Professional Documents
Culture Documents
Hugh E. Wilkinson
and so the influence of suffixes like -ado, -ido and the rest has
worked to change -aço, -iço, -eça and the like to -azo, -izo, -eza
and so on, in words where these segments are clearly
recognizable as suffixes.
assume, with Ford and Mussafia, that this is the natural outcome
of -TJ- in those areas where single voiceless consonants were
regularly voiced between vowels. But there are also many
instances in all the languages of -TJ- giving a voiceless consonant
in popular words — not only when preceded by another
consonant — and here I think we may most conveniently suppose
that the -T- was doubled in Vulgar Latin; there are also other
cases where a voiceless consonant is the rule, and that is in
(semi-)learned borrowings. One of the points in which Spanish
and Portuguese differ is in the fact that Portuguese has more
examples of ç, not only in words where Spanish has innovating
-azo, -izo from -ACEU, -ICIU, but in words where Latin has -TJ-.
It is here that I am not happy with Malkiel’s assumption of ç as
the normal reflex; if we take the other position and assume that z
was the natural reflex of -TJ-, and that inherited ç must go back to
-TTJ-, we do not need to go to special lengths to explain the
presence of z in Sp. (agu)-zar, -ez(a), -azón, -izón, thus confining
Malkiel’s anomalous “derivational patterns” to -azo and -izo.
(Malkiel does not make it clear whether he thinks his ç comes
from -TJ- or -TTJ-; if it is the latter, then I must rephrase my
objection and say that I believe Spanish inherited forms with z <
-TJ- as well as forms with ç < -TTJ-, as did the other Romance
languages.) However, at the same time, we shall have to account
for the ç in forms like Ptg. -ação, -ição and others, where these
correspond to Spanish forms with z. In this case, one obvious
explanation is the influence of (semi-)learned borrowings, and in
5
both cases, while the latter has -θ-, -tt-, -tts- for both, according
to the dialect (kapiθa, puttu etc.). It is true that there are some
exceptions in Rumanian, but these must be due to special causes;
for example, from ERICIU we have ariciu for expected *ariŃ, for
which we may suppose either derivation from *ERICE or
back-formation from *ariciune < *ERICIONE (cf. Fr. hérisson).
In southern Italian, too, the outcome of both is zz. Standard
Italian (Tuscan), on the other hand, is instructive in that it does,
in general, distinguish -KJ- from -TJ-, with cci for the former and
zz for the latter, but it differs in two respects from Western
Romance. Firstly, it does not voice intervocalic voiceless
consonants, and, secondly, it is prone to geminate, so that we
7
maça, Prov. massa, Fr. masse (mache), massue, NIt. maça, Ven.
mazuco, Engad. mazzüc, It. mazza, mazzocca, Rum. măciucă.
Then you have *PLATTEA giving Sp. plaça, Ptg. praça (both
showing less than popular development), Cat. plaça, Prov. plassa,
Fr. place (plache), NIt. plaça/platsa, Rh. plazza, It. piazza, and
*PETTIA giving Sp. pieça, Ptg., Cat. peça, Prov. pessa, Fr. pièce
(pieche), NIt. peza/peça, Rh. pezza/piaza, Friul. piezze, It.
pezza/pezzo, Sard. petha/peza. We also find evidence for
*ARBUTT-EU/-EA in Arag. alborço, Cat. arboç, arboça, Prov.
arboz, arbossa, NIt. (Lig.) arbossa; Fr. arbouse, arbouce seem to
reflect OFr. arboisier (from -TEU) influenced by Prov. arbossa.
Limited to Hispano-Romance we find LAUTIA giving Sp. loça,
Ptg. louça, with the preceding AU, which fell in with AL,
preserving the voiceless consonant (there is evidently no
connection with Cat. loça, Prov. lossa and Fr. louche (a northern
form for earlier louce), all meaning ‘ladle’).
