|1. Prounciation of Latin vowels
|1.1 Classical Latin|
|1.2 Vulgar Latin|
|2. Diphthongization of /ε/ and /ɔ/|
|3. Vowel raising|
|4. Syncope of intertonic vowels|
|5. Apocope of final /e/|
1.1 Classical Latin
The Classical Latin orthography had five vowel letters, which can be transliterated as a, e, i, o and u (although there is only a partial resemblance between the latter forms and the equivalent letters in Latin handwriting or Roman cursive). Originally, each of these five vowels had both a long and a short pronunciation. Vowel length can be indicated by placing either a macron or a breve above the letter in question, as shown below, but it should be remembered that this is a practice that was developed by scholars writing about Latin rather than something that the ancient Romans did themselves.
ā, ē, ī, ō, ū (macron, long pronunciation)
ă, ĕ, ĭ, ŏ, ŭ (breve, short pronunciation)
The five Latin vowel letters each denoted a pure vowel when used individually. There were, however, three diphthongs, designated by the digraphs ae, oe and au, as in Aegyptus ‘Egypt’, foetidus ‘stinking’ and taurus ‘bull’. Originally, these diagraphs probably represented the diphthongs [aj], [oj] and [aw] respectively.
1.2 Vulgar Latin
The vowel system of (late) Vulgar Latin, the source of the modern Romance languages, differed in certain key respects from the Classical system just described. Most importantly, differences in vowel length were either lost or replaced by differences in aperture, i.e. tongue height. In the case of the two most open vowels, ă and ā, the length contrast was simply lost and the two vowels merged into /a/. Among the two high vowels, the long variant in each case retained its original degree of aperture, ī continuing as /i/ and ū as /u/, but the short variant merged with the long variant of the corresponding mid-high vowel. Thus ĭ and ē merged into /e/ while ŭ and ō merged into /o/. As regards the short variants of the mid-high vowels, viz. ĕ and ŏ, these remained distinct from their long counterparts, but only in stressed syllables. In unstressed ones, they joined the above-mentioned mergers, ĕ being absorbed into /e/ and ŏ into /o/.
These changes imply that the ancestral vowel system of Spanish can be reconstructed as in Table 1 below.
Stressed syllables only
|Vulgar Latin||Classical Latin source||Vulgar Latin||Classical Latin source|
|/e/||ē, ĕ, ĭ||/ɛ/||ĕ|
|/o/||ō, ŏ, ŭ||/ɔ/||ŏ|
Three additional points should also be noted.
2. Diphthongization of /ε/ and /ɔ/
Two vowels that appear to have enjoyed phonemic status in Vulgar Latin do not survive in the modern Spanish phonemic inventory. These are /ε/ and /ɔ/, corresponding respectively to ĕ (as well as ae) in stressed syllables and to ŏ in the same context. These sounds underwent vowel breaking, the sound change whereby a pure vowel evolves into a diphthong (or a triphthong). In the case of /ε/ and /ɔ/, each of these vowels came to be preceded by a homorganic semivowel, viz. [j] and [w] respectively, resulting in the diphthongs [je] and [wo]. In the vast majority of cases, the former survives unamended into modern Spanish:
mĕtum [ˈmεto] > [ˈmjeðo] miedo ‘fear’
nĕb(ŭ)lam [ˈnεbola] > [ˈnjeβla] ‘fog’
In contrast, while the [wo] diphthong survived into Italian, as is indicated by words like fuoco ‘fire’, ruota ‘wheel’ etc., it evolved to [we] in most Iberian varieties of Romance:
fŏcum [ˈɸɔko] > [ˈɸwoko] > [ˈfweɣo] fuego ‘fire’
prŏbam [ˈpɾɔba] > [ˈpɾwoβa] > [ˈpɾweβa] prueba ‘proof/evidence’
Posner (1996:158) alludes to the misspelling NIEPOS for NEPOS ‘nephew’ in an inscription dating from 120 AD, which suggests that diphthongization of ĕ may have been underway by the second century. According to the same author, diphthongization of ŏ is conventionally assumed to have begun somewhat later.
A following syllable-final nasal had the effect of slightly raising the tongue height of /ɔ/ and so a number of words containing ŏ + nasal + consonant escaped diphthongization. For example, the modern reflex of mŏntem is monte ‘mountain’, rather than *muente. Similarly, hŏm(i)nem ‘man/human’ delivers modern hombre (although huembre is sporadically attested in older forms of Spanish).
