preserved till quite a late date, at least for living things. It is
also worth nothing that both ILLE and ILLUM are represented in the various
forms of the article el, lo, o.

When we come to the neuters we find that WR, unlike ER, reduced the
three genders of Latin to two, most of the neuters being absorbed into the
masculines. However Rhaeto-Romance has unitary plurals like la bratscha
‘the (two) arms of the body’, dua bratscha ‘two ells’ from Latin BRACCHIA,
and Old French has similar combinations of a numeral with a former
neuter plural to give a measurement, as in the cinquante care ‘fifty
cartloads’ of the Roland, from Latin CARRA ‘carts’. A plural sense is also
traceable in the Spanish Castilla, which goes back to CASTELLA ‘castles’.
Other neuter plurals, especially those designating objects which could

25
be taken in pairs or groups, became feminine singulars, like Fr. brasse, Sp.
braza ‘fathom’, Fr. corne ‘horn’, Sp. cuerna ‘antler’, and Sp. hueva ‘roe’,
a process which was not limited to WR. Some 3rd declension neuters,
which did not fit into any existing type, were later remodelled; thus, COR,
LAC, SAL, MEL, FEL, OS, VAS became *CORE, LACTE, SALE, *MELE, *FELE,
*OSSO, VASO. (This is shown by the existence of final e in Italian and
Rumanian and diphthongs in French and Italian.) In the case of the other
third declension neuters, some areas generalised the nominative/accusative
form and some the oblique. Those ending in -R appear in Spanish in the
oblique form: roble, mármol, pebre (contrast It. marmo, pepe). The neuters
in -US originally kept their final -s in IR as in WR, thus OSp. uebos, pechos,
peños, tiempos, cuerpos, but these were soon felt to be plural forms and
new forms without the -s were created for the singular. In most cases there
is no trace of the oblique form or the old plural (as found, for example, in
OIt. pettora, corpora and It. Tempora ‘ember days’, and extended widely
in Rumanian), but prenda < PIGNORA ‘pledge’, Témpora exist as singulars,
and modern Spanish has estiércol, with the extended stem seen in
STERCORA, in contrast to OSp. estierco, Ptg. esterco, It. sterco. Spanish also
has the two forms polvo and polvora, while Italian has polve, polvere and
French has old pous now poussière, poudre (Cat., Prov. pols, pólvora,
polvera). In the neuters of the NOMEN type Spanish differs from
Portuguese and Catalan and the rest in generalising the oblique form: Sp.
nombre, Ptg. nome, Cat. nom.10 Other Latin words were attracted to this
type, giving *FAMEN, *SANGUEN, *CONSUETUMEN, and the corresponding

26
oblique forms, which appear in Spanish: Sp. hambre, sangre, costumbre,
as against Ptg. fome, sangue, costume, Cat. fam, sang, costum.

The only possible clue to the dating of the loss of the case system in
all this is to be found in the Germanic nouns (including compaño and
gasalianes, discussed under Vocabulary) which appear with case forms.
This sets an earliest time limit, but the probability is that the case system
survived till quite late, in view of the traces appearing in Catalan.

With regard to the adjectives, IR partook in the Vulgar Latin tendency


to adapt the 3rd declension adjectives to the -US/-A type; compare Spanish
agro/agra with “ACER NON ACRUS” in the Appendix Probi. In the
formation of the comparative and superlative, IR now uses MAGIS where
French and Italian use PLUS, but the distinction was not always clear cut,
as OPtg. had chus and OCat. pus, while the Glos. Emil. use plus and maius
together. In modern Provençal both forms are still used. In the superlative,
IR follows the general Romance practice of using the comparative form
with the article. Catalan forms a bridge between IR and GR in having
formerly had the comparative adverbs mils/mills and pitz/pis and then
replacing them by millor and pitjor as in Spanish (mejor, peor).

Numerals

These are developed from the Common Romance (CR) forms


UNU/UNA, DUOS/DUAS (with rare remains of DUA), TRES (with rare

27
remains of TRIA), *QUATTRO (for *QUATTOR by metathesis or from use
before a vowel, written quatrum in the Formulae Andecavenses), *CINQUE
(with assimilation of the ending to that of cuatro; OPtg. still has examples
of cinque) etc.; WR *ŬNDECIM, CR *DODECIM (dodece in a Christian
inscription), WR *CATTORDECIM etc. SEDECIM persists in OSp. sedze, OPtg.
seze, thus bringing the older stages of IR into line with WR usage. OSp.,
OPtg. veínte, viínte with i by metaphony as in WR and dissimilation of the
two i’s in Spanish. Treínta, triínta by analogy (an etymological treyenta in
the Fuero Juzgo). Cuarenta, quarenta from a CR form without the d (cf.
qarranta in a Christian inscription) and with qu- maintained, in contrast to
catorce, catorze, because of the secondary accent. The forms in -enta
(earlier -aenta) are another example of Iberian retention of earlier forms of
Latin. OSp. has nonaenta from the CR form, later replaced by novaenta,
noventa. (For the Aragonese forms in -anta see under History.) For the
‘hundreds’ the general tendency in Romance was to create new analytical
forms, but OSp. dozientos, trezientos, Ptg. duzentos, trezentos agree with
Prov. dozent, trezent, OIt. dugento. Spanish and Portuguese are alone in
retaining quinientos, quinhentos.

The old Romance ordinals up to ten were generally retained, if


sometimes with a specialised use. Spanish, like Provençal and Catalan,
also formed new ordinals with the Latin suffix -ENUS.

28
Pronouns
a) Personal.

IR shows the CR distinction between conjunctive and disjunctive


forms, and the use of ILLE for the 3rd person (and definite article). In
general the modern languages (but not Rumanian) have rejected one or
other of the accusative and dative forms of the 1st and 2nd persons in the
disjunctive position. IR (and OCat.) retains the dative in this function in
contrast to French, Italian etc. However the dative mi appeared in certain
areas in Old French, and southern Italian has meve, teve, seve. This meve,
like the Mozarabic mibi, comes from a Vulgar Latin MIBI, which probably
ousted MIHI in the spoken language; modern Spanish mi could come from
MIBI as i from IBI and o from UBI — as can Rum. mie, with normal loss of
intervocalic B.

In the third person IR and Catalan show no trace of the VL masculine


nominative singular *ĪLLĪ, with the first I lengthened by metaphony in WR
(unless it is to be seen in the il which, with isti for este, occurs in some
texts of the Fuero Juzgo, and is also found in Galician which, like Asturian,
also has iste, ise and aquil), though a later formation elli occurs, or of the
accented datives *ILLUI, *ILLAEI, ILLORUM. This *ĪLLĪ may have been of
limited diffusion; its use in Provençal is restricted, as is that of similar
forms in the demonstratives, and furthermore it is generally used here as a
feminine rather than a masculine form. (Spanish meismo > mismo seems to

29
have an i coming from *ĪPSĪ, with the metaphonic Ī, but this may be a
loanword from old French, which was more likely to have had derivatives
of *ĪPSĪ.) The datives were of wide diffusion, appearing in all the other
areas which are relevant to our theme. We cannot say whether or not they
were ever used in Iberia; we can only observe that the dative CUI is
likewise absent, and this must have been in existence formerly (there are
just slight traces of cui in Old Catalan, as also of lui and lei, though these
are all in texts which show Provençal influence), and there must at least
have been some accented dative form in the 3rd person as long as it was
necessary. Except in Rumanian, these datives had lost their dative force
and become an alternative to the accented accusative; IR opted for the
accusative (the nominative in the case of el, êle) and so the datives became
unnecessary. Nor were they needed, as in French and partly in Italian, to
supply new conjunctive forms, as the phonetic conditions in IR did not
bring about the fusion of the accusative and dative forms in the plural. So
the fact that these dative forms are not found need not make us necessarily
suppose that they never existed.

It. glielo finds its counterpart in OSp. gelo, Ptg. lho (from which a
new dative lhe was derived; cf. similar formations mhe, che, xe in the other
persons). Gelo became confused with se lo when ge became pronounced
[∫e]. As with the nouns, the nominative/accusative case system has been
lost, but el, êle come from the old nominative, and the latter has produced
a new plural êles. OSp. and OPtg. also had INDE and IBI (en/end(e) and i)

30
used pronominally as in the rest of the Western Empire.

b) Possessive.

Early Latin had weak forms MUS, TUS, SUS, SA, SUM, as well as the
regular forms, and similar forms are attested later for the other parts of the
possessives. In Spanish, new weak forms were derived from the
apocopated feminines coming from MEA, TUA, SUA; Old Spanish had
mio/mi(e), to/tu(e), so/su(e). Old Portuguese has weak forms ma, ta, sa,
which seem to go back to the Latin weak forms. (These weak forms also
occur in the Catalan-type Pyrenean dialects, as in Catalan itself.)

SUUS was probably the general third person possessive in the West,
though in Rumania it is only a reflexive (singular and plural), and the use
of ILLORUM for the plural probably came later. (Sardinian has issoro, but
the open quality of the accented o and the presence of final o make it seem
as if this was a late formation on the model of It. loro.) Catalan has llur,
and Aragonese lur, but they probably came in from Provençal, where lur is
an unaccented variant of lor.

c) Demonstrative.

IR preserves the original Latin distinction of three points of reference


in (aqu)este, (aqu)ese, aquel. The same three forms are found in OSard.
custe, cusse, cullu, and in southern Italian, while Italian itself has

31
produced a new series of three, questo, codesto and quello. French and
Rumanian have reduced the series to two, as has Catalan, though it
preserves the three forms, so it seems that IR perpetuated what was
probably the general Romance state of affairs. As in the case of ILLE, there
is no sign of the dative forms found elsewhere; here again Provençal is in a
zone of transition, having very few occurrences of such forms. The
combination ECCE HOC does not occur in IR, but HOC itself exists in pero
and in o que in the Cid (1. 1136); here Catalan has the neuter pronoun ho,
and the variants açò, això, earlier also ço, axò, aquò.

The pronouns of identity Sp. mismo, mesmo are probably taken, as we


have seen, from the Gallo-Romance forms, which also spread to Italian.
The native forms are seen in OLeon. simediso, OSp. misso (in the Crónaca
General) and OPtg. medês, which are comparable to Prov. mezeis/mateis,
Cat. mateix/old meseis/meseix. Interesting also are OSp. eleisco, OPtg.
elesso, OCat. ele(i)x from ILLU IPSU, which are to be compared with Rum.
acelaş. In this latter the -aş is invariable and can only come from a form
with a yod (Latin X gives s in Rumanian); the Rumanian derivatives of
plain IPSE are însu-, însa-. This may indicate that the Spanish and Catalan
forms (and also the Cat. aqueix, mateix) come from forms with a yod; it
also leads one to speculate whether the invariable ipsius in Tardif’s
Cartons des Rois was accented on the first syllable.

d) Interrogative-Relative.