The interesting thing here, from our point of view, is the fact that
neither Spanish nor Portuguese has forms with the popular
ending, as this can be compared in some respects with the
development of PIGRITIA. Here Portuguese has only preguiça
(old priguiça, pegriça), while Spanish has both pigricia, with
OLeon. (Alex. O) pegricia, pegriçosos, and pereza; Catalan and
French have popular forms, pere(s)a and paresse, while
Provençal has pereza, pigreza and pigricia, and Italian has
pigrizia (‘laziness’) and pigrezza (‘sluggishness’). Apart from
OIt. (Veronese) pegro, the adjective PIGRU has not survived in a
popular form, and in many areas it must have been driven out by
PIRU, PIRA (compare It. pero, pera with peritare < PIGRITARE).
But why PIGRITIA did not also survive in a popular form in
Portuguese is a mystery. That the -iça suffix is not popular seems
to me clear; for one thing there is no alternative *pigricia (as no
*cupidicia), and the OPtg. priguiça seems to reflect *pigric(i)a
with -cia > -ça as in graça < gracia, while the pegr-, preg-
forms (and OLeon. pegr-) might reflect a popular root (cf. the
combination of popular cobi- with early borrowed suffix -iça).
Again, though Malkiel speaks of preguiça as the popular
development, the correspondence of maiça to Sp. malicia
(alongside popular maeza corresponding to Sp. maleza) and of
cobiiça to Sp. cobdicia tells me that we should equate
preguiça/priguiça with pigricia/pegricia. -Iça in Portuguese is
quite a rare ending, as we shall see later, and its use here may
have some special significance; was there, for instance, any
21
from Latin -CTIONE etc., forms which were not always learned;
thus we have lição, feição, afeição, eleição, correição, destruição,
conceição, canção, torção, punção, whereas Spanish has the
corresponding -çón in only a few words like cançón (Alex. O),
torçón, punçón. Add to this the Portuguese preference for
semi-learned forms (e.g. multidão where Spanish has
muchedumbre), and the homophony of inherited and early
borrowed -ença (a point not spotted by Malkiel), which accounts
for the frequency of -ença in Portuguese where Spanish only has
-encia except in simienza, fimienza, with some other words found
in Alex. O, such as creença, fallença, repentença. (Portuguese
words with -ência are later borrowings.). In the case of -ANTIA
the two languages do not differ so much in the ratio of
-anza/-ança to -ancia/-ância forms. Add too the possible effect
of carnaça on carnação, and the influence of other words ending
in -ção, such as caparação, mação, carapução, cabeção, pellição,
and it is clear that for this particular suffix the odds were
weighted very heavily against the survival of *-zão. Another
point to consider is the fact that the reflexes of -ITIONE containing
a voiced consonant, such as Sp. -izón, Fr. -ison, must have been
remodelled on those of -ATIONE, as intertonic -I- would normally
be lost; cf. OFr. mensonge, parçonier, vençon, besson/boisson
(and norreçon with a vowel of support), Prov. menzonja,
parsonier, Cat. monçonja, parçoner, Sp. parzonero, Ptg. parceiro,
Sard. parthone, It. menzogna, Rum. minciună (these last perhaps
showing loss of -TI- by haplology rather than loss of intertonic -I-,
27
different ACIA), and Asturian tšaz, Gal. lazo (if this reflects
*GLACE) to Cat. glas/glaça, Prov. glatz/glassa, Fr.
(ver-)glas/glace, Rh. glatsch/-a, Friul. glace, It. ghiacci-o/-a,
Rum. ghiaŃă, which are all derived from *GLACCIE/*GLACCIA.