It should also be noted that the diphthong [je] (< ĕ) was sometimes reduced in Old Spanish to [i], particularly before /ʎ/ and, to a lesser extent, before syllable-final /s/. Similarly, the diphthong [we] (< ŏ) underwent occasional reduction to [e], the favoured context being after /l/ or /ɾ/:
castĕllum > Old Sp. [kasˈtjeʎo] castiello > Mod. Sp. [kasˈtiʎo] castillo ‘castle’
flŏccum > Old Sp. [ˈflweko] flueco > [ˈfleko] fleco ‘fringe’
3. Vowel Raising
In the history of Spanish, the palatal semivowel [j] (often called yod) has been involved in a very large number of sound changes. In the vowel system, its principal effect was the raising of the stressed vowel’s tongue height in many words, an effect brought about by the very high tongue position of [j] itself. This raising effect can be regarded as an instance of metaphony, the name given to any process that leads to the approximation of one vowel’s quality to that of another. In the Spanish case, the metaphony was regressive, in the sense that the influencing sound, [j], always came after the vowel that underwent raising (though not necessarily immediately after).
The vowels that were capable of being affected were /a, ε, ɔ, e, o/. When affected, these raised to /e, e, o, i, u/ respectively, implying that they each raised by one degree of aperture.
The [j] which triggered the metaphony had a variety of sources, including (unstressed) front vowels that semivocalized after the loss of the hiatus and velar or lateral consonants that semivocalized in syllable-final position. In most cases the [j] did not itself survive into Old Spanish, being either elided altogether or, more commonly, absorbed into an adjacent consonant. As a general rule, the earlier the [j] disappeared the weaker its effect was in terms of causing vowels to raise. In the extreme case of the [j] that was involved in the palatalization of Latin /t/ and Latin /k/, which resulted ultimately in the creation of modern /θ/, disappearance occurred at a very early date, leaving no time for vowel raising to take place.
Metaphony did not apply uniformly across the vowel system. The vowels that were most consistently affected were /ε/ and /ɔ/, corresponding to the short vowels ĕ and ŏ in stressed syllables. The vowels that were least affected were /a/ and /e/. In the case of /a/, metaphony only seems to have occurred if the [j] was – or came to be – in the same syllable as the /a/.
Examples of vowel raising are given in Tables 2a and 2b below.
/a/ > /e/
/ε/ > /e/
/e/ > /i/
|lacte [lajte] > leche ‘milk’||
praemium [ˈpɾεmjo] > premio ‘prize’
|lĭmpi(d)um [ˈlempjo] > limpio ‘clean’|
|bāsium [ˈbajso]* > beso ‘kiss’||
lĕctum [ˈlεjto] > lecho ‘bed’
|mētiō [ˈmetjo] > mido ‘I measure’|
|vĕnio [ˈwεnjo] > vengo ‘(I) come’|
|rĕg(u)lam [ˈrεjla] > reja ‘ploughshare’|
|* Assuming metathesis of [s] and [j].|
/ɔ/ > /o/
/o/ > /u/
|ŏctō [ˈɔjto] > ocho ‘eight’||plŭviam [ˈploβja] > lluvia ‘rain’|
|pŏdium [ˈpɔdjo] > poyo ‘stone bench’||mŭltum [ˈmojto] > mucho ‘much’|
|fŏlia [ˈhɔlja] > hoja ‘leaf’||pŭgnum [ˈpojno] > puño ‘fist’|
|ŏstream [ˈɔstɾja] > ostra ‘oyster’|
The /ε/ > /e/ and /ɔ/ > /o/ changes, which are not immediately obvious from a comparison of Spanish with Latin, can be inferred from the failure to diphthongize. For example, the [ɔ] corresponding to the ŏ in the word ŏctō must have been raised to [o] because, otherwise, the expected modern form would be *[ˈwetʃo] and not [ˈotʃo] ocho.
In addition to the foregoing changes, which affected stressed vowels, a parallel phenomenon affected /e/ and /o/ in unstressed initial syllables, with raising to /i/ and /u/ respectively:
caemĕntum [keˈmjento] > cimiento (whence cimientos ‘foundations’)
cognātum [kojˈnato] > cuñado ‘brother-in-law’
cochleāre [koˈkljaɾe] > cuchara ‘spoon’
Note also that stressed vowels were occasionally raised under the influence of final /i/. For example, modern Spanish vine ‘I came’ is descended from Latin vēnī, a derivation which implies first of all metaphony in relation to the root vowel /e/ and, subsequently, replacement of final /-i/ by /-e/. Analogously, the imperative form vĕnī, where the root vowel is /ε/, delivers ven in modern Spanish rather than *vien.