32
In the unaccented forms of the relative the nominative qui eventually
gave way to que in Spanish; it is not found at all in Old Portuguese.

Like the personal pronouns, the CR interrogative and accented


relative (without an antecedent or after a preposition) originally had three
cases, from Latin QUI, QUEM, CUI, and the various areas in the Western
Empire dropped one of the oblique ones; here again IR differs from the
rest in its choice, retaining the accusative, Sp. quien, Ptg. quem. (These
forms, like Rum. cine, must surely go back to a Latin form with a final
supporting -e which still survives in Rumanian. It is true that to produce
the Spanish diphthong it is not necessary for there to have been an open
syllable, but the history of QUEM must have been the same as that of REM,

which produced a diphthong in French rien; a simple vowel is seen in the


corresponding OPtg. rem, OCat. ren, Prov. re(n), these being areas which
did not diphthongise. The Rumanian ending -ne was extended to the
accented accusatives of the personal pronouns, giving mine, tine, sine.)
Catalan in common with the other languages retained cui, under Provençal
influence, but this soon gave way to qui, as it did later in French.

IR also differs from the rest (but agrees with Sardinian) in its use of
the possessive of the relative, retaining CUIUS in its pre-classical form as a
fully declined adjective. The classical Latin invariable CUIUS is nowhere
found, having been ousted in the other parts by the “dative of the
possessor” CUI (‘Fons cui nomen Arethusa fuit’). The same grouping is

33
found in the use of ALIENUS as opposed to *ALTERUI. It looks as if here
there was a variation dating back to Imperial Latin times.

In the other pronouns there is no special separate development to


comment on. OArag. res, ren, taken from Catalan, had a negative function
like Fr. rien. Nadie, nada have parallels in OProv. nat, nada, hom nat, It.
anima nata.

Adverbs, prepositions, conjunctions.


IR shares in the common formation in the Western Empire of adverbs
with MENTE, originally not agglutinated to the adjective, as also in Old
Provençal and Old Italian. The only example of a MENTE adverb in
Rumanian is aimintre/aiminterea (from ALIA MENTE), which also shows
the intrusive r that appears in OSp. mientre, perhaps coming from the
Latin -ER of adverbs formed from third declension adjectives (cf. OProv.
seguentre, auzentre, vezentre), or the -RE of QUARE and other words
formed with the use of RE. A similar r is found in other forms such as OSp.
ajubre, Ptg. alhur(es), Prov. alhor(s), OCat. alloure, Fr. ailleurs, Rum.
aiure(a) from ALIUBI (found in the Cartons des Rois, Muller & Taylor, pp.
202, 204); this comparison shows up the faultiness of the derivation from
ALIORSU generally given for Fr. ailleurs, a mistake which should have
been apparent anyway from the vowel in French. The final -s in some of
these forms is also a Common Romance feature in adverbs, shared by IR.

34
In the prepositions IR shows the general Romance tendency to create
new compound prepositions. In Vulgar Latin, those beginning with DE-

must have come to be used for prepositional phrases dependent on a noun,


as is the case in Rumanian; Latin eschewed “homo in luna”, but Vulgar
Latin found “homo de in luna” acceptable. (A similar construction
obtains in Japanese, where such phrases have to be linked by use of the
possessive marker no.) In the older stages of the Iberian languages such
familiar Romance words appear as suso, yuso, arriedro, (en)tro a, which
were later superseded. The conjunctions correspond to regular Vulgar
Latin formations.

Reference may be made here to the use of the prefixes MALE-, MINUS-
and DIS- in a negativing sense, in which IR follows the general pattern. It
has been said that the French mes- and for- (appearing in Italian as mis-
and for-/fuor-) come from Germanic, but the existence of Sp. menoscabo,
menospreciar, Cat. menysprear (as OFr. meschief, Fr. mépriser, OIt.
mispregiare), and the forfecho of the Fuero Juzgo, OCat. forsfeit (as Fr.
forfait, OIt. forfatto), together with the ‘foris facio’ quoted by St. Isidore,
show that the origin is Romance, whatever influence Germanic may
subsequently have had. Another point to be mentioned is the partitive use
of de in the old stages of the Iberian languages, similar to the use still so
widespread in French and Italian; thus ‘no nos darán del pan’ in the Cid (1.
673) and also comer de, beber de, with similar usage in Old Portuguese.

35
Verbs.

It is when we look at the verbs in modern Spanish and Portuguese that


we are most struck by the contrast with the other Romance languages, and
it is to this subject that I wish particularly to address myself. The position
we find is: three conjugations instead of four, or, to put it better, three
infinitive types, seeing that it is only Rumanian that preserves any other
distinction between the 2nd and 3rd conjugations; a virtual absence of -SI
and -UI perfects and their replacement, in Spanish, by what appear to be
4th conjugation forms; as these same 4th conjugation endings are also used
in Spanish in those parts of the perfect paradigm of the remaining “strong”
verbs which are accented on the ending, a remarkable degree of uniformity
has been achieved; the same uniformity to be seen in the past participles of
these verbs, nearly all being formed in -ido; analytical perfects all formed
with the use of HABERE and none with ESSE; the preservation of the Latin
pluperfect and future perfect indicative; a change of accent in the 1st and
2nd persons plural in certain tenses; and above all the absence of any -ir
conjugation with infix, balanced by the presence of an -ecer type. In
addition, there are certain other changes belonging to the internal history
of the languages on which no light is thrown by a comparison with the
other languages, and which thus remain outside the scope of this paper; I
am referring particularly to the process of metaphony (both in Spanish and
Portguese) which has led to the divergence between the types sentir/siento
and servir/sirvo in Castilian, and sentir/sinto, pedir/peço in Portuguese,

36
and for which it is so difficult to find a satisfactory explanation.

I propose to examine all the parts of the verb in turn, to show how the
earliest stages of the languages followed the common Romance pattern,
but first let us see the points where the modern Iberian languages still
agree with the rest.

a) General Vulgar Latin changes.


IR partook of all the changes common to the Romance languages.
Apart from certain changes which I shall discuss below, there was firstly
the loss of the passive endings and their replacement by analytical forms
made up of parts of ESSE plus the past participle, or by a construction using
the reflexive. At the same time the deponent verbs took the forms of the
active. The future was lost, becoming phonetically unviable, and was
almost everywhere replaced by HABERE plus the infinitive (in addition to
other constructions). In Rumanian, where the infinitive was replaced by
the subjunctive except in a few special cases, the future is formed with
HABERE plus the subjunctive; however, in the conditional both
agglutinated and unagglutinated forms exist which seem to consist of the
infinitive plus some form of HABERE (though it can also be argued that the
verb concerned was *VOLERE). Rumanian also has another future derived
from *VOLERE plus the infinitive or the subjunctive, like modern Greek thá
from thélo hína; compare the use of shall and will in English. The present
participle was limited in use to an adjectival function, and even here had a

37
rival in -TOR or -TORIU added to the thematic vowel a, e, i and not formed
from the past participle; the other functions were taken over by the gerund.
The imperfect and perfect subjunctives disappeared in most areas,
becoming indistinguishable from the future perfect indicative; it is argued
that the former survives in the Portuguese “personal infinitive”, but all the
evidence is against this. A new analytical perfect and pluperfect were
created, using HABERE or ESSE plus the past participle; ESSE was used in the
case of intransitive verbs, conceivably on the model of deponent forms
such as NATUS ESSE, MORTUUS ESSE or constructions like ITUM EST (in late
Latin we can find OBITAE SUNT as MORTUAE SUNT, PROCESSI ERANT as
PROGRESSI ERANT, DEVENTI ESSENT as (CON)VENTUM ESSET; Väänänen, § 342).
The Latin perfect forms were refashioned, with the virtual elimination of
the reduplicated perfect and the extension of -SI forms and others discussed
later. Apart from Rumanian, the Romance languages replaced the supine
by the infinitive, and in general, apart from Sardinian, the imperfect
subjunctive by the pluperfect subjunctive, and introduced certain other
changes discussed below. IR partook of all these changes, and so it is
dangerous to speak as Elcock does (The Romance Languages, p. 141) of a
Vulgar Latin belonging to a later period than that of Spain, at least as far as
verb formations are concerned. I will now proceed to a discussion of
various particular forms with a view to accounting for the differences
between present-day Spanish and Portuguese and the other Romance
languages and showing IR to have sprung from a common Latinity in this
regard.

38
b) Infinitive.
In IR we find the old -ĔRE verbs not only in the -er but also in the -ir
conjugation, as hacer, decir. There are also some scant traces of the former
state of affairs; in Old Spanish the infinitive fer is found, which must be
derived, as Italian fare, French faire etc., from a Latin form *FAGERE, with
substitution of GE for post-tonic CE (cf. Fr. plaid, It. piato with the same
vowels as OFr. fraile, It. frale < FRAGILE; also vide, vuoto etc. from
*VOGITU for *VOCITU (VACARE); note too, conversely, Sp. recio, Rum.
rece from *RICIDU for RIGIDU, with Rum. răcoare corresponding to RIGORE,
this already used by Chiro to mean ‘cold(ness)’, and the Vulgar Latin
confusion in the spelling DIGITUS/DICITUS). The future forms haré, diré,
OSp. aduré (Ptg. farei, direi, OPtg. adurei) also contain unaccented forms
of infinitives with the same substitution of GE for CE but with loss of this
middle syllable (cf. It. farò, Friul. farai, Fr. ferai, Prov. farai, Cat. faré
from *FARE HABEO, and so also for *DIRE and, in part, *ADDURE). Apart
from these and the similar Ptg. trarei, the only traces of the old infinitive
are those found in derivatives of VIVERE which appear in place names (see
Orígenes del Español, p. 356). However, mediaeval texts from the
Catalan-type dialects of the Pyrenean regions of Navarre and Aragon
contain the infinitives defendre, creistre, treire, prendre, perdre, but these
forms cannot really be considered Iberian in the sense in which I am using
the word. Navarro-Aragonese also has esser from *ESSERE, but there is no
indication of where the accent falls.

39
In the case of the new -ir infinitives for old -ĒRE verbs in Spanish,
some have parallels in other parts of Romania, but these changes have not
always been universal. For instance, beside Sp. lucir we have Ptg. luzir,
Cat. lluir, Prov. luzir, Fr. luisir and luire, Rum. luci, It. lucére, rilúcere,
Sard. allúgere; beside henchir, Ptg. encher, Cat. omplir, Prov., OFr. emplir,
It. émpiere and empire, Rum. úmple and umpleá. (These -IRE conjugations
may have arisen by association with -ESCERE forms, see below.)

IR certainly participated in such general Vulgar Latin new creations


as *FUGIRE, *MORIRE, *CAPĒRE, *CADĒRE, *SAPĒRE, Sp. huir, morir,
caber, caer, saber. What is not clear from the modern forms is whether it
also originally participated in the other trend found in all the other
languages — including Sardinian, where -ĔRE has completely ousted -ĒRE,
— of changing certain -ĒRE verbs into -ĔRE. By and large, Vulgar Latin
“regularised” the 2nd and 3rd conjugations by relegating all the verbs with
-UI perfects (cf. CAPERE, CADERE, SAPERE above and also *VOLERE,
*POTERE) to the 2nd conjugation, and those with -SI perfects (often new
formations) to the 3rd — I use grammatical terms to describe a process
which occurred, of course, by spontaneous analogy without any conscious
sense of the grammatical categories into which we classify the Latin verbs.
Thus we have *RESPONDĔRE/*RESPOSI, *RIDĔRE/RISI, *MORDĔRE/*MORSI,
*TERGĔRE/TERSI, *TORCĔRE/TORSI. (This was not carried to its ultimate
conclusion, however, so that we have REMANĒRE/*REMA(N)SI and
CRESCĔRE/*CREVUI etc.) Seeing that this was such a widespread change, we
can be justified in surmising that IR also took part in it, and see where

40
that leads us.

It is interesting to speculate on the cause of the change in the


infinitives in Spanish. Such a change cannot come about arbitrarily, nor
can analogy alone (the attraction of the minority to the majority) account
for it, to judge by the other languages where it is mainly the verbs in -CERE
and -VERE that have been affected and the change has more often been
from -ĒRE to -ĔRE. Nor have these languages made any similar changes
from -ĔRE to -IRE. It remains to be seen, then, what part was played by
phonetic causes. An examination of the hypothetical forms the
etymological descendants of the -ĔRE infinitives would have taken in
Castilian produces some interesting results, which make it appear not
unreasonable to suppose that phonetic causes brought about the death of
this infinitive type.

Regular phonetic evolution in Castilian would have produced such


forms (besides fer) as (and I leave them unstarred): trel (by dissimilation
as in cárcel, roble etc.), ler, crel, dir, piérdel, muérdel, ril, tiérzel, tuércel,
espárzel, mueldre, porne or pondre, cínzer or cindre, tínzer or tindre
(perhaps even trinde, cf. prenda), destrúir, adúir etc. from *TRAGĔRE,
LEGĔRE, CREDĔRE, *DIGĔRE, PERDĔRE, *MORDĔRE, *RIDĔRE, *TERGĔRE,
*TORCĔRE, SPARGĔRE, MOLĔRE, PONĔRE, CINGĔRE, TINGĔRE, *DESTRUGĔRE,
*ADDUGĔRE, and perhaps also, with analogical influence, ber, escrir, vir
from BIBĔRE, SCRIBĔRE, VIVĔRE (cf. Fr. écrire, boire by analogy). As is seen,
the multiplicity of forms is great, especially as the number of words

41
with r in the root syllable is high, so a sufficient number of infinitive
endings would have undergone change to have destroyed the sense of a
unified conjugation. In one case, destrúir, the normal change to a rising
diphthong has produced the modern form accented on the i. Nor would it
take much to change fer, trel (trer), ler, crel (crer), dir, ril (rir), adúir, ber,
escrir, vir into fazer, traer, leer, creer, dizir, riir, aduzir, bever, escrivir,
vivir, by remodelling on the stem, thus producing the (early) historical
forms divided between -er and -ir according to what was originally the
stem vowel. This would also set the pattern for the present distribution,
where, with few exceptions (later borrowings from Latin also excepted),
verbs with the vowels i or u in the 1st person singular of the present have
their infinitive in -ir, and those with e, o, ie, ue the infinitive in -er. It is
true that this is all purely speculative, but the results are interesting, to say
the least.

c) Present indicative and subjunctive; imperative.


The main difference from the other languages observable in IR is in
the 3rd person plural, where -ENT replaced -UNT and -IUNT from Vulgar
Latin times (cf. vadent, colligent, custodent in Peregrinatio ad Loca
Sancta). The same phenomenon is seen in Catalan and partly in Provençal.
It is not clear whether both forms existed in French; font and OFr. dient
must come from *FACUNT and DICUNT, while plaisent seems to come from
PLACENT, though it could be analysed as the stem plais- plus a standard
termination derived from -UNT. Italian and Rumanian have both

42
generalised -UNT, with varying treatment of the preceding consonant.

Old Portugese shows a greater retention of -IO, -IA(M) forms than the
other languages (in most of which they are only found after a single
consonant), producing arço, perço, menço, senço, dormio, servio in stems
with two consonants where the other languages only have forms derived
from *SENTO, *DORMO, *SERVO etc., though the I appears in one French
participial form, the substantival sergent (which may be a semilearned
borrowing), and -IAMUS, -IATIS are the basis for all the 1st and 2nd person
plural forms in the subjunctive in French and Italian. (Ouço, like the oza of
the Fuero Juzgo, comes into this category, as the U of AU- acted as a
consonant in IR, where preconsonantal AU and AL coalesced, so the ç is
parallel to that in verça, garça, old vergonça.) In the case of arço, perço,
the forms with ç must be due to analogical extension of the yod (cf.
parallel forms ard/arz, pierd/pierz in Rumanian), but the others seem to be
etymological, in which case we can only suppose Portuguese to have been
exceptionally conservative. It is certainly more conservative than Castilian
in retaining faço, faça and posso, possa. In the latter case other dialects
show similar forms; possa, posca in Leonese, poscan, puescan, puysquen
in Navarro-Aragonese. These dialects also preserve sia from *SIAM where
the rest have replaced it with derivatives of SEDEAM. Other dialect forms
are Leonese día, estía, paralleling Italian dia, stia, Prov. and OCat. estia,
Rum. dea, stea.

43
As in French, Provençal and Italian, there are a few sporadic remains
of the old third conjugation in the 1st and 2nd persons plural. (One factor
that would hasten their demise was the existence of future forms
identical with the present of the 2nd conjugation; when DICIS, DICIT became
identical with DICES, DICET, DICEMUS would be likely to oust DICIMUS

because of its resemblance to PLACEMUS.) The forms found in Old Spanish


are femos, feches, and the imperatives fech, tred, from *FAGIMUS, FAC(I)TIS,
FAC(I)TE, *TRAGITE; the modern froms vamos, vais probably do not go back
to VADIMUS, VADITIS but replaced earlier imos, ides. Old Portugese has
treides (used as a imperative) and the imperative form treide from
*TRAGITIS, *TRAGITE. Both languages also have derivatives of the
imperatives DIC, DUC — di, adu; haz, faze are re-formations, though OPtg.
has fa < FAC, and OSp. has fay, seemingly from *FAGE, while dialectal Sp.
fes may come from a similar form plus -s, like OCat. fe, now fes (cf. Rum.
fă, zi, adu). In the imperative in general IR is more conservative than GR
in deriving the plural forms from the Latin imperative and not the present
indicative.

Also noteworthy is the use of the present subjunctive as the negative


of the imperative. Here there is a clear cleavage in Romance. Old French,
Italian and Rumanian use the infinitive in the 2nd person singular and the
imperative in the plural (in French and Italian this is, however, identical
with the subjunctive in the verbs given below). Provençal (most areas) and
Catalan side with IR. In the other persons the use of the subjunctive is, of

44
course, universal. For the affirmative of the 2nd persons, I have established
that IR, at one time, also shared in the French, Italian, Rhaeto-Romance,
Provençal and Catalan use of the subjunctive for the imperative in the
verbs HABERE, *VOLERE, SAPERE, *ESSERE (also found to some extent in
Rumanian and Sardinian), with the substitution of QUAERERE and SEDERE

(in the subjunctive) for *VOLERE and *ESSERE; however, imperative forms
were also in use alongside these, and are now the only ones in use. The
reason why the subjunctive was used in these verbs is that the imperative
in such cases is a wish rather than a command.

d) Imperfect.
IR preserves the VL endings -ea, -ia, also found in Old Italian
alongside -eva, -iva, with the usual change of e to i in hiatus; these endings
are widespread throughout Rumania in the -ERE and -IRE conjugations,
suggesting an early loss of the B. The variants in -ee, -ie, also found in the
various languages in the subjunctive formed from *SIA as well as in the
imperfect, also seem to be widespread in Romance, being found in Old
Italian, Provençal and Old Catalan; in French, OFr. -eit from -eet is the
basis of the regular modern forms. The change of accent in the 1st and 2nd
persons plural in IR probably spread from the perfect paradigm, and will
be discussed later. It is also paralleled in Catalan and occasionally in
Provençal, and occurs in many Italian dialects. Aragonese notably has -eva,
-iva rather than -ia. Wherever forms with -v- occur, analogical reformation
on the basis of -ava has been suggested, but it seems to me that the forms

45
with -v- may have come from Latin forms accented as -ÉBAM and those
without -v- may have come from those accented as -EBÁMUS, so that one
had, for example, HABEBAM, *HABE(B)AMUS, with later levelling in both
directions.

e) Future and Conditional.

Though these tenses are formed in almost all the Romance languages
from a common original of infinitive plus HABERE, and have all arrived at
agglutinated forms in the modern languages (except Rumanian, see above,
and Sardinian), the process of agglutination took place at different times in
the different languages, being slowest in Iberia, especially in Portugal.
Unagglutinated forms comparable to those found in Old Spanish and old
and modern Portuguese can also be seen in Old Provençal and Old Catalan.
IR follows WR (GR) usage in selecting the conditional formed from the
infinitive plus the imperfect of HABERE. In modern Italian the conditional
derived from the infinitive plus the perfect of HABERE is used, but the other
also existed in the old language, and still exists in the northern and
southern dialects.

f) Present participle and gerund.

The position in early Romance is not clear but many languages


(Rumanian, Sardinian, Portuguese and Catalan) show gerund forms

46
derived from -AND-, -END-, -IND- (Rumanian -ând represents both -AND-
and -END-). Others (Italian, Provençal and probably Spanish and French)
derive from just two forms -AND- and -END-. However, other participial
forms seemingly based on -IEND- occur, such as Fr. sachant, vaillant,
puissant, but these have probably been formed on the stem of the
subjunctive. It is therefore most likely that Spanish forms like sirviente,
sirviendo do not come from the classical Latin forms (note that SERVIO,

SERVIAM etc. do not survive), but were remodelled on those of the -er verbs,
seeing that this process has gone so far in the other parts of the two
conjugations that they now differ only in the infinitive (and its derivatives)
and the 1st and 2nd persons plural of the present indicative, as far as the
endings are concerned.

g) The Latin perfect.


Latin -AVI. The standard endings all derive from a common Vulgar
Latin -AI, -ASTI, -AUT, -AMUS (? -A?MUS), -ASTIS, -ARUNT. The form of the
3rd person singular agrees with the Italian form, whereas French,
Provençal, Catalan and Rumanian have forms without U. Aragonese seems
to have both: compare duplicaot and betait in the Glosas Silenses and a
form mena from Huesca (in a Catalan-like form of language) which
appears to be a perfect; betait may be a Latinised spelling of a Romance
vedé with the -é extended from the 1st person rather than representing any
inherited Latin form, as there are no traces of -AIT elsewhere in WR, only
in Sardinian. The 1st person plural is identical in form with the present

47
(but is given a different pronunciation in the Portuguese of Portugal); all
the other languages made changes in one or other tense, or both, in order to
avoid confusion. The forms in French and Italian suggest that there may
once have been another consonant (the D from DEDIMUS?) before the m,
and the single m in the Iberian forms would not necessarily be inconsistent
with such a possibility, but in any case the forms in the two tenses must
have been similar enough in many places to make for confusion. (I am
supposing that -AV(I)M- would have given -aum- > -om-, and that *-AMM-
would not have given French -âmes.)

Latin -IVI. The IR forms derive in the main from the Vulgar Latin -II,
-ISTI, -IUT, -IMUS, -ISTIS, -IRUNT. As with -ó from -AUT, Sp. -ió, Ptg. -iu from
- IUT agree with an Italian form -io, which is dialectal but is also found
extensively in Dante, for example, while they differ from the other
languages and also, in this case, standard Italian, where no U is found. The
change of accent in the Spanish diphthong is phonetic; Portuguese -iu
retains the old accent. The same observations apply to -imos as to -amos.
In the third person plural Portuguese and many Spanish dialects derive
from Common Romance -IRUNT: Ptg. ouviram, Asturo-Leonese oyron,
pediron, Arag. partiron, Pyrenean dialects establiren. Other dialects,
including Castilian, have -ieron, which will be discussed in connection
with the -EDI perfects.

Romance -EDI. This derivation is generally accepted for the

48
weak preterite of the -er verbs, though Menéndez Pidal and Nunes
strangely make no reference to it. It is assumed to have started from
perfects like VENDIDI with the accent moved to the root syllable of the
simple verb and its root vowel adopted, as happened with other compound
verbs, cf. *RETÉNET for RÉTINET. The second D was then lost between
vowels or assimilated to a following consonant in WR. The accented e
seems to have been open, judging from the diphthongisation in Sp. and
OFr., but in modern Portuguese and Catalan it is close, as also in Italian.

In the 1st person singular, Portuguese, Spanish, Catalan and Old


French have the ending -i. In the case of the last three this can be explained
as coming from an earlier -iei diphthongised from -ei (in Catalan this
necessitates supposing that the -i acted as a yod); the ending -iei is also
found in Provençal alongside -ei with open e. In Portuguese (and perhaps
rather in Catalan) it seems as if the -i must have come by metaphony from
-ei, as fiz from FECI; perhaps the e became close to come in line with the e
of the infinitive etc. In the 2nd person we must suppose an original WR
-eísti with accented i by metaphony; this would account for the i in Old
Portuguese, Spanish, Old Catalan and French. Modern Portuguese and
Provençal have substituted the vowel of the third person. In the third
person an open e seems to have been the original. Old Portuguese has -e
which then became -eu (by analogy), with the e becoming close according
to the normal phonetic development in Portuguese. Catalan has -e,
similarly becoming close in the modern language. Old Provençal has -et

49
with open e and t from -DIT. Old French has -iet similarly formed. Old
Spanish has various forms, -iot (Glos. Sil.), -io, -ieu (Arag.), -eo (Ast.), all
with final o/u added by analogy with the other weak perfects; Aragonese
also has -ie without u. -Ió came from -ieu as Dios came from Dieus, or old
mio from mieu.

In the case of the first person plural, the -EDIMUS form may give us a
clue to the process whereby the accent in the strong perfects changed in
Romance. It is usually said that DÉDIMUS and the other perfects in -ĬMUS
changed to DEDÍMUS etc. on the analogy of the second person, but this does
not account for the occurrence of earlier diemos, diestes, preceding dimos,
distes. From DÉDIMUS we should expect *DED’MOS; perhaps this actually
happened and this -D’MOS is the basis of the -mes in French and the -mmo
in Italian, which then spread to the other conjugations. In the case of
Spanish, *DE’DMOS would naturally have given the OSp. diemos, which
perhaps then influenced the change from DE(D)ISTIS to diestes. Conversely,
*VID’MOS, which naturally gave vimos, could have been affected by
VI(D)ISTIS > viestes, to give an old form viemos as an alternative to vimos.
This vimos, for its part, generated vistes, and these set the pattern for dimos,
distes. The corresponding French -EDIMUS, -EDISTIS forms must originally
have contained ie, but French then generalised i throughout the paradigm,
in line with the 1st and 2nd persons singular, and -îmes, -îtes became the
endings of all the ‘regular’ perfect forms of verbs like vendre. In Spanish,
-iemos, -istes became the endings in all the ‘weak’ perfects like vendiemos,

50
vendiestes, and were extended from these to the strong perfects like OSp.
oviemos, oviestes; these endings were then confused with the -imos, -istes
(later -isteis) of the -ir verbs (but not in Asturo-Leonese or Aragonese)
because of the identity of the singular endings. In the modern language
both conjugations have -imos, -isteis, but in the old language we find
-iemos, -iestes in the -ir verbs as well as the -er verbs. Similarly, in the 3rd
person plural, it is natural to suppose that the -ieron coming from
DÉDERUNT > dieron became the standard ending in the -er verbs, and was
extended, in Castilian, to the -ir verbs, where it matched the singular -ió.
(In Aragonese, the -ieu, -ie of the 3rd person singular was also extended to
the -ir verbs.) I do not think it likely that the Spanish forms -iemos, -iestes,
-ieron could have come from Latin -IIMUS, -IISTIS, -IERUNT, as we have no
evidence for such forms having survived in Vulgar Latin, and also II would
have changed to Ī and IE naturally changes to ee > e, as in PARIETE > pared,
just as UO changes to oo > o, as in DUOS > dos.

It is also stated that -ieron and Portuguese -eram, in both strong and
weak perfects, are derived from forms containing the classical Latin
ending -ĒRUNT. This again, I think, is highly unlikely. To begin with, as we
have seen already, IR is so obviously derived from the same post-classical
Latin as the other Romance languages that it is dangerous to start
postulating special derivations for such a common verb part, where all the
other languages derive from forms based on -ĔRUNT11. We may even
question whether -ĒRUNT ever reached Spain, having possibly developed

51
(as a cross between -ĔRUNT and -ĒRE) after the first introduction of Latin
into Iberia. Next, there is no trace of it in -aron, which comes from
-AVĔRUNT, already paired with -ARUNT (which may be the earlier form) in
pre-classical Latin, or in Ptg. -iram, Sp. dial. -iron or in OSp. foron, Ptg.
foram, all of which agree with the forms in the other Romance languages.
Also the diphthong -ie- is not confined to this person but is also found in
the future perfect and pluperfect, which had no Ē, so we have to find a
derivation which fits all these forms. In fact, the diphthong -ie- suggests
derivation from an accented open Ĕ in Latin; nor would Latin DEDĒRUNT

have given dieron but *deeron/deron, as seer/ser from SEDĒRE. It therefore


seems to me most likely that dieron from DÉDĔRUNT (cf. dial. viron from
VÍDĔRUNT) is the starting point for the -ieron ending, and that this type was
analogically extended in Spanish to the -ir verbs and the strong verbs. Let
us now examine how this process took place.

Portuguese keeps three separate types, -eram with open e (strong -er
verbs), -eram with close e (weak -er verbs) and -iram (-ir verbs). The
dialects bordering on Portuguese show two, -ieron (or undiphthongised
-eron) for the strong and the weak -er verbs and -iron for the -ir verbs,
while Aragonese similarly has -ieron and -iron. Texts from the Pyrenean
parts of Navarre and Aragon (in which there is no diphthongisation) also
have -iren for the -ir verbs. It seems as if IR originally had three inherited
types, like the other Romance languages, but the original root-accented
forms (see the next paragraph) disappeared at an early date; the confusion
of all three types is limited to the area which diphthongises, and where

52
also old -ĔRE verbs with an -EDI perfect, such as rendir, have been
transferred to the -ir conjugation. In this area vendí, vendiste, vendió
agreed with partí, partiste, partió, and rendiemos belonged to an -ir verb
rendir, setting the pattern for a wholesale merger. The pluperfect
subjunctive followed the vowel of the other tenses, as it has in the other
Romance languages.

In the case of the strong verbs the change of accent from the root to
the ending in both Spanish and Portuguese (supposing an original
*HÁBUĚRUNT, *MÍSĔRUNT as the base of the IR forms as of those in the
other Romance languages) can be explained in two ways. One is to
suppose that they followed the influence of the future perfect HABUÉRIMUS,
HABUÉRITIS, which I think is not so likely, as the rest of the paradigm, and
the pluperfect indicative, did not have accented Ĕ. The other, which seems
to me much more likely, is to suppose that once again the -EDI verbs
provided the type on which these ones were remodelled. We have already
seen that the 1st person plural forms like DÉDIMUS, VÍDIMUS may have been
the starting point for a change in the 1st person plural of the strong
perfects; the strong perfects and -EDI perfects also already agreed in their
accentuation of the 2nd persons singular and plural, and of the pluperfect
subjunctive, so it would be quite natural if the 3rd person plural of the
perfect and the whole of the pluperfect and future perfect indicative
followed suit. There is also support for this theory in the fact that the the
same thing happened in Catalan and Provençal. In the earliest Catalan
documents there were forms such as hágren, ágra, vólgra, and these

53
were soon remodelled to become haguéren, haguéra, volguéra. Similarly
in Provençal tráistron gave way to traisséron, while French also shows
examples of strong verbs remodelled with -EDI endings, e.g.
conduisis–conduisirent, naquis–naquirent. Italian too has -ei perfects such
as potei, dovei, which replaced the potti, dibbi coming from -UI forms; the
starting point here was forms like potesti, dovesti coming from the -UI
perfect (with loss of U) but being indistinguishable from those of the -ei
one. Portuguese -eram in the strong perfects keeps the original vowel of
-ĔDI, in contrast to the weak perfects of the -er verbs, which, as I have
suggested, may have adopted the close e of the present (and also of -eu).

