Diphthongization in Romance Language

Introduction

Comparative linguistics is a field that has added to the seemingly chaotic diachronic change of languages a strong degree of systematism; and of this, no language family has been more studiously studied as the descendents of Latin. The vowel system of Classical Latin and its gradual modification have been of considerable fancy in philological circles, in that, the past two millennia have seen the fall of an inventory of vowel quantity and the rise of the modern Romance tendