Latin to Spanish: Phonology

10 (/i/ [ɪ], /i:/, /e/ [ɛ], /e:/, /u/ [ʊ], /u:/, /o/ [ɔ], /o:/, /a/ and /a:/)

Classical Latin vocalic system: number of vowels?

/i/ ([ɪ]), /i:/, /u/ [ʊ], /u:/

Classical Latin vocalic system: degree of aperture, high vowels?

/e/ [ɛ], /e:/, /o/ [ɔ], /o:/

Classical Latin vocalic system: degree of aperture, mid vowels?

/a/, /a:/

Classical Latin vocalic system: degree of aperture, low vowels?

/i/ [ɪ], /i:/, /e/ [ɛ], /e:/

Classical Latin vocalic system: front vowels (note that /a/ and /a:/ are neither front nor back)

/u/ [ʊ], /u:/, /o/ [ɔ], /o:/

Classical Latin vocalic system: back vowels (note that /a/ and /a:/ are neither front nor back)

AURUM

oro (Classical Latin vocalic system: AU [au̯] is one of three Latin diphthongs, the others being AE [ai̯] and OE [oi̯])

HĪC/HIC

aquí/este (Classical Latin vocalic system: minimal pairs based on contrast of vowel length. However, this contrast was lost no later than the first century AD, and the functional load carried by this feature was transferred to that of aperture, where the formerly long vowels were slightly higher than the short vowels. Meanwhile, /a/ and /a:/ simply merged, meaning that the spoken Latin vocalic system consisted of nine phonemes: /i/, /ɪ/, /e/, /ɛ/, /u/, /ʊ/, /o/, /ɔ/, /o/ and /a/). However, this system, which depended on the fragile distinction of five degrees of aperture, was soon modified. In tonic position, no later than the first century AD, there were two mergers: one between /ʊ/ and /o/ in favour of /o/, and one between /ɪ/ and /e/ in favour of /e/. This left seven vowels in tonic position in spoken Latin: /i/, /e/, /ɛ/, /a/, /ɔ/, /o/, /u/)

VĒNIT/VENIT

vino/viene (Classical Latin vocalic system: minimal pairs based on contrast of vowel length. However, this contrast was lost no later than the first century AD, and the functional load carried by this feature was transferred to that of aperture, where the formerly long vowels were slightly higher than the short vowels. Meanwhile, /a/ and /a:/ simply merged, meaning that the spoken Latin vocalic system consisted of nine phonemes: /i/, /ɪ/, /e/, /ɛ/, /u/, /ʊ/, /o/, /ɔ/, /o/ and /a/. However, this system, which depended on the fragile distinction of five degrees of aperture, was soon modified. In tonic position, no later than the first century AD, there were two mergers: one between /ʊ/ and /o/ in favour of /o/, and one between /ɪ/ and /e/ in favour of /e/. This left seven vowels in tonic position in spoken Latin: /i/, /e/, /ɛ/, /a/, /ɔ/, /o/, /u/)

ŌS/OS

boca/hueso (Classical Latin vocalic system: minimal pairs based on contrast of vowel length. However, this contrast was lost no later than the first century AD, and the functional load carried by this feature was transferred to that of aperture, where the formerly long vowels were slightly higher than the short vowels. Meanwhile, /a/ and /a:/ simply merged, meaning that the spoken Latin vocalic system consisted of nine phonemes: /i/, /ɪ/, /e/, /ɛ/, /u/, /ʊ/, /o/, /ɔ/, /o/ and /a/. However, this system, which depended on the fragile distinction of five degrees of aperture, was soon modified. In tonic position, no later than the first century AD, there were two mergers: one between /ʊ/ and /o/ in favour of /o/, and one between /ɪ/ and /e/ in favour of /e/. This left seven vowels in tonic position in spoken Latin: /i/, /e/, /ɛ/, /a/, /ɔ/, /o/, /u/)

MĀLUM/MALUM

manzana/mal (Classical Latin vocalic system: minimal pairs based on contrast of vowel length. However, this contrast was lost no later than the first century AD, and the functional load carried by this feature was transferred to that of aperture, where the formerly long vowels were slightly higher than the short vowels. Meanwhile, /a/ and /a:/ simply merged, meaning that the spoken Latin vocalic system consisted of nine phonemes: /i/, /ɪ/, /e/, /ɛ/, /u/, /ʊ/, /o/, /ɔ/, /o/ and /a/. However, this system, which depended on the fragile distinction of five degrees of aperture, was soon modified. In tonic position, no later than the first century AD, there were two mergers: one between /ʊ/ and /o/ in favour of /o/, and one between /ɪ/ and /e/ in favour of /e/. This left seven vowels in tonic position in spoken Latin: /i/, /e/, /ɛ/, /a/, /ɔ/, /o/, /u/)

COLUMNA NON COLOMNA

columna (merger of /ʊ/ and /o/ in favour of /o/: evidence from the Appendix Probi)

OCTŌ

ocho (metaphony (=assimilatory raising of vowels by one degree of aperture in anticipation of a high vowel or a glide, which in this case emerges through the palatalization of syllable-final /k/) of tonic vowels due to the emergence of yod 1:

/ɔ/ + off glide [i̯] in the same syllable > [o] ([‘ɔi̯to] > [‘otʃo])

Note that yod 1 affects all possible vowels other than /e/, a deep effect that is probably down to the survival of the glide for centuries where it arose through palatalization of /k/ before /t/ and /s/, through metathesis or through loss of an intervocalic consonant)

LĀICU

lego (metaphony (=assimilatory raising of vowels by one degree of aperture in anticipation of a high vowel or glide, which in this case emerges through the reduction of /i/ in hiatus with /a:/) of tonic vowels due to the emergence of yod 1 + off glide [i̯] in the same syllable:

/a/ + off glide /i̯/ in the same syllable > /e/

Note that yod 1 affects all possible vowels other than /e/)

NERVIU

nervio (metaphony (=assimilatory raising of vowels by one degree of aperture in anticipation of a high vowel or glide, which in this case is the on glide [j] which results from earlier atonic /i/ in hiatus) of tonic vowels due to yod 2:

/ɛ/ + intervening consonant(s) + on glide [j] in the following syllable > /e/.

Note that yod 2 affects all possible vowels other than /a/)

MĀTERIA

madera (metaphony (=assimilatory raising of vowels by one degree of aperture in anticipation of a high vowel or glide, which in this case is the on glide [j] which results from earlier /i/ in hiatus) of tonic vowels due to yod 2:

/ɛ/ + intervening consonant(s) + on glide [j] in the following syllable > /e/.

Note that yod 2 affects all possible vowels other than /a/)

VINDĒMIA

vendimia (metaphony (=assimilatory raising of vowels in anticipation of a high vowel or glide, in this case the on glide [j] resulting from from earlier /i/ in hiatus) of tonic vowels due to yod 2:

/e/ + intervening consonant(s) (m) + on glide [j] in the following syllable > /i/.

Note that yod 2 affects all possible vowels other than /a/)

PODIU

poyo (metaphony of tonic vowels due to yod 3, where the vowel is followed by [ǰ] (which here is the result of earlier [dj]):

/ɔ/ + [ǰ] (>[dj]) in the following syllable > /o/.

Note that yod 3 affects only /ɛ/ and /ɔ/.)

FOVEA

hoya (metaphony of tonic vowels due to yod 3, where the vowel is followed by [ǰ] (which here is the result of earlier [bj]):

/ɔ/ + [ǰ] (>[bj]) in the following syllable > /o/.

Note that yod 3 affects only /ɛ/ and /ɔ/. The fact that only two vowels are affected perhaps suggests a fairly rapid (but not immediate) absorption of the glide into the adjacent /d/, /g/ or /b/)

FOLIA

hoja (metaphony of tonic vowels due to yod 4, where the vowel is followed by [ʎ], which here is the result of earlier [lj]:

/ɔ/ + [ʎ] (>[lj]) > /o/.

Note that yod 4 affects only /ɛ/ and /ɔ/. The fact that only two vowels are affected perhaps suggests a fairly rapid (but not immediate) absorption of the glide into the adjacent /l/)

CUNEU

cuño (metaphony of tonic vowels due to yod 5, where the vowel is followed by [ɲ] (from earlier [nj] or [gn]:

/o/ (product of merger between /ʊ/ and /o/) + [ɲ] (here product of [nj]) > /u/.

Note that yod 5 affects only /o/ (limited effect perhaps indicates that a glide adjacent to /n/ survived the shortest period (except for the glide following /t/ or /k/)).

PARIETE

pared (outcome of hiatus that is not metaphony: loss of the weaker vowel in hiatus)

OSSUM

hueso (diphthongization: first we see the lengthening of tonic vowels (but not atonic vowels), possibly because of Germanic influence; immediate effect of this is the raising of tonic vowels (other than /i/ and /u/). This process gives rise to potential confusion, so the diphthongs /ie/ and /ue/ were created in order to preserve comprehension. The glide [j] may not occupy syllable-initial position, so the result of diphthongization of word-initial /ε/ is a sequence of fricative consonant (/ǰ/) + /e/. The phonological effect of this is that two of the seven spoken Latin tonic vowels came to be interpreted as sequences of two other vowels or of fricative consonant plus vowel. This meant that the number of tonic vowel phonemes in Spanish was reduced to five.

In the case of hueso, the diphthong /ue/ was created in order to preserve comprehension as the lengthening and raising of tonic vowels other than /i/ and /u/ meant that [‘ɔso] ‘bone’ must have become very similar to [‘oso] ‘bear’. For this, the newly lengthened low-mid vowel /ɔ/ was moved downwards and then fractured to give [oɔ], and further differentiation led to the true diphthongs already observable in early Spanish and no doubt by then interpreted as sequences of two vowel phonemes: [oɔ] > [uɔ] > [wo] > [we] (=/ue/).)

BENE

bien (diphthongization: first we see the lengthening of tonic vowels (but not atonic vowels), possibly due to Germanic influence. The immediate effect of this is the raising of tonic vowels other than /i/ and /u/, a process that gives rise to potential confusion, so the diphthongs /ie/ and /ue/ were created in order to preserve comprehension. In the case of BENE, the /ε/ is moved downwards and then fractures to give [eε], and then further differentiation led to the true diphthongs already observable in early Spanish and no doubt by then interpreted as sequences of two vowel phonemes: [eε] > [iε] > [je] (=/ie/). The phonological effect of this is that two of the seven spoken Latin tonic vowels came to be interpreted as sequences of two other vowels (or of a fricative consonant followed by a vowel). This meant that the number of tonic vowel phonemes in Spanish was reduced to five)

PETRA

piedra (diphthongization: first we see the lengthening of tonic vowels (but not atonic vowels), possibly due to Germanic influence. The immediate effect of this is the raising of tonic vowels other than /i/ and /u/, a process that gives rise to potential confusion, so the diphthongs /ie/ and /ue/ were created in order to preserve comprehension. In the case of PETRA, the /ε/ is moved downwards and then fractures to give [eε], and then further differentiation led to the true diphthongs already observable in early Spanish and no doubt by then interpreted as sequences of two vowel phonemes: [eε] > [iε] > [je] (=/ie). The phonological effect of this is that two of the seven spoken Latin tonic vowels came to be interpreted as sequences of two other vowels (or of a fricative consonant followed by a vowel). This meant that the number of tonic vowel phonemes in Spanish was reduced to five)

EQUA

yegua (diphthongization: first we see the lengthening of tonic vowels, possibly due to Germanic influence. The immediate effect of this is the raising of tonic vowels other than /i/ and /u/, a process that gives rise to potential confusion, so the diphthongs /ie/ and /ue/ were created in order to preserve comprehension. In the case of EQUA, the glide [j] may not occupy syllable-initial position, so the result of diphthongization of word initial /ε/ or /hε/ was a sequence of the voiced palatal fricative consonant /ǰ/ + /e/: so EQUA [εkwa] > [ǰeɣwa])

HERBA

yerba/hierba (diphthongization: first we see the lengthening of tonic vowels, possibly due to Germanic influence. The immediate effect of this is the raising of tonic vowels other than /i/ and /u/, a process that gives rise to potential confusion, so the diphthongs /ie/ and /ue/ were created in order to preserve comprehension. In the case of EQUA, the glide [j] may not occupy syllable-initial position, so the result of diphthongization of word initial /ε/ or /hε/ was a sequence of the voiced palatal fricative consonant /ǰ/ + /e/: so HERBA [εɾba] > [ǰeɾba])

POENA

pena (treatment of Latin diphthongs: OE > /e/ in spoken Latin, i.e. the diphthong is reduced to a simple vowel)

CAELUM

cielo (treatment of Latin diphthongs: AE > /ε/ in spoken Latin, that is, early enough for the product of AE to be involved in the diphthongization process. In a few cases, AE was reduced to /e/, giving rise to undiphthongized outcomes: e.g. SAETA > /séta/ > seda)

PAUCU

poco (treatment of Latin diphthongs: AU > /o/. Note that this change was at first limited to non-patrician varieties and to certain words, and it was not until later that reduction of AU became regular in Spanish)

TAURU

toro (treatment of Latin diphthongs: AU > /o/. Note that this change was at first limited to non-patrician varieties and to certain words, and it was not until later that reduction of AU became regular in Spanish)

MĀTERIA

madera (creation of new diphthongs: as a result of the transfer of a glide to a preceding syllable, MĀTERIA came to be pronounced /matɛiɾa/. However, because the diphthong consisted of a front vowel or /a/ + palatal glide, there was a reduction of the sequence, through assimilation, to /e/, so /matɛi̯ɾa/ > /madeɾa/)

CORIU

cuero (creation of new diphthongs: as a result of the transfer of a glide to the preceding syllable, CORIU came to be pronounced /kɔi̯ro/. This new diphthong consisted of a back vowel + palatal glide, and was dramatically modified to /ue/, no doubt in part because by this time the diphthong /ue/ (the product of tonic Ŏ) was A) extremely frequent and B) because diphthongs stressed on the first element were rare)

DŌRIU

Duero (creation of new diphthongs: as a result of the transfer of a glide to the preceding syllable, DŌRIU came to be pronounced /doi̯ro/. This new diphthong consisted of a back vowel + palatal glide, and was dramatically modified to /ue/, no doubt in part because A) by this time the diphthong /ue/ (the product of tonic Ŏ) was extremely frequent and B) because diphthongs stressed on the first element were rare)

HABUĪ

ove/hube (creation of new diphthongs: it was possible for a labiovelar glide ([w]) to be transferred to a preceding syllable. In this case it would combine with tonic /a/ to produce /o/. However, the only examples of this are certain irregular preterites)

CASTELLU

castiello/castillo (modification of tonic vowels in the medieval period: after diphthongization and the emergence of the system of five tonic vowels, there were no further modifications to this system, but there were some modifications of individual vowels, with /ié/ occasionally reduced to /í/ (and /ué/ to /é/). /ié/ was reduced to /i/ in Old Spanish principally when followed by /ʎ/, probably through assimilation of /e/ to the higher phonemes (/i/ and /ʎ/) by which it was surrounded. Other examples include CULTELLU > cuchiello > cuchillo and SELLA > siella > silla)

FLOCCU

flueco/fleco (‘fringe, tassel’) (modification of tonic vowels in the medieval period: after diphthongization and the emergence of the system of five tonic vowels, there were no further modifications to this system, but there were some modifications of individual vowels, with /ué/ occasionally reduced to /é/, probably for reasons of assimilation, after /ɾ/ or /l/, so FLOCCU ([flɔko] > [flwéko] > [fléko]. Other examples include FRONTE > OSp. fruente > frente and COLOBRA > culuebra > culebra)

PLICĀRE

llegar (atonic vowels in initial position: Ĕ and Ŏ merged early with their long counterparts, so the conditions for diphthongization (i.e. the need to keep the low-mid vowels, /ɛ/ and /ɔ/, separate from the high-mid vowels, /e/ and /o/) did not arise, meaning that diphthongization could not take place.

Atonics in initial position are less energetically articulated than tonic vowels, but are the most energetic of the atonics. They therefore show a greater degree of merger than the tonics but less than other atonics. These mergers probably began earlier, but are not clearly attested until the third century AD, where they are revealed by statements in the Appendix Probi such as SIRENA NON SERENA.

In this position, the ten vowels of literary Latin were reduced in spoken Latin to /i/ (from Ī), /e/ (from Ĭ, Ĕ and AE), /a/ (from Ā and Ă), /o/ (from Ō, Ŏ, OE, AU and Ŭ) and /u/ (from Ū). In the case of PLICĀRE, the change is Ĭ > /e/)

SĒCŪRU

seguro (atonic vowels in initial position: Ĕ and Ŏ merged early with their long counterparts, so the conditions for diphthongization (i.e. the need to keep the low-mid vowels, /ɛ/ and /ɔ/, separate from the high-mid vowels, /e/ and /o/) did not arise, meaning that diphthongization could not take place.

Atonics in initial position are less energetically articulated than tonic vowels, but are the most energetic of the atonics. They therefore show a greater degree of merger than the tonics but less than other atonics. These mergers probably began earlier, but are not clearly attested until the third century AD, where they are revealed by statements in the Appendix Probi such as SIRENA NON SERENA.

In this position, the ten vowels of literary Latin were reduced in spoken Latin to /i/ (from Ī), /e/ (from Ĭ, Ĕ and AE), /a/ (from Ā and Ă), /o/ (from Ō, Ŏ, OE, AU and Ŭ) and /u/ (from Ū). In the case of PLICĀRE, the change is Ē > /e/)

RĒNIŌNE (from imparisyllabic RĒN)

riñón (atonic vowels in initial position: metaphony also occurs in the case of initial atonic vowels. A palatal glide [j] in the following syllable means that Ē in atonic position, rather than giving /e/, is raised to /i/ because of metaphony/assimilation in anticipation of the glide, so RĒNIŌNE > [re’njone] > [ri’ɲon])

CŌGNĀTU

cuñado (atonic vowels in initial position: metaphony also occurs in the case of initial atonic vowels. A palatal glide (the result of Latin GN) means that Ō in atonic position, rather than giving /o/, is instead raised to /u/ because of metaphony/assimilation in anticipation of the glide, so CŌGNĀTU > [koi̯’nato] > [kuɲaðo])

MANŪS

mano (atonic vowels in final position: the change in nature of the Latin accent (from a pitch-accent (where pitch is the predominant element and all the vowels have a similar loudness, i.e. the energy/stress is more uniformly deployed) to a stress-accent (where energy-deployment is the dominant element. The increasingly uneven deployment of energy over the word (more to the tonic, less to the atonics) accounts in large part for the different historical treatment of the Latin vowels in different positions, as tonic vowels are well-differentiated and preserved because of the concentration of energy on them (and the greater audibility this brings), while lesser degrees of energy devoted to initial, final and intertonic vowels (in decreasing order) imply greater degrees of merger and loss) brought a severe decrease of energy to final syllables. The result was intense merger and some loss. Where Latin final vowels survived into Spanish, they had the following outcomes: /e/ (from Ī, Ĭ, Ē, OE and Ĕ), /a/ (from Ā and Ă) and /o/ (from Ŏ, Ō, AU, Ŭ and Ū), although there is evidence that the contrast between final /i/ and /e/ and final /u/ and /o/ was maintained in spoken Latin)

PATRĒS

padres (atonic vowels in final position: the change in nature of the Latin accent (from a pitch-accent to a stress-accent) brought a severe decrease of energy to final syllables. The result was intense merger and some loss. Where Latin final vowels survived into Spanish, they had the following outcomes: /e/ (from Ī, Ĭ, Ē, OE and Ĕ), /a/ (from Ā and Ă) and /o/ (from Ŏ, Ō, AU, Ŭ and Ū), although there is evidence that the contrast between final /i/ and /e/ and final /u/ and /o/ was maintained in spoken Latin)

VĒNĪ

vine (atonic vowels in final position: the change in nature of the Latin accent (from a pitch-accent to a stress-accent) brought a severe decrease of energy to final syllables. The result was intense merger and some loss. Where Latin final vowels survived into Spanish, they had the following outcomes: /e/ (from Ī, Ĭ, Ē, OE and Ĕ), /a/ (from Ā and Ă) and /o/ (from Ŏ, Ō, AU, Ŭ and Ū). Note that final Ī gives rise to metaphony (so Ē > /i/), which is evidence that the contrast between final /i/ and /e/ was maintained in spoken Latin)

MEA (atonic vowels in final position)

mía/míe/mi (atonic vowels in final position: the change in nature of the Latin accent (from a pitch-accent to a stress-accent) brought a severe decrease in energy to final syllables. The result was intense merger and some loss. Where Latin final vowels survived into Spanish, they had the following outcomes: /e/ (from Ī, Ĭ, Ē, OE and Ĕ), /a/ (from Ā and Ă) and /o/ (from Ŏ, Ō, AU, Ŭ and Ū). In Old Spanish, final /a/ (and final /o/) is relatively stable: when in hiatus with a high tonic vowel, /a/ is raised to /e/ and occasionally lost, thus the varied forms of the first person feminine possessive pronoun mía, míe and mi (of which mi would win out) (other examples include duas > dues and the imperfect endings of -er/-ir verbs: -ia > -ie))

TERTIĀRIUS

tercer (atonic vowels in final position: the change in nature of the Latin accent (from a pitch-accent to a stress-accent) brought a severe decrease in energy to final syllables. The result was intense merger and some loss. Where Latin final vowels survived into Spanish, they had the following outcomes: /e/ (from Ī, Ĭ, Ē, OE and Ĕ), /a/ (from Ā and Ă) and /o/ (from Ŏ, Ō, AU, Ŭ and Ū). In Old Spanish, final /o/ (and final /a/) is relatively stable: it is lost in a few words which habitually preceded a noun or adjective. Other examples: SĀNCTUS > san, DOMINUS > don. Evidence in the Glosas emilianenses: the form ‘terzero’ appears in the gloss for tertius ueniens: ‘elo terzero [diabolo] uenot’, whereas in the PMC the apocopated form tercer appears)

PARIETE

pared (atonic vowels in final position: the change in nature of the Latin accent (from a pitch-accent to a stress-accent) brought a severe decrease in energy to final syllables. The result was intense merger and some loss. Where Latin final vowels survived into Spanish, they had the following outcomes: /e/ (from Ī, Ĭ, Ē, OE and Ĕ), /a/ (from Ā and Ă) and /o/ (from Ŏ, Ō, AU, Ŭ and Ū). In Old Spanish, final /e/ is highly unstable (unlike final /o/ and final /a/): it suffers elimination in two periods of the history of Old Spanish. The first period was can be seen in the pre-literary texts of the 10th and 11th centuries, where /e/ is lost after a dental or alveolar consonant (except /t/) which at that stage was ungrouped. Other examples include: MERCĒDE > merced, PĀNE > pan and MARE > mar (but SEPTEM > siete). Later, especially in the thirteenth century, urban, educated varieties of Castilian suffered sporadic loss of /e/ in other phonological environments, possibly under the influence of French: examples include nuef (nueve) and noch (noche). However, the forms with /e/ never disappeared from the written record and likely continued to characterize popular styles of Castilian speech, which reasserted themselves towards the end of the thirteenth century and have been perpetuated into the modern language)

PISCE

peçe/pez (atonic vowels in final position: the change in nature of the Latin accent (from a pitch-accent to a stress-accent) brought a severe decrease in energy to final syllables. The result was intense merger and some loss. Where Latin final vowels survived into Spanish, they had the following outcomes: /e/ (from Ī, Ĭ, Ē, OE and Ĕ), /a/ (from Ā and Ă) and /o/ (from Ŏ, Ō, AU, Ŭ and Ū). In Old Spanish, final /e/ is highly unstable (unlike final /o/ and final /a/): it suffers elimination in two periods of the history of Old Spanish. The first period was can be seen in the pre-literary texts of the 10th and 11th centuries, where /e/ is lost after a dental or alveolar consonant (except /t/) which at that stage was ungrouped. Other examples include: MERCĒDE > merced, PĀNE > pan and MARE > mar (but SEPTEM > siete). Later, especially in the thirteenth century, urban, educated varieties of Castilian suffered sporadic loss of /e/ in other phonological environments, possibly under the influence of French: examples include nuef (nueve) and noch (noche). However, the forms with /e/ never disappeared from the written record and likely continued to characterize popular styles of Castilian speech, which reasserted themselves towards the end of the thirteenth century and have been perpetuated into the modern language. The word pez is one of the few apocopated forms from this period that were preferred in the modern language. All of these forms show a consonant group preceding /e/ that was simplified in later Old Spanish to /ts/ = ç. This probably took place because, after the loss of /e/, these words ended with the same (now ungrouped) phoneme as many other previously existing words, e.g. paz and cruz)

SPECULUM

espejo (treatment of intertonic vowels in atonic position: these are the vowels which suffer most from the development of a stress-accent in spoken Latin. With the exception of the vowel /a/, Latin intertonic vowels have been entirely eliminated. This began in spoken Latin, where intertonic vowels in contact with /ɾ/ or /l/ (and sometimes with /n/ or /s/) were frequently lost. This conditioned loss is attested in the Appendix Probi: SPECULUM NON SPECLUM)

SEPTIMĀNA

semana (treatment of intertonic vowels in atonic position: these are the vowels which suffer most from the development of a stress-accent in spoken Latin. With the exception of /a/ (e.g. RAPHANU > rábano), Latin intertonic vowels have been entirely eliminated. In pre-literary Spanish, almost all surviving intertonics other than /a/ were eliminated. Other examples include MANICA > *mánega > manga and TEMPORĀNU > temprano). Note that the consonants brought into contact by the loss of intertonic vowels frequently undergo modification)

PARIETE

pared (treatment of hiatuses: the shift from a pitch-accent to a stress-accent militated against hiatus, which requires that its constituent syllables should be of similar intensity. In a stress-accent, however, the neighbouring vowels received markedly different degrees of energy, and cases of hiatus were reduced to monosyllabic pronunciation, either by loss of one of the vowels or by reducing the higher of the two to a (non-syllabic) glide (the latter was more frequent). The outcome pared is an example of the vowel /i/ being lost, but others include DUŌDECIM > doce and QUIĒTUS > quedo)

LANCEA

lanza (treatment of hiatuses: the shift from a pitch-accent to a stress-accent militated against hiatus, which requires that its constituent syllables should be of similar intensity. In a stress-accent, however, the neighbouring vowels received markedly different degrees of energy and cases of hiatus were reduced to monosyllabic pronunciation. This was achieved either by loss of one of the vowels or, more frequently, by reducing the higher of the two to a (non-syllabic) glide. The weaker vowel was usually the first, and if it was a front vowel it was converted into the front (i.e. palatal) glide [j], while if it was a back vowel it was converted into the back (i.e. labiovelar) glide [w]. This change was a case of neutralization, as all front/back vowels came to be realized in the same way when one of them was originally the weaker member of a sequence of vowels in hiatus. This neutralization produced uncertainty of spelling, and alternation between the spellings E/I and O/U under these conditions was common in non-literary Latin. In an attempt to redress this uncertainty, the Appendix Probi comments: LANCEA NON LANCIA, CLOACA NON CLUACA. The impact of the glide on consonants led to the creation of a new order of palatal consonants, thus making the consonant system more complex)

IŪSTUS

justo (creation of palatal order: word-initial /i/ was non-syllabic from early times in Latin, probably articulated as the frictionless glide [j], so IŪSTUS was pronounced [ju:stus]. In spoken Latin, the frictionless glide [j] became a consonant, presumably the voiced palatal fricative [ǰ]. The fricative nature of the initial phoneme is reflected in misspellings such as SUSTUS. The appearance of the voiced palatal fricative as the phoneme represented by word-initial I in spoken Latin can be considered the first step in the creation of a palatal order of consonants)

MARTIU

março (creation of palatal order: /t/ + [j] > [ts], so MARTIU > [maɾtso]. Note that the development of the glide is the early palatal stage of development, but the outcome /ts/ is a dento-alveolar phoneme)

PUTEU

pozo (creation of palatal order: /t/ + [j] > [ts], so PUTEU > [potso]. As /ts/ is preceded by a vowel and is therefore intervocalic, it undergoes voicing, so [potso] > [podzo]. Note that the development of the glide is the early palatal stage of development, but the outcome /dz/ is a dento-alveolar phoneme)

CALCEA

calça (creation of palatal order: /k/ + [j] > [tʃ], so CALCEA > [kalʃa]. This /tʃ/ then merges with /ts/ in favour of /ts/, so [kalʃa] > [kaltsa]. Note that the development of the glide is the early palatal stage of development, but the outcome /ts/ is a dento-alveolar phoneme)

ĒRĪCIU

erizo (creation of palatal order: /k/ + [j] > [tʃ], so ĒRĪCIU > [eɾitʃo]. This /tʃ/ then merges with /ts/ in favour of /ts/, so [eɾitʃo] > [eɾitso]. As the /tʃ/ is preceded by a vowel (and is therefore intervocalic), it undergoes voicing, so [eɾitso] > [eɾidzo]. Note that the development of the glide is the early palatal stage of development, but the outcome /dz/ is a dento-alveolar phoneme)

MATTIĀNA

mançana (creation of palatal order: geminate /tt/ was also palatalized by following [j]. This produced in spoken Latin the geminate [tts], which merged with the product of the following: /kk/ + [j], /pt/ + [j], /kt/ + [j] and /sk/ + [j] (all of which also gave [tts]). Like other geminates, [tts] was simplified through lenition to /ts/ in Old Spanish, spelt ç or c. Note that the development of the glide is the early palatal stage of development, but the outcome /ts/ is a dento-alveolar phoneme)

BRACCHIU

braço (creation of palatal order: geminate /kk/ was also palatalized by following [j]. This produced in spoken Latin the geminate [tts], which merged with the product of the following: /tt/ + [j], /pt/ + [j], /kt/ + [j] and /sk/ + [j] (all of which also gave [tts]). Like other geminates, [tts] was simplified through lenition to /ts/ in Old Spanish, spelt ç or c. Note that the development of the glide is the early palatal stage of development, but the outcome /ts/ is a dento-alveolar phoneme)

FOLIA

foja/foia (later hoja) (creation of palatal order: /l/ + [j] > /ʒ/. /l/ + [j] widely became the palatal lateral [ʎ] in spoken Latin. In pre-literary Castilian, this [ʎ] was modified to the voiced pre-palatal fricative /ʒ/, possibly out of a need to maintain contrast/avoid merger between words originally containing /l/ + [j] and those containing the geminate /ll/, since the /ll/ in words like GALLU had also become [ʎ] in spoken Latin)

ARĀNEA

araña (creation of palatal order: /n/ + [j] > /ɲ/, spelt nn or with the contraction ñ)

PODIU

poyo (creation of the palatal order: /d/ + [j] > [ǰǰ] (midpalatal fricative geminate), which is subject to reduction through lenition and merges with intervocalic -I-. It is spelt y in Modern Spanish and and y and i in late Old Spanish and early Modern Spanish)

VIDEŌ

veo (creation of the palatal order: /d/ + [j] > [ǰǰ] (midpalatal fricative geminate), which is subject to reduction through lenition (>[ǰ]) and merges with intervocalic -I-. However, when preceded by a front vowel, the consonant is lost through assimilation to the vowel)

EXAGIU

ensayo (creation of the palatal order: /g/ + [j] > [ǰǰ] (midpalatal fricative geminate), which is subject to reduction through lenition (>[ǰ]) and merges with intervocalic -I-. It is spelt y in Modern Spanish and and y and i in late Old Spanish and early Modern Spanish)

HABEAM

haya (present subjunctive of haber) (creation of the palatal order: the labials /b/ + [j] and /m/ + [j] were largely unaffected by palatalization, which occurs in a very few cases. Otherwise, the consonants survive unchanged (apart from the fricatization of /b/) into Old and Modern Spanish, e.g. NOVIU > novio and PRAEMIU > premio)

BĀSIU

beso (palatalization: /s/ + [j] (as well as /p/ + [j] and /ɾ/ + [j]) were subject only to metathesis, without palatalization of the consonant (though the presence of the glide led to the metaphony/raising of /a/ to /e/). Other examples include CAPIAM > quepa and AUGURIU > agüero)

PĀCE

paz (creation of the palatal order: /k/ > /tʃ/ > /ts/. Syllable-initial velars /k/ and /g/ were palatalized in later spoken Latin when followed by a front vowel. They contributed to the creation of the Old Spanish phonemes /ts/, /dz/ and /ǰ/. Latin /k/ probably always had a fronted allophone used before a front vowel (/i/, /e/ or /ɛ/). In spoken Latin, this assimilatory fronting process was continued and exaggerated, so that the allophone of /k/ used before front vowels came to be palatal [tʃ]. In some areas, including Spain, it was further fronted to [ts].

This change had consequences for the phonemic system of Latin. Firstly, it implies merger of /k/ before a front vowel with /t/ + [j] and /k/ + [j]. So PĀCE [‘patse] has the same internal consonant as FACIĒ [‘fatse].

Secondly, the sound [ts] could now occur before any of the spoken Latin vowels, and was therefore well on the way to phonologization. When QUI-, QUE- became [ki] and [ke] (as the sequence QU retained the glide [w] for a sufficiently long period to ensure that this /k/ is always treated like /k/ preceding a non-front vowel, e.g. QUĪNDECIM > quince), /ts/ was fully phonologized, contrasting with /k/ in all environments, e.g. /ts/ in PRETIĀRE [pre’tsare] was in minimal opposition to the /k/ of PRĒCĀRE

Note that the development of the glide is the early palatal stage of development, but the outcome /ts/ is a dento-alveolar phoneme)

CIRCĀ

cerca (creation of the palatal order: /k/ > /tʃ/ > /ts/. Syllable-initial velars /k/ and /g/ were palatalized in later spoken Latin when followed by a front vowel, but remained velar when followed by a back vowel. So CIRCĀ [kiɾka:] > [tʃeɾka] > [tseɾka]. Note that the development of the glide is the early palatal stage of development, but the outcome /ts/ is a dento-alveolar phoneme)

DĪCIT

dize (creation of the palatal order: /k/ > /tʃ/ > /ts/. Syllable-initial velars /k/ and /g/ were palatalized in later spoken Latin when followed by a front vowel, but remained velar when followed by a back vowel. So DĪCIT [di:kit] > [ditʃit] > [ditsit]). As the /k/ was intervocalic, it was subject to lenition, and therefore appears in Old Spanish as the voiced phoneme /dz/ (spelt z). Before back vowels, it remained velar and became voiced /g/ in intervocalic position, e.g. IOCU > juego, meaning that Latin /k/ in this position split into two Old Spanish phonemes, /dz/ and /g/)

FLACCIDU

llacio (later lacio) (creation of the palatal order: few words with geminate /kk/ before a front vowel have survived into Spanish, but the Latin group /sk/ was frequent before such vowels. Both groups became geminate /tts/ in Cantabrian spoken Latin, thereby merging with the product of /kk/ + [j], /tt/ + [j], etc., and producing Old Spanish /ts/ (spelt (s)ç or (s)c) through the usual Western Romance simplification of geminates)

PISCĒS

peces (creation of the palatal order: few words with geminate /kk/ before a front vowel have survived into Spanish, but the Latin group /sk/ was frequent before such vowels. Both groups became geminate /tts/ in Cantabrian spoken Latin, thereby merging with the product of /kk/ + [j], /tt/ + [j], etc., and producing Old Spanish /ts/ (spelt (s)ç or (s)c) through the usual Western Romance simplification of geminates)

GEMMA

yema (creation of the palatal order: like voiceless /k/, the voiced velar /g/ no doubt always had fronted allophones when followed by a front vowel. Likewise, these fronted allophones were further fronted in later spoken Latin and became the voiced palatal fricative [ǰ]. This meant that the product of /g/ before front vowels in word-initial position merged with the product of non-syllabic I-. The result was that words like GEMMA and IŪSTUS had the same initial sound. Palatal /ǰ/ from these sources was sometimes preserved in word-initial position in Old Spanish, as was the case with yema)

(However, note that loss or modification of /ǰ/ was frequent in early Old Spanish, e.g. the atonic /ǰe/ of words like *yermano would have appeared anomalous (because /ǰe/ was usually the result of the diphthongization that was limited to tonic syllables, e.g. EQUA > yegua) and was thus replaced with /e/, giving ermano (later hermano))

GERMĀNUS

hermano (creation of the palatal order: /g/ always had fronted allophones when followed by a front vowel, which were further fronted in later spoken Latin and became the voiced palatal fricative [ǰ], thus merging with the product of non-syllabic I-. Palatal /ǰ/ from these sources was sometimes preserved in word-initial position in Old Spanish, e.g. GEMMA > yema. However, loss or modification of /ǰ/ was frequent in early Old Spanish. One reason for this was because word-initial /ǰe/ in an atonic syllable would have appeared anomalous, as it was more common for /ǰe/ to arise due to the diphthongization process, which was limited to tonic syllables. Therefore, the atonic /ǰe/ of words like *yermano was brought into line with the prevailing morpho-phonological pattern by replacing /ǰe/ by /e/)

(Note that another reason for the modification of /ǰ/ was possibly the influence of Medieval Latin, e.g. in IŪSTU > yusto > justo, /ǰ/ (the product of non-syllabic /i/) is modified to /ʒ/, spelt as j or i.)

SAGITTA

saeta (creation of the palatal order: intervocalic /g/ before front vowels was also palatalized, but this time to simple /ǰ/ rather than to the geminate /ǰǰ/. The /ǰ/ was rapidly eliminated, either by total assimilation to the following vowel, or through the lenition process. However, before non-front vowels intervocalic /g/ survived as a velar sometimes, but was also lost at times (e.g. NEGĀRE > negar vs LIGĀRE > liar). This means that the effects of the phonemic split (where Latin /g/ gave palatal results in initial position before front vowels, but remained velar before other vowels) are partially obscured in intervocalic position)

DULCE

dulce (creation of the palatal order: when /k/ and /g/ occurred as the second element of an internal group in Latin, in principle they evolve in the same way as in word-initial position. Thus, in the case of consonant + /k/, we see the same phonemic split as in the case of initial /k/, namely /k/ before non-front vowels and Old Spanish /ts/ before front vowel)

IŪNCU

junco (creation of the palatal order: when /k/ and /g/ occurred as the second element of an internal group in Latin, in principle they evolve in the same way as in word-initial position. Thus, in the case of consonant + /k/, we see the same phonemic split as in the case of initial /k/, namely /k/ before non-front vowels and Old Spanish /ts/ before front vowel. Note that cases of consonant + /g/ are notoriously thorny, especially given that many of the examples are verbs, where analogical interference with phonological development is likely to be involved. For example, before a front vowel, /ng/ gives no fewer than three results: GINGĪVA > enzia > encía (i.e. /ng/ > /ndz/ > /nθ/), RINGERE > reñir (i.e. /ng/ > /ŋ/), and QUĪNGENTŌS > quinientos (i.e. /ng/ > /n∅/, loss)

CABALLU

cavallo (later caballo) (creation of the palatal order: the Latin geminate /ll/ was palatalized to /ʎ/)

ANNU

año (creation of the palatal order: the Latin geminate /nn/ was palatalized to /ɲ/)

PLĀGA

llaga (creation of the palatal order: word initial PL- was palatalized to /ʎ/ (as was word-initial CL- and FL-). It is likely that the /l/ of these groups already had a palatalized pronunciation in the spoken Latin of some areas. In pre-literary Spanish, the initial consonant was in most cases assimilated to the following /ʎ/.

However, there are some popular words, e.g. PLATEA > plaça, CLĀVICULA > clavija and FLOCCU > flueco, which show retention of the group)

CLĀMĀRE

llamar (creation of the palatal order: word-initial CL- was palatalized to /ʎ/ (as was word-initial PL- and FL-). It is likely that the /l/ of these groups already had a palatalized pronunciation in the spoken Latin of some areas. In pre-literary Spanish, the initial consonant was in most cases assimilated to the following /ʎ/.

However, there are some popular words, e.g. PLATEA > plaça, CLĀVICULA > clavija and FLOCCU > flueco, which show retention of the group)

FLAMMA

llama (creation of the palatal order: word-initial FL- was palatalized to /ʎ/ (as was word-initial PL- and CL-). It is likely that the /l/ of these groups already had a palatalized pronunciation in the spoken Latin of some areas. In pre-literary Spanish, the initial consonant was in most cases assimilated to the following /ʎ/.

However, there are some popular words, e.g. PLATEA > plaça, CLAVĪCULA > clavija and FLOCCU > flueco, which show retention of the group)

PLATEA

plaça (later plaza) (creation of the palatal order: word-initial PL- was palatalized to /ʎ/ (as was word-initial CL- and FL-). It is likely that the /l/ of these groups already had a palatalized pronunciation in the spoken Latin of some areas. In pre-literary Spanish, the initial consonant was in most cases assimilated to the following /ʎ/.

However, there are some popular words, e.g. PLATEA > plaça, CLAVĪCULA > clavija and FLOCCU > flueco, which show retention of the group)

CLAVĪCULA

clavija (creation of the palatal order: word-initial CL- was palatalized to /ʎ/ (as was word-initial PL- and FL-). It is likely that the /l/ of these groups already had a palatalized pronunciation in the spoken Latin of some areas. In pre-literary Spanish, the initial consonant was in most cases assimilated to the following /ʎ/.

However, there are some popular words, e.g. PLATEA > plaça, CLAVĪCULA > clavija and FLOCCU > flueco, which show retention of the group)

FLOCCU

flueco (later fleco) (creation of the palatal order: word-initial FL- was palatalized to /ʎ/ (as was word-initial PL- and CL-). It is likely that the /l/ of these groups already had a palatalized pronunciation in the spoken Latin of some areas. In pre-literary Spanish, the initial consonant was in most cases assimilated to the following /ʎ/.

However, there are some popular words, e.g. PLATEA > plaça, CLAVĪCULA > clavija and FLOCCU > flueco, which show retention of the group)

VĪTA

vida [βiða] (appearance of /β/: just as non-syllabic I was consonantized to the voiced fricative /ǰ/ (and merged with the product of /d/ + [j] and /g/ + [j]), so too was non-syllabic V ([w]) in words like VĪTA [wi:ta] and AVIS [a:wis] consonantized to the voiced bilabial fricative /β/.

In intervocalic position, the difference between the product of non-syllabic V and /b/ was neutralized in spoken Latin, as suggested by the frequent interchange of the spellings b and v.

There was some neutralization of initial B- and V- in spoken Latin, but it seems that in initial position the Latin phonemes represented by B- and V- maintained their contrast into Old Spanish period, since Old Spanish spelling practice for the most part continues to distribute initial b and v according to whether the etymon had B- or V-, e.g. BUCCA > boca and VACCA > vaca. That said, in a minority of instances, Old Spanish spelling does show confusion of initial b- and v: this is sometimes due to the dissimilation of initial /β/ from a following /β/, e.g. VĪVERE > bevir, and sometimes due to the increasing neutralization of the two phonemes, e.g. boz for voz.

Full merger was not accomplished until the late medieval and early modern periods).

CABALLU

cavallo (later caballo) (appearance of /β/: the extraordinary frequency of interchange of the spellings b and v suggests that V was identical to B in intervocalic position (both likely pronounced as the voiced fricative /β/) in spoken Latin, e.g. PLEBES NON PLEVIS from the Appendix Probi. This pronunciation was inherited by Old Spanish and spelt v or b)

SICCU

seco (lenition: beginning in the last centuries of the Empire & continuing through the Dark Ages, Western Romance was affected by lenition (interrelated series of consonantal changes). Almost all intervocalic consonants & all geminates were involved in these changes. One of major outcomes of lenition was the provision of further voiced fricative phonemes: /β/, /z/, /ʒ/ and /ǰ/.

Reason for lenition changes is much disputed: some see in them the influence upon spoken Latin of Celtic speech, since similar weakenings of intervocalic consonants occur in the history of the Celtic languages, and there is considerable (but not complete) correspondence between areas of originally Celtic population and the areas of Romance speech where lenition is evident, i.e. Gaul, northern Italy, western Spain.

Others seek purely Latin explanations: increase in spoken Latin of the incidence of geminates led to an unbalanced consonantal system, and this imbalance was redressed by the simplification of the geminates. This process caused a chain-reaction of further changes: voicing of voiceless intervocalics and fricatization/loss of voiced intervocalics.

Increase in incidence of geminates in spoken Latin is well attested, e.g. Appendix Probi: PERSICA NON PESSICA. Other examples: DORSUM > DOSSUM, IPSE > ISSE. Following this increase, the geminates were reduced to simple intervocalic consonants, e.g. /kk/-/k/. This simplification put pressure on the existing intervocalic consonants to change, so originally voiceless simple intervocalics became voiced, e.g. /k/ > /g/ in order to maintain the original contrast between geminate and simple. In turn, the voicing of voiceless intervocalic phonemes threatened merger with the pre-existing voiced phonemes, so the latter (if originally plosive) became fricative, e.g. /g/ > /ɣ/), in order to avoid merger. The chain reaction was completed when the pre-existing voiced fricative /ǰ/ (result of palatalization of /g/ before front vowels) was eliminated from the words in which it occurred in order to avoid merger with the new voiced fricatives.

The simplification of geminates was contemporary with the voicing of the plosives, so /kk/ aws reduced to /k/ and then remained a voiceless plosive, i.e. it was not affected by voicing, fricatization or loss.

Note that the Latin geminates /nn/, /ll/ and /ɾɾ/ were not reduced to the corresponding simple consonants. They were simplified, but the resulting phoneme differed somehow from the predicted phoneme, probably because the Latin simple consonants /n/, /l/ and /ɾ/ were neither geminate, voiceless, voiced plosives nor fricatives, i.e. they did not qualify as the input of any of the four lenition processes. The consequence of this lack of change is that straightforward simplification of /nn/, /ll/ and /ɾɾ/ would automatically bring about merger of the geminates with the simple consonants. No doubt in order to preserve these distinctions, /nn/ and /ll/ were simplified and palatalized, while /ɾɾ/ was simplified by changing the spoken Latin sequence of syllable-final /-ɾ/ + syllable-initial /ɾ-/ to the single syllable-initial vibrant /r/.

Also note that when an internal consonant was preceded by a vowel and followed by a glide (or vice versa) at the spoken Latin stage, voicing of /p/, /t/ and /k/ sometimes, but not always fails to occur, e.g. /p/ is voiced when preceded by [u̯] (e.g. CAUSA > cosa ['koza]). However, when followed by [j], /p/ remained voiceless (e.g. SAPIAM > sepa))

CUPPA

copa (lenition: simplification of the geminates (process 1), /pp/ > /p/)

SCOFFĪNA

escofina (lenition: simplification of the geminates (process 1), /ff/ > /f/)

FLAMMA

llama (lenition: simplification of the geminates (process 1), /mm/ > /m/. Note that this is the only example of the total neutralization/merger of a contrast between geminate and simple consonant in Latin)

GUTTA (lenition: simplification of the geminates (process 1), /tt/ > /t/)

gota

*PETTIA

pieça (lenition: simplification of the geminates (process 1), /tts/ > /ts/. Thus the spoken Latin distinction between geminate and simple consonant was maintained in Old Spanish as a contract between voiceless and voiced consonant, spelt ç and z respectively. However, this contrast was lost in the sixteenth century)