-SI perfect. The endings have in the main been discussed already. As
in the case of other strong perfects, Spanish uses -o as the third person
singular ending to distinguish it from the first person; in the early stages of
IR these two persons were distinguished in some cases because of the
metaphony which took place in the root vowel of the first person, but
Spanish extended this vowel throughout, giving 1st and 3rd person forms
such as prise/priso, respuse/respuso, puse/puso from Latin *PRESI/-IT,
*RESPOSI/-IT, *POSI/-IT. Portuguese maintained the old distinction, so did
not need to add -o: pris/pres, pus/pos. (The verb querer has i throughout
the perfect in Portuguese as well, perhaps from compound forms; compare
the Romance derivatives of -CLUDERE alongside CLAUDERE.)

Though this type of perfect has practically vanished from the modern
languages, as from Catalan also, there are enough traces of the forms in

54
the old languages to show that in Vulgar Latin times it flourished as much
in Iberia as elsewhere, with the same extension beyond its use in classical
Latin, producing such forms as respuso, priso, tanxo, fuxo, espiso and a
Latin form fraisit in Old Spanish, pris, respos in Old Portuguese, besides
others derived from classical -si perfects such as OSp. cinxo, tinxo, raxo,
coxo, destruxo, escriso, riso, miso, remaso, OPtg. arsi, adussi, ersi, masi,
in addition to the surviving forms, dije, disse etc..

-UI perfect. Spanish uses the same means to differentiate the


persons as in the case of the -SI perfects. Metaphony of ō to u before the
final i took place in conuve from *CO(G)NŌVUI, extended also to pude from
PǑTUI (puse is from *PŌSI not PŎSUI). A pattern similar to that of the -SI
verbs was established in Portuguese for the verbs with root vowel e:
crive/creve, sive/seve (etymologically from *CRĒDUI, *SĒDUI), tive/teve
(analogical from TĔNUI). These verbs appear in Spanish with a change of
vowel: crove/crovo, sove/sovo, tove/tovo, together with other verbs
crove/crovo, atrove/atrovo from *CREVUI (CRESCO), ATTRIBUI. The origin
of this change is not certain, but compare OFr. crui, dui, reçui; in some
cases, this process helped to distinguish the perfect from the present.

In the verbs with root vowel a, this a combined with the u of the
ending, giving OSp. ove, sope, yogue, as it did in the Provençal forms saup,
caup (and cf. receup etc). The same combination took place in French
through loss of the intervening consonant.

55
As with the -SI perfects, though their numbers are few in the modern
languages there is ample evidence that this class flourished in the earliest
IR: after liquids — OPtg. dolverom, valvera; after nasals — Ptg. tive, OSp.
tove; after labials — OSp. sope, cope, ove; after velars — OSp. yogue,
plogue; after dentals — pude, OSp. crove, sove; with further extensions as
in OSp. troxe, visque, nasque, these last two agreeing with the forms found
in Catalan, Provençal and French.

Other strong perfects. Vi, viste etc. with i throughout in Portuguese


and Spanish dialects, coming from the stem vowel in the 1st and 3rd
persons singular and plural, by metaphony in the 2nd person singular and
by analogical extension in the 2nd person plural. In Castilian vi is confused
with the perfects of the -ir verbs.

DEDI has already been discussed. Latin STETI gave estide/estiedo in


earliest Spanish (i in estide by analogy with OSp. fize?), 3rd person estede
in Old Portuguese. The later forms estovo, estudo, estive, esteve etc. are
probably analogical, although a Latin *STETUI is the basis for Catalan,
French, Italian and Rumanian; it is possible that estide etc. also came from
*STETUI, as pude from POTUI. (Provençal has forms modelled on DEDI, but
also estec; Rumanian has made two perfects, one based on the
stem-accented stete, the other on ending-accented stătuşi.)

Latin FECI, VENI gave OSp. fize, vin, Ptg. fiz, vim by metaphony, and
this vowel was extended, except in the Portuguese third person singular,

56
vẽo > veio, with added -o.

In the case of Latin FUI, in common IR, as everywhere else in


Romania including Sardinian, the forms were derived from a contracted
FŪI, FŪSTI (both with metaphony), FŬT, FŬMUS, FŬSTIS, FŬRUNT. Alongside
these, Spanish also had forms with -ue- (-ui-), which became the dominant
ones, and the question is whether these were derived from the classical
Latin forms (with the normal change of accent in a diphthong from FÚERO
to fuéro) or whether they were produced within the history of Spanish
itself. As in the case of -ie-, I incline to the latter view, which does not
necessitate supposing the continued existence of a form lost elsewhere.
The change centres around the same diphthongising territory as in the case
of -ie-, too, which lends support to the idea of a late development. The
starting point for this would be the 1st person fui and the 3rd person foi, to
quote the Portuguese forms; the latter person had an -e added to the end of
the form derived from FŬT in Portuguese as well as Spanish, a thing which
happened frequently to a final accented vowel in verb terminations, cf. vai
from earlier va and old futures in -ae, and also OIt. fue. *Fóe gave fue in
Castilian, setting the pattern for the other persons, and replacing the
unfamiliar endings -úi/-omos with the familiar -í/-emos (later -imos).

From all the above it will be seen that IR started out with the same
multifariousness of perfect forms as the other Romance languages but was
very quick to simplify them and reduce them to standard types, Spanish
going further than Portuguese in the matter of combining the -i- and -ie-

57
types.

h) The Latin future perfect and pluperfect indicative, and pluperfect


subjunctive.

The accented vowels of the endings agree with those of the perfect
third person plural and have been discussed above. These tenses are all
noteworthy for having the accent on the same syllable throughout in IR,
including the 1st and 2nd persons plural, and this phenomenon was
extended to the imperfect indicative (and consequently the conditional) too.
This position was, of course, inherited from Latin in the case of future
perfect forms such as AMARIMUS, AMARITIS, VIDERIMUS, VIDERITIS, and
perhaps spread from here to the other tenses. Italian, Catalan and
Rumanian have the same uniform accent in the tenses derived from the
pluperfect subjunctive; French and Old Provençal preserved the old accent,
but the accent has been moved backwards to the previous syllable in some
of the modern Occitan dialects. With regard to the derivatives of the
pluperfect indicative, preserved in this area, we may note that in Rumanian
and modern Catalan and Occitan the 1st and 2nd persons plural of the
perfect have forms resembling the pluperfect formations (but newly
formed on the analogy of the 3rd person plural), which are also accented
on the same syllable as in Spanish (in the case of Catalan, persons 2, 4, 5
and 6 are identical in the two tenses). In IR the order of change seems to
have been future perfect–pluperfect subjunctive–pluperfect indicative and

58
finally imperfect indicative/conditional. This last change is characteristic
of IR (including Catalan) but a few sporadic examples also occur in the
imperfect and conditional in Old Provençal, e.g. vezíatz, faríatz, and
besides these we have in the modern -ar conjugation cantávam, cantávats,
while forms like -ávamo, -ávate were formerly also usual in Italian, now
being limited to dialectal use. It thus seems that IR radically extended a
tendency which was already nascent in general Romance. It should be
noted here, however, that Galician preserves the Latin accent in the
imperfect and pluperfect indicative, which is another pointer to the
supposition that the change of accent first took place in the pluperfect
subjunctive.

IR is also remarkable in the way it retained the future perfect and


pluperfect indicative, these tenses still flourishing in modern Portuguese,
the former with much the same usage as in Latin, while the latter is limited
to use as a pluperfect in subordinate clauses such as relative clauses or
reported speech, or serves as a past subjunctive or conditional in the
protases or apodoses of conditional clauses, as it does also in Spanish. In
the rest of Romania the pluperfect came to be used as a conditional or a
simple past tense and its use gradually waned; the future perfect is not
even found in the other WR languages, but is still found regularly in late
Latin documents in Italy and France, sometimes formed from the Vulgar
Latin perfects as, for example, the preserit, falierit, inpinxerit found in the
Salic Law in France. It therefore seems to have enjoyed some sort of
existence even outside Iberia till a fairly late date, at least as a literary

59
form; it is quite possible that in Iberia too it entered the spoken language
from the literary language. It is also to be noted that in Navarro-Aragonese
the use of the future perfect (future subjunctive) wanes as one approaches
Catalan territory.

i) Past Participle.
In the modern languages there are virtually only two types of past
participle, those in -ado for the -ar verbs and in -ido for all the rest. But in
the earliest recorded stages of IR there was as much variety as in the
perfect, with all the common Romance types perpetuated. As a rule, the -SI
perfects had past participles in -so or -to (-cho), and the -UI and -EDI
perfects past participles in -udo. Thus in Old Spanish we have priso-preso,
espiso-espeso, miso-miso, raxo-raso, cinxo-cinto, tinxo-tinto, coxo-cocho,
troxo-trecho, aduxo-aducho, destruxo-destruto, respuso-respuesto,
escriso-escrito, quiso-quisto; dijo-dicho, puso-puesto still survive. We also
find isolated strong participles, such as confuso, defeso, enceso, ascuso,
recluso, junto, roto, enjuto. In Old Portuguese we find adusse-aduito,
trouxe-treito, quis-quisto, pris-preso, ersi-ereito, and in modern
Portuguese disse-dito, pus-pôsto, with isolated past participles such as
confuso, defeso, aceso, recluso, junto, rôto, enxuto. Some other strong
perfects also have past participles in -to: hecho/feito, visto (plus an OPtg.,
OGal. veudo, cf. Fr. vu, It. veduto, Rum. văzut); note also nado and
muerto/morto. There is also the same extension of -echo/-eito from the
-ECTU of Latin COLLECTU, which gave cogecho/colheito as OFr. coilleit,

60
OIt. colletto, so OPtg. tolheito as OFr. toleit, OIt. tolletto, and of the -erto
of abierto etc. to Sp. sufierto, substantival oferda, profierta (Gifford &
Hodcroft, Nos. 114, 16, 29), rifierta (Alex.), Ptg. substantival oferta, old
referta, cf. Fr. souffert, offert, Prov., Cat. sofert, ofert, It. sofferto, offerto.
The past participles in -udo are found firstly in the original Common
Romance verbs with perfects in -UI, -EDI12, and then are extended, as in the
other WR languages, to the verbs which changed over to the weak
conjugation with -EDI perfects, before finally giving place to new
formations in -ido. Thus we have in Old Spanish such forms as avudo,
sabudo, tenudo, temudo, moludo, cabudo, recebudo, atrevudo, vençudo,
conosçudo, vendudo, perdudo, rendudo, batudo, creudo, and then
(a)prendudo, metudo, defendudo, esparzudo, ardudo etc., and in Old
Portuguese sabudo, avudo, teudo, temudo, devudo, mudo (moer), cabudo,
recebudo, atrevudo, vençudo, conhoçudo, vendudo, perdudo, rendudo,
batudo, creudo, and then metudo, def(f)endudo, aprendudo, onjudo,
correjudo, corrudo etc., and also veudo (ver) and veudo/vĩudo (vir), the
latter parallelled by OSp. venudo and recalling Cat., Prov. vengut, Fr. venu,
It. venuto. An old -ĬTU participle, BIBITU, survives as an adjective in Sp.
beodo/dial. bébedo, Ptg. bêbedo/bêbado, cf. Fr. en boite.

j) Periphrastic perfect and passive.


These tenses are formed in Romance by use of the auxiliaries HABERE
and ESSE, but Portuguese has replaced HABERE by TENERE in modern times.
The modern Iberian languages differ from the others in always using

61
HABERE/TENERE to form the perfect, with an invariable participle, but in the
old languages ESSE was used with intransitive and reflexive verbs, with the
participle agreeing with the subject, while the participle after HABERE
agreed with the object, as in the other Romance languages. (Also the
distinction between ser and estar was less rigid, it being possible to say, as
in the Alex., estava bien pagado for a past action, not a state.)

k) The -ecer verbs.


One question which has particularly intrigued me is that of the history
of the -ecer verbs in Iberia, and it is one that I find has not received much
attention generally. It seems to have been largely assumed, without further
investigation, that this conjugation descended without change from the
Latin -ESCERE verbs, and without any connection with the -IRE verbs. For
example, Entwistle says: “Ibero-Romance made no use in conjugation of
inceptive forms (-ESCERE), used in Gallo-Romance and Catalan to
regularise the accent in the -ir conjugation.”13 And he is echoed by Elcock
when he speaks of -ESCERE continuing as a suffix in Iberia after it had lost
its inceptive sense, and being used to form new verbs in -ecer.14 Menéndez
Pidal also speaks of the use of the suffix to create new verbs and refers
specifically to its use in creating doublets of -ir verbs, eventually
supplanting the latter, but does not enlarge on this.15 Nunes and Williams
are similarly unenlightening on the position in Portuguese. This stimulated
me to investigate what situation we should have expected to find in IR
from an examination of the other Romance languages, and what the actual

62
position recorded in the earliest documents was, the question in my mind
being, was the assumption that there was never any -ir conjugation with
-esc- infix in IR justified?

When we examine the position in the rest of Romania we find two


main divisions and one sub-division. In the first group is Sardinian, which
separated from the main stream at an early date; here there is no extension
of the -ESCERE type to the -IRE verbs.16 The other group consists of Catalan,
Gallo-Romance, Rhaeto-Romance, Italian and Rumanian; these languages
have a mixed -IRE/-ESCERE conjugation, with the distinction that in most of
Italy and France, and parts of Catalonia, the -esc- infix was changed to
-isc- under the influence of the -IRE forms. In view of what we have
discussed so far, we would expect IR to fit into this class, so let us examine
further.

We need first to study the origins of the infix conjugation. In so doing,


we should beware of using the language of the grammarian as Entwistle
does when he speaks of using the infix “to regularise the accent in the -ir
conjugation”. People do not think of conjugations when they speak; they
are only conscious that words which perform the same functions ought to
behave in the same way. Equally, if it was a question of regularising the
accent, why was the process limited to one conjugation, and to certain
verbs only within that conjugation?

We shall get the right perspective when we take the Latin -ESCERE

63
verbs, which form the nucleus of this group, as our starting point rather
than the -IRE verbs; when we ask ourselves, as it were, not how did the -IRE
verbs come to get an infix, but how did the -ESCERE verbs lose it? In Latin,
it will be remembered, the -ESCERE verbs were first the inceptives17derived
from -ERE verbs, and then came to be derived from adjectives, with the
sense of ‘becoming’: NIGRESCERE, ‘become black’, MOLLESCERE, ‘become
soft’, GRANDESCERE, ‘become large’. At the same time there were -IRE
verbs derived from the same adjectives with the idea of ‘making’:
MOLLIRE, ‘make soft’, GRANDIRE, ‘make large’. There thus arose pairs of
words both indicating a change of state, the one transitive, the other
intransitive. Moreover the -ESC- forms existed only in the imperfective
paradigm, and these verbs had no separate perfect or past participle. So
MOLLITUS EST would come to be associated with MOLLESCERE in the sense
of ‘it has softened’ rather than ‘it has been softened’. (Compare the
transitive and intransitive use of soften, in English.) We can only speculate
about how the two verbs eventually coalesced: perhaps the occurrences of
the intransitive verb in the third persons singular and plural of the present
outweighed those of the transitive verb, whereas the position was reversed
in the first and second persons; and then the first and second persons
singular fell in line with the third persons, as they do in other conjugations,
all being root-accented in contrast to the 1st and 2nd persons plural.
Eventually the distribution of the infix probably became at first as it is in
Italian and Rumanian, that is, in all persons of the singular and the third
person plural of the present indicative and present subjunctive, and the

64
singular of the imperative. In the other languages various extensions of the
infix have taken place: to the plural of the subjunctive in Catalan and old
Provençal; Provençal also extended it to the present participle in ancient
times, and in the modern dialects the infix has the same distribution as in
French, that is, the whole of the imperfective paradigm with the exception
of the infinitive, an interesting point to bear in mind when making
comparisons with the situation in IR. Italian dialects also show extensions
to the first and second persons plural of the imperfective indicative, and to
the infinitive, giving, for example, Calabrian addormiscire/addormiscere,
guarniscire/guarniscere.

Meanwhile, the existence of parallel verbs in -IRE and -ESCERE


indicating a change of state affected other -IRE verbs containing a similar
idea, such as FINIRE (‘bring to an end’), producing a parallel *FINESCERE
(‘come to an end’), and, eventually, by back-formation, *FINUS ‘in a
finished state’, ‘perfect’, from which comes the English fine. Similarly
PERIRE and ADDORMIRE, being intransitive in meaning, developed forms
*PERESCERE, *ADDORMESCERE (always keeping the -IRE forms in the
perfective paradigm). A certain number of other -IRE verbs (partly learned)
also became attracted, irrespective of meaning, such as *PATIRE,
OBOEDIRE, NUTRIRE, IMPEDIRE, but the vast number of the popular -IRE
verbs remained unaffected in the old languages, though they have often
changed over in the modern languages, especially as the -IRE conjugation
with infix became the standard for the learned or semilearned words
reintroduced from Latin, such as French envahir, gémir, convertir,

65
Italian digerire, convertire. Rumanian has also adapted Slavonic
loan-words to this conjugation: primi, citi. It should be noted here that it
was only the -IRE verbs and not those in -ĒRE which were affected in this
way. So when we find verbs like It. fiorire, aborrire, abolire, languire, Fr.
pourrir corresponding to Latin verbs in -ĒRE, we should not derive them
from the latter but rather from the corresponding -ESCERE forms,
FLORESCERE etc., which then generated -IRE forms (FLORIRE, for example,
is amply attested). (In some cases the newly generated -IRE form seems to
have continued an independent existence; thus *LUCIRE, based on
LUCESCERE, has remained as a simple verb, without any infix, in many
areas, as has also *IMPLIRE.) At the same time there seem to have been
-ESCERE forms continuing to exist in parallel with -ĒRE ones; French and
Provençal have paraître, pareisser alongside OFr. paroir, Prov. parer
(similarly with the compounds, whereas Italian has only one infinitive
apparire, but with apparisco and appaio forms, the infinitive apparere
having gone out of use; but in this case cf. OPied. appareiser, OLomb.
pareser, which correspond to the OFr., Prov. forms), and Catalan has
merèixer alongside OCat. merir, which corresponds to OFr., Prov. merir,
raising the problem of which of the new formations came first (was the
form without infix liable to confusion with the merd- verb?). In these cases
Spanish has only parecer, merecer; the simple *parer was driven out by
parir, whereas in France the reverse happened. The existence of this
unmixed -ESCERE conjugation perhaps accounts for the special extension
of -ecer in IR where Italian, for example, has -ere, such as Sp. acaecer,

66
permanecer, pertenecer, Ptg. esquecer compared with It. accadere,
permanere, appartenere, scadere. Carecer also seems to belong here
though it is tempting to see a connection with CARUS (cf. encarecer) in the
sense of ‘dearth’ (from this adjective French and Italian have chérir,
(r)enchérir, rincarire).

One more significant addition to this conjugation in Western


Romance is the Germanic -ian verbs; Fr. guérir, garnir, bâtir, fournir, OFr.
escharnir; Prov. g(u)arir, g(u)arnir, bastir, fornir, escarnir; Cat. guarir,
guarnir, bastir, fornir, escarnir; It. guarire, guarnire, bastire, fornire,
schernire. We will examine later the relation of the Iberian forms.

When we turn to the Iberian languages the first thing that strikes us is
the broad range of correspondence between this infix conjugation, as
exemplified elsewhere, and the -ecer verbs. We find not only the Latin
-ESCERE verbs derived from adjectives, as Sp. endurecer, enmollecer,
(en)verdecer, empobrecer, envejecer, embellecer, ensordecer, amortecer,
establecer and parallel Portuguese formations (cf. Fr. endurcir, amollir,
verdir, empauvrir, vieillir, embellir, assourdir, amortir, établir, with
parallels in the other languages, including Rumanian); we have also
fenecer, padecer, perecer, (Sp., OPtg.) nodrecer, empecer, obedecer and
adormecer, and the Latin -ESCERE verbs derived from -ĒRE verbs —
florecer/florescer, aborrecer, (Ptg.) languescer, podrecer/apodrecer (there
is no abolecer but abolir, which, like (Sp.) arrecirse, empedernir, (Ptg.)

67
languir (beside languescer) and some others, is defective in the modern
language in precisely those parts where we would expect the infix, making
it seem likely that such forms also once existed and also further reinforcing
the argument for a mixed conjugation); finally we have the Germanic
words guarecer, guarnecer, bastecer, fornecer, escarnecer. The notable
difference from the other languages is that the learned borrowings from
Latin, apart from a few like those included in the above list, belong to the
plain -ir conjugation and not the -ecer one.18

Next we need to ask ourselves what we would have expected to


happen to this conjugation in the light of what we know about IR. We have
seen how IR has by and large shared in all the changes undergone by WR,
Italian and Rumanian; with regard to the verbs, in particular, there is the
loss of the Latin passive, deponent and future forms and the creation of a
new analytical perfect and passive and agglutinated future; the change in
conjugation of certain verbs, the participial use of the gerund, the -ea, -ia
imperfect, extension of the -UI and -SI perfects and past participle in -UTU;
in common with all the major languages except Rumanian it has merged
the present tenses of the Latin 2nd and 3rd conjugations, used the infinitive
for the supine and the pluperfect subjunctive for the imperfect subjunctive,
and created a new past tense on the basis of -EDI (we do not know what
happened here in Rumanian as even the descendants of the original -DIDI
verbs, crede, pierde, vinde, have perfects in -ui; there is an ending -éi to be
seen, however, in the first person of the strong perfects, replacing an
earlier unaccented -i, and the z of crezui might indicate an earlier

68
*crezedi from *crediedi). We have also seen how prone IR is to regularise;
how quick it was to substitute weak perfects and past participles for the
Latin -SI, -SU and -SI, -TU forms in all but a few cases, and how soon the -UI
perfects disappeared, with even the new participles in -udo eventually
following suit. How it also merged the infinitives of the 2nd and 3rd
conjugations, and how Castilian produced one single set of endings for the
perfect paradigm of the weak and strong -er verbs and the -ir verbs
(including strong venir). Also how HABERE came to be used to form the
analytical perfect of all verbs, including intransitive and reflexive verbs.

In view of all this, the situation we might expect to find in Spanish,


given the fact that in the modern language the -ir verbs have no infix
while the -ecer verbs have -ec- throughout, is that in the early stages of
the language we would find traces of an earlier -ir conjugation with infix
still surviving alongside forms with the infix throughout. This, I am
convinced, is the position we actually find in Old Spanish. But before
turning to Old Spanish, it is worth remarking that the list given above is
in itself a cogent pointer in the same direction. If there had never been
any connection between -ESCERE and -IRE verbs in Spanish, how could
such forms as fenecer, guarecer have ever arisen? *FINESCERE might
have been created and have coexisted side by side with FINIRE (but there
must have been close connection to have given the first e of fenecer,
which could not have arisen without the intervention of fenir, itself
derived from finir by dissimilation), though it is unlikely that this

69
situation would have remained unaffected by the situation throughout the
rest of Romania, but *PATESCERE would have been a less likely creation,
and it is hardly conceivable that the Germanic words should have been
imported in two forms — in which they do actually appear — if these
forms were not intimately related. Equally, if two separate but related
conjugations existed, how is it that the new Spanish verbs in -ir, such as
reir, vivir, were not affected?

Turning now to Old Spanish texts we find that though the -ecer verbs
have in the main -ec- (or -esc- etc.) in all forms of the verb, -ir forms are
also found in those parts of the verb where we would expect to find them,
for example, in the infinitive (and the future), the first and second persons
plural of the present indicative, the second person plural of the imperative,
the perfect and the past participle (always excepting verbs like parecer,
which never had any -IRE forms). A notable exception is the present
subjunctive, where it is conceivable that Spanish, like Catalan and
Provençal, extended the infix early to all persons. It is equally noteworthy
that the -ir forms do not appear in places where we would expect the infix,
that is, there are no *grado, *grades, *grade, *graden (from gradir =
agradecer), and the Portguese verbs in -ir, where they have survived
alongside -ecer forms, similarly lack such parts of the verb, thus making it
highly unlikely that separate -ir and -ecer verbs formed from the same root
were in existence side by side (allowing always for the fact that new verbs
might be attracted to the -ecer class). There is, it is true, one notable
exception found in the Cid and elsewhere and that is fallen from

70
fallir/fallecer, but as this verb appears without infix in Fr. faillir, Prov.
falhir, Cat. fallir and, formerly, It. fallire, this may be regarded as an
example in Spanish of an -ir verb without infix being attracted to the infix
class; later, the two verbs were differentiated, as also in Portuguese. To
quote other forms without infix from the Cid, we have the infinitives fallir,
guarnir, guarir, future escarniremos, present indicative gradimos,
imperatives guarnid, gradid, perfect paradigm cuntio (= aconteció), gradio,
adurmio, falliessen and past participles fallido, escarnidas, cuntida,
guarnidos, while from an -er verb which was either later attracted to the
-ecer class or else had parallel forms (cf. parecer etc. above) we have
remanidas, remandran, remanga (compare Istro-Rumanian, which also
has rămărescu beside rămăru). In other Castilian documents we find fallir,
guarida; in Asturian-Leonese guarido; and from Navarro-Aragonese
affranquitz, fayllis, fayllissen, goarida, goarnida, goarnir, goarnidos,
establie (perf.), establiren, guarnir, establido, falra, fallen, establimos
(pres.), fornido, guarnido, perir, nudrir and also reman, romanga,
romanientes, with ensordién, ensordidas in the Alex. P; we also find a
Mozarabic baštit, while the consonantism of Sp. marchito betrays a
Mozarabic origin when compared with Ptg. marcido. Above all we have
many past participles in -ido in Berceo, such as enflaquido, empobrido,
enloquido, esclarido, amortido. Some of the Navarro-Aragonese forms
occur in documents written in the Catalan-type Pyrenean dialects where
the infix conjugation is to be expected; the Navarro-Aragonese documents
also contain other forms with infix (succedexca, complezca, destribuescan,

71
possedezcan) where Castilian has none, suggesting that the infix
conjugation lasted long enough there to absorb some of the later learned
borrowings as in French, Italian etc. Besides these forms, Menéndez Pidal
lists seguir and seguecer, aburrir and aborrecer, pudrir and podrecer,
bastir and bastecer, endurir and endurecer, embravir and embravecer,
enflaquir and enflaquecer, padir and padecer,20 without mentioning in
which parts of the verb the -ir forms are found. Similarly Old Portuguese
has fallir, adormir, aborrir, bastir, escarnir, guarir, guarnir, fĩir, florir,
fornir, nodrir, podrir, gouvir, oferir beside -ecer forms, and participial
forms esmarrido (Sp. desmarrido), esmorido. From all this it is seen that
the Spanish -ecer verbs of all types (old -ESCERE verbs, old -IRE verbs and
Germanic verbs) which correspond to -IRE verbs with infix in the other
languages had traces of -ir forms in the old language, whereas verbs such
as merecer, parecer, acaecer did not (though the Fuero Jugo has acayó
from acaer). Similarly, the verbs which vacillated in the other languages or
came to adopt the infix later on, and the later learned borrowings which
joined the -ir, -ire conjugation with infix in French, Italian etc. because
this was the dominant -ir type, did not come to have -ecer forms in
Spanish. In the latter case this is because the infix verbs, having mostly
lost their -ir forms, were no longer felt to represent the -ir conjugation, so
the new verbs were conjugated on the model of partir. We can thus date
the introduction of new words into Spanish, those which were introduced
while the links with France were maintained, such as the Germanic words,
joining the -ecer class, those that came in later joining the simple -ir class.

72
The behaviour of the Germanic words is also evidence for the case that
Spanish did not diverge radically until after the Moorish conquest.

As a final comment let me say also that it is not so remarkable to


suppose that Spanish extended the infix throughout what was once a mixed
conjugation when we consider how far French has gone in the same
direction, extending the infix throughout the imperfective paradigm with
the exception of the infinitive. In fact it might even have gone further had
it not already reached the position where all the endings began with a
uniform i (in point of fact, we do find some rare OFr. future forms in
-istra). This still does not clear up the question as to whether there was an
actual mixed conjugation, although the extension of the -ecer forms at the
expense of the -ir forms suggests this. It may be that in this respect the
northern part of the Iberian peninsula was the meeting place of two
streams of Latinity. On the one hand, an older, more conservative stream,
as found in Mozarabic (which seems to have had no -UTU participles, cf.
miššita as Sp. mecida, mejida), on the other, a later stream sharing features
with Gallo-Romance.

l) Verbs formed with the use of other suffixes.


Three other common Romance suffixes which IR shares with the
other languages also need just to be mentioned; they are the Latin -IARE,
-ICARE and -IZARE (pronounced as if -IDIARE). Examples of -IARE are: Ptg.
alçar, Sp. alzar, Cat. alçar, Prov. alsar, Fr. hausser, It. alzare, Rum. înălŃa,

73
from Latin *ALTIARE. Examples of -ICARE: Ptg. carregar, Sp. cargar, Cat.
carregar, Prov. cargar, Fr. charger, It. caricare, Rum. încărca, from late
Latin CARRICARE. In the case of -IZARE Rumanian differs in using -IZ- as
an infix like -ESC-: Ptg. mane(j)ar, Sp. manear (later replaced by manejar),
Cat., Prov. manejar, Fr. manier, It. maneggiare, from Latin *MAN-IZARE,
cf. Rum. înmânez etc. from înmâna. Other formations with parallels
elsewhere are those based on present and past participles, Sp., Ptg., Cat.
espantar, Prov. espaventar, Fr. épouvanter, It. spaventare; Ptg. ousar, Sp.
osar, Cat. gosar, Prov. gausar, Fr. oser, It. osare.

m) General comment on the verbs.

From all the foregoing it will be seen that the IR verb can be
presumed to have been identical with the verbs in the other Romance
languages until close to the period immediately preceding the first records.
In fact it is remarkable how uniform the verb parts were throughout
Romania, with no major differentiation before the separation of the various
languages. It has been said that Eastern Romance favoured -SI perfects and
Western Romance weak perfects, but the change referred to is a late one
that was not completed until after the appearance of the first documents in
Romance.

74
Conclusion

The picture of IR that emerges is one of a differentiation by stages. In


the first stage, within the period of Romance unity, IR developed a special
regional vocabulary and then became remarkable for the conservatism of
its vocabulary in the face of the changes taking place elsewhere. During
this period it underwent the same phonological and morphological changes
as were taking place throughout Vulgar Latin, and the further phonological
changes only found in WR. Development as a distinct group of languages
had its first beginnings in the 5th century; by the time the first Spanish
documents emerge, the various Iberian dialects are already differentiated,
with some of them, particularly Castilian, showing violent phonetic
departures from the WR original. At the same time, the morphological
changes peculiar to IR are still relatively limited compared with the
position as it was to be several centuries later in the early modern period of
the languages. Notable such changes are the transfer of the -ĔRE verbs to
the -er and -ir conjugations; the introduction of -EDI endings into the
perfective paradigm of the strong verbs, producing an accent on the ending
instead of the stem, and the confusion, in certain areas, of the -EDI and -IVI
perfects; the gradual extension of the infix throughout all the parts of the
infixed -ir conjugation, producing a new class of -ecer verbs; and the loss
of the two-case system in the nouns and adjectives.

Thus we see that there is no reason for supposing that IR dates back

75
to any of the pre-Common Romance features such as the classical perfect
ending in -ĒRUNT; infinitives of the type ARDĒRE, RIDĒRE, RESPONDĒRE,

MORDĒRE; -ESCERE verbs uncrossed with -IRE; or that verbs such as


MORDERE, DEFENDERE, INTENDERE did not go through the stage of having a
perfect in -SI, even though it is not attested. All the forms in present-day
Spanish can be explained as having developed, many of them quite
recently, out of the same type of Vulgar Latin as produced the other
Romance languages, barring those such as Sardinian which were isolated
from an early date. There is thus no conflict between the development of
the phonology, where the evolution in common with the rest of WR is
obvious, and the development of the morphology, which seems at first
sight to have diverged at an early date; and the Latin ancestor of Spanish
and Portuguese is seen to have differed from the Latin of other parts only
to the extent to be expected in the local dialect of one unified language.

BIBLIOGRAPHY
W.D. Elcock, The Romance Languages, London, 1960.
R. Menéndez Pidal, Manual de Gramática Histórica Española, 9th ed.,
Madrid, 1952.
Orígenes del Español, 3rd ed., Madrid, 1950.
W.J. Entwistle, The Spanish Language, London, 1936.
D.J. Gifford & F.W. Hodcroft, Textos Lingüísticos del Medioevo Español,
1st ed., Oxford, 1959.
V. Fernández Llera, Gramática y Vocabulario del Fuero Juzgo,

76
Madrid, 1929.
Dr. J.J. Nunes, Compêndio de Gramática Histórica Portuguesa, 5th ed.,
Lisbon, 1956.
E.B. Williams, From Latin to Portuguese, Philadelphia, 1938.
A. Badía Margarit, Gramática Histórica Catalana, Barcelona, 1951.
J. Anglade, Grammaire de l’Ancien Provençal, Paris, 1921.
E. Bourciez, Précis historique de phonétique française, 8th ed., Paris,
1937.
Éléments de linguistique romane, 5th ed., Paris, 1967.
H.F. Muller & P. Taylor, A Chrestomathy of Vulgar Latin, New York,
1932.
M.L. Wagner, La Lingua Sarda, Berne, 1951.
G. Nandriş, Colloquial Rumanian, London, 1951.
C.H. Grandgent, An Introduction to Vulgar Latin, New York, 1962.
V.Väänänen, Introduction au Latin Vulgaire, Paris, 1967.

FOOTNOTES
1.
Examples taken from the muwaššahas contained in Selection 53 of
Textos Lingüísticos del Medioevo Español, Gifford & Hodcroft (1st
edition).
2.
In spite of their age these texts and the Glosas Silenses sometimes contain
more innovations than are found in later texts from other areas or even in
the modern standard language; compare, for example, naisceset, abierat,
sapieret with nasquieran in the Cid (alongside forms based on nac-),

77
MSp. hubiera, supiere. This development is typical of Aragonese.
3.
See Selections 72, 93, 96 and 106 in Gifford and Hodcroft; 93 is more
latinised but evidently belongs to the same dialect. The editors speak of
“fuertes influencias extranjeras” (p. 128), but it is surely a question not of
outside influences but of an indigenous non-Spanish dialect, as the
language of the texts is remarkably homogeneous and is also distinct from
Catalan or Provençal. This dialect may have had a wider extension than
we would suppose from its limited use in documents; many mediaeval
Navarro-Aragonese texts which are written in a Spanish-type dialect
contain words, particularly names, agreeing in form with this dialect.
4.
Compare the use of manducare in Peregrinatio ad Loca Sancta. Trobar
is also found in one version of the Fuero Juzgo, as well as elsewhere in
poetry, with the meaning of ‘find’.
5.
See “Celtic Lenition and Western Romance Consonants” by André
Martinet, Language, Vol. 28, 1952.
6.
Muller & Taylor, Chrestomathy of Vulgar Latin.
7.
Regarding the origin of this -ANEM accusative type, besides the existence
of parallel forms such as MUCIUS/-UM, MUCIO/-ONEM for the masculines,
may not the monosyllables REM, QUEM have had some effect? To judge by
the French, Spanish and Rumanian forms rien, quien, cine, their direct
ancestor must have been *RENE, *QUENE to have produced the diphthong
in French and the final e in Rumanian, and these might have been felt as
*RENEM, *QUENEM, producing a new declension RES/*RENEM,

78
QUI*/QUENEM. In some cases this -ANEM ending has been modified in the
modern languages under the influence of -ano < -ANUM. So we have Sp.
escribano (earlier escribán), old ermitano (ermitán in Berceo, perhaps by
elision), and It. scrivano, eremitano/romitano from *SCRIBANEM,
*EREMITANEM, formed as new accusatives of SCRIBA, EREMITA; however,
the Portuguese plurals escrivães, ermitães (beside ermitãos/-ões) point to
the original ending reflected in old escripvam, now escrivão, old
hermitan, now ermitão, cf. cam > cão/cães < CANEM/CANES.
8.
Preste is said to be a borrowing from French, in view of the absence of
any diphthong, but I suspect that PRESBYTER must have left some direct
descendant in Iberia, even if that was affected by the French form (or by
church Latin). Also there is the question of the loss of final r. In Italian
and Rumanian it is lost in nouns, but retained with metathesis in adverbs
and prepositions, thus It. frate, moglie, sarto, marmo, pepe, prete, Rum.
frate, împărat, preot, but It. sempre, Rum. între, spre. IR agrees with the
last three (cf. Sp. siempre, entre, sobre) and I suspect that in its treatment
of the nouns it may also have agreed and that preste, fray, sor, old maese
< MAGISTER and Ptg. tredo, pai, mãe, along with Cat. mossén, sor,
represent the normal Iberian development, while sastre and OCat. misser
with their r are loans from GR. (On the other hand, Cat. sastre may be a
native formation, like Pestre Yohan ‘Prester John’ from PRESBYTER

alongside prevere from the oblique cases.) The retention of r in adverbs


and prepositions (and also in the descendants of QUATTUOR) may have
come from the fact that they were closely associated with a following

79
word beginning with a vowel (cf. semper eadem, inter alia). I also
suspect that Sp. mujer/dial. muller, Ptg. mulher and their cognates
elsewhere derive their palatalised l from the nom. MÚLIER > MÚLJER, cf. It.
moglie, as MULÍEREM would have given *muler, cf. pared < PARÍETE; the
existing forms have come from a blend of the two.
9.
But sierpe etc. may come from a different accusative SERPEM, formed on
SERPE(N)S.

10.
Nunes states (Compêndio de Gramática Histórica Portuguesa, pp. 108,
111), on the strength of the OPtg. forms vimen, vermẽ, that all these
nouns must originally have ended in -ẽ, and were therefore derived from
the oblique form. If he is right, then Portuguese agrees with Spanish in
this respect.
11.
Where was the accent in Old Sardinian petterun, vennerun etc.?
12.
Latin -UTU was extended to all -UI perfects on the analogy of words like
TRIBUI-TRIBUTU, ousting the unaccented -ITU. It became attached to the
-EDI perfects, which also originally had -ITU as VENDITU, on the analogy
of BATTUERE, CONSUERE, which lost their U for phonetic reasons and
formed new perfects in -EDI while retaining their participle in -UTU. It is
not necessary to suppose that the verbs which changed to -UTU past
participles went via -ITU, in order to account for such substantival forms
as Ptg. fenda, Fr. fente, Sp. muebda, Fr. meute, though this may have
happened in some cases: it is more natural to suppose that the endings
-da, -te were extended analogically from forms like venda, vente, where

80
they were inherited. The diphthong in meute is difficult to reconcile
historically with the retention of unvoiced t if this came from *MOVITA;
we would expect either *motte or *meude, and may reasonably suppose
the influence of other root-accented forms from the same verb, as in the
case of boite < BIBITA, which follows boire, bois. Deuda, dívida, dette
show the natural development.
13.
W.J. Entwistle, The Spanish Language, p. 59.
14.
Ibid. p. 90. W.D. Elcock, The Romance Languages, 123.
15.
R. Menéndez Pidal, Manual de Gramática Histórica Española, § 125.1.
16.
Elcock includes the south Italian dialects in the same group, but the
Calabrian forms addormiscere, guarniscere, which he quotes, and the
parallel forms addormiscire, guarniscire, all with the infix vowel -i-
coming from the -ir conjugation, seem to throw some doubt on it.
Grandgent says that -ESCERE is not found in Veglia, the Abruzzi,
Sardinia and a part of Lorraine. In the case of Sardinia he is certainly
mistaken. While there does not seem to be a mixed conjugation, there
are parallel forms in -ire and -eskere such as albire/albeskere and
pudire/pudeskere; pudire shows that this area partook in the formation
of new -IRE verbs to match -ESCERE verbs, cf. FLORIRE as FLORESCERE.

Also Maurer quotes a Sardinian (Logudorian) form mereskere, but this


may possibly be the Sp. merecer or Cat. merèixer given a Sardinian
guise.
17.
The term ‘inceptive’ (or ‘inchoative’) is misleading as these verbs
express a change of state as opposed to a completed state; thus

81
FLORESCERE means ‘come out’ and FLORERE ‘be out’. Therefore it is not
true to say that -ESCERE lost its inceptive sense in words like endurecer,
empobrecer.
18.
I was happy to find my conclusions about the origin of the infix
conjugation corroborated by Theodoro Henrique Maurer, Jr. in his paper
“The Romance Conjugation in -ēscō (-īscō), -īre, Its Origin in Vulgar
Latin”, Language, Vol. 27, 1951. He goes into the whole subject in great
detail very convincingly, but without any special emphasis on the state
of affairs in IR. Also I feel his conception of the extent of the diffusion
of the infix conjugation in IR is too limited, as he excludes in particular
the original -ESCERE verbs derived from adjectives, which were the
origin of the whole conjugation, in spite of the fact that these verbs have
-ir forms in IR (e.g. endurir, enverdir, establir).
19.
All examples taken from Gifford & Hodcroft, op. cit.
20.
R. Menéndez Pidal, op. cit., § 125.1. Padir may be another example of a
verb that was attracted to the -ecer class later, as the old forms pades,
padan are found.

*****

This is an edited version of a paper originally published in Ronshu, Vol.8


(Tokyo, Aoyama Gakuin University, 1967).

82

You might also like