4. Syncope of intertonic vowels
Intertonic or unstressed internal vowels occupied a position of relative weakness in words and so were prime candidates for syncope (i.e. loss). In the examples below, the syncopated vowel is underlined in the Latin etyma (as can be seen from the hostal example, loss of the intertonic vowel often triggered some readjustment of the consonant cluster produced by the syncope):
ciudad ‘city’ < cīvĭtātem
pueblo ‘people’ < pŏpŭlum
ojo ‘eye’ < ŏcŭlum
hostal ‘hotel’ < hŏspĭtāle
The only Latin intertonic vowel to survive into modern Spanish is /a/, the vowel with the highest degree of sonority (i.e. acoustic intensity). With the exception of /a/ all intertonic vowels had been lost in Spanish by the end of the Early Middle Ages, although this result was achieved in two distinct phases.
The first phase, which appears to have taken place before the end of the Latin period, was phonologically conditioned, in the sense that syncope occured only when the relevant vowel was adjacent to /ɾ/ or /l/ or, occasionally, /s/. Thus already in the Appendix Probi we find recommendations such as the following:
speculum non speclum (> espejo ‘mirror’)
viridis non virdis (> verde ‘green’)
Subsequently, in pre-literary Spanish, there was a generalized loss of the intertonic vowel (although /a/, as noted above, was unaffected). This process played an important role in determing the phonetic shape of a large number of words. Some examples are given below:
vĭndĭcāre > vengar ‘to avenge’
līmĭtāre > lindar ‘to border’
delĭcātum > delgado ‘thin’
domĭnĭcum > domingo ‘Sunday’
5. Apocope of final /e/
Final /e/ was unstable in Old Spanish and was subject to two waves of apocope. The more significant of these, in terms of the formation of modern Spanish, is a generalized sound change that occurred towards the end of the first millennium (it post-dates both lention and the loss of the intertonic vowel). This change resulted in the permanent loss of final /e/ after intervocalic dental and alveolar consonants, except in enclitics such as the pronoun le. A good number of familiar modern words display the effect of this change, including the following:
mar ‘sea’ (< mare [ˈmaɾe])
pan ‘bread’ (< pānem [ˈpane])
sed ‘thirst’ (< sĭtĭm [ˈsete], later [ˈseðe])
hostal ‘hotel’ (< hŏspĭtāle [ospeˈtale], later [osˈtale])
mes ‘month’ (< mēnsem [ˈmense], later [ˈmese])
The other wave of apocope, associated primarily with the twelfth and thirteenth centuries, did not (in the general case) lead to any permanent change and usually involved variable or stylistic elision. This variable elision was not phonologically conditioned, as was the permanent sound change just mentioned, but occurred after almost any consonant. The example below illustrates apocope after the palato-alveolar affricate /tʃ/, the word leche being shortened to lech:
La lech e la manteca e el queso deuen uenir al uso de las monias.
‘Milk and butter and cheese should be for the use of the nuns.’
(Documentos castellanos de Alfonso X - León)
Variable elision of final /e/ in enclitic pronouns is also well attested in the written texts of the period. The example below has diol for diole:
& diol un colpe tan grand; que luego a pocos de dias fue muerto.
‘And he gave him such a blow that within a few days he died.’
(Estoria de España I)
The phonologically unconditioned elision of /-e/ died out in the Late Middle Ages. However, in a small small number of words, final /e/ was permanently lost after consonants other than the ungrouped dentals and alveolars which defined the context for the generalized loss of this element. The words mil ‘thousand’, él ‘he’ and piel ‘skin’ owe their phonetic shape to this circumstance, each item having originally been a disyllable with an internal geminate:
mīlle [ˈmille] > [ˈmiʎe] > [ˈmil] mil
ĭlle [ˈelle] > [ˈeʎe] > [ˈel] él
pĕllem [ˈpεlle] > [ˈpjeʎe] > [ˈpjel] piel
The geminate lateral in these words lenited to /ʎ/, which depalatalized as a by-product of the apocope, Spanish phonotatics disallowing /ʎ/ in word-final position.