Considering that both groups of words show an alternation
between -IE and -IA, we may well say that the development of
-ACIE affected that of -ITIE. And we may even go so far as to
suppose that -ITIE, -ITIA tended to become *-IC(C)IE, *-IC(C)IA,
judging from the confusion (admittedly, inconclusive) we find in
mediaeval Latin, which writes -ICIA for -ITIA (even earlier than
this we find IUSTICIA recorded in an edict of Diocletian at the
end of the 3rd c.), and also -ARITIA for -ARICEA. This latter
confusion is also sometimes reflected in a variation in the
Romance forms; thus, corresponding to Fr. forteresse, or
fortelesce with dissimilation, Provençal has both fortaressa,
fortalessa and fortareza, fortaleza, the last of which was adopted
by Catalan, Spanish and Portuguese. (In contrast, Ptg. calvareça
seems to agree with fortaressa.) In view of this, it is not
inconceivable that this -ICEA affected -ITIA, and in turn -ITIE,
producing the voiceless consonants of Fr. -esse, Ptg. -ece etc. We
also see in Hispano-Romance that -ARĬC-EU/-EA were affected
by -ARĪC-EU/-EA, producing forms like Sp. caballer-izo/-iza,
vaqueriza, porqueriza, hortaliza, Ptg. cavalar-iço/-iça, porcariço,
hortaliça, Cat. hortalissa, in contrast to OFr. chevalerez,
vacherece, porcherece, Prov. vacaressa, ortalessa, It.
vaccareccia, porcareccia, Rum. călăr-eŃ/-eaŃă, lungăreŃ, so it
31
would also be possible to see a popular origin for Ptg. -ice, -iça
based on a substitution of *-ĪCIE, *-ĪCIA (based on -(AR)ĪCEA)
for -ĬTIA, though the correspondence of -iça to Sp. -icia
(semi-learned) seems to point against this. Incidentally,
substitution of -ĪC-EU/-EA for -ĬC-EU/-EA will also explain Ptg.
cortiça for expected *corteça < CORTICEA (It. corteccia), and
carriço < *CARĬCEU (cf. It. dial. carezza).
10. Popular words derived partly from -TJ- and partly from
-TTJ- forms. In the case of -ITIA we have seen that some of the
Romance languages have forms that require us to posit -TTJ-
instead of -TJ-, while others, including Portuguese, have both;
and in the case of -ATIONE, -ITIONE the forms with ç, which are
peculiar to Portuguese, can be explained by learned influence.
But we now come to a group of popular words in which
Portuguese has only ç representing -TJ-, where Spanish has z; and
once again there is support for the Portuguese forms from other
languages, not on such a large scale as in previous instances, but
still sufficient to make us suspect, as a possible solution, that we
have to do once more with alternative -TTJ- forms in Common
Romance, rather than concluding that ç is the natural Portuguese
outcome of -TJ- with a single -T-. To take first a group of verbs,
we have Ptg. aguçar, esmiuçar, adelgaçar in contrast to Sp.
aguzar, (d)esmenuzar, adelgazar, Cat. aguar, Prov. aguzar,
amenuzar, Fr. aiguiser, (a)menuiser, It. a(g)uzzare, (s)minuzzare
(and the noun minugia, perhaps from French), and also Ptg.
32
meio, raio (with both raia and raja), ensaio, so that it is equally
possible to believe in a double outcome of -TJ-, especially given
the evidence of the other languages. In fact, though there is a
body of words (*CAPITTIU etc.) where doubling is found
consistently, there are virtually no words where a voiced
consonant is found in every language, even though the majority
have one; we need not be unduly worried, therefore, if
Portuguese shows sometimes z and sometimes ç in a different
distribution from Spanish, especially when we consider that the
hesitation goes right back to Latin times, as seen in the use of -TI-
for -CI- and vice versa (SOLATIUM for SOLACIUM, cf. Prov.
solatz, solassar, OFr. solas, solacier (solachier) reflecting the
latter, and PECIOLUS for PETIOLUS below). In view of all this,
we also need to question whether the fluctuation between ç and z
in Spanish simply represents a confusion of orthography, or
whether there are also underlying variations in the pronunciation,
with, say, the use of the pronunciation -azo for -aço gradually
spreading from word to word. At the same time we need to be
clear that there is no need to suppose that z has replaced earlier ç
in Spanish words derived from Latin words with -TJ-, as Malkiel
would have it when he posits *avareça, *adelgaçar etc., though
there may sometimes have been doublets like poço/pozo, as
elsewhere.
11. Spanish -azo, -izo. We now come to the cases where Spanish,
and occasionally Portuguese, has z for expected ç from -KJ-.
37
Malkiel neatly accounts for some of them; for example, the rise
of -rz-, -nz- from -RGe,i-, -NGe,i- and the suffix -zillo from
-ICELLU changed arçón, cenceno to arzón, cenzeno, while the
latter spread its influence to OGal. anzella, and to anzuelo, OPtg.
anzol from *HAMICEOLU, which contrast with Gal. amocelo.
Azero for *acero (but note Alex. P açero), and also OPtg. azeiro
beside aceiro, have been affected by az, whose development we
looked at just now. But these are just isolated instances, and the
bulk of the cases we need to look at concern words with -azo,
-izo in Spanish. (Portuguese has -zo, -za rather than expected -ço,
-ça in borrowings like granizo, gozo, granza (beside garança),
though I also note an alternative reading lazes for laços in Nunes,
Crestomatia Arcaica, p. 48.) These words are at the core of
Malkiel’s theory, and while I feel that attraction to other suffixes
containing a voiced consonant, as he makes out, may have played
a part, I would prefer to speak of an extension of previously
existing -azo, -izo suffixes, rather than the creation of new ones,
to replace *-aço, *-iço in the cases where “derivational
transparency” is in operation. (But actually in his thesis there are
also cases where primitives are involved, and some of those
words he lists as containing transparent suffixes, such as
amenazar, can hardly be considered so in the context of Spanish
alone.) The -azo, -izo suffixes I refer to come from -ACE, -ICE >
-az, -iz, with -o (or -a) added to characterize the gender, a point
which Malkiel mentions, but does not emphasize. But in his
previous article, to which I referred in § 3, he points out the
38
and Portuguese words like goço (gozo), baço, raça. Here goço
is evidently the original form and gozo may be remodelled on
pozo; Sp. goço < GAUDIU (attested in Alex. O) matches Ptg. ouço
< AUDIO (cf. the other 1st pers. forms above) and Leon. oça
beside ozo, oza, and resembles loça, Ptg. louça < LAUTIA, as
well as loçano, Ptg. loução, which probably go back to a root
*FLAUTJ-. In other cases we may see a substitution of familiar
-TIA for unfamiliar -DIA, as in Sp. grança, vergüença, verça
(now berza) from GRANDIA, VERECUNDIA, VIRIDIA (here we
find Ptg. grança, vergonça, verça; also fronça/frança <
*FRONDIA), or -TEOLU (-CEOLU) for -DEOLU in Sp. orçuelo <
HORDEOLU (Ptg. torçol/terçol/treçol and torção; crossed with
*TRITICEOLU, TORTIONE?). In this case there may have been
confusion with URCEOLU, as Rumanian has ulcior ‘stye’ (cf. It.
orzaiuolo) beside urcior ‘jug’, as It. orciuolo. The simple
HORDEU is also represented in Spanish and Portuguese, with a
voiceless consonant appearing in some forms: Moz. uarso,
uaryo,(17) (cf. horchata, also from Mozarabic), OGal. orjo.
Elsewhere we find Cat. ordi, Prov. ordi, orge, Fr. orge, It. orzo,
Rum. orz. This substitution would presumably have taken place
at the Latin stage; we have already seen how, at a later stage, -nz-,
-rz- were substituted for -nç-, -rç-, a change matched in Italian by
that in garzone (= Fr. garçon) and Tuscan scorza < SCORTEA
with voiced z, though Sp. grança, vergüença etc. are paralleled
by It. pranzo < PRANDIU, with voiceless z in some areas.
(Incidentally, Sp., Ptg. grança ‘siftings’ must not be confused
42
References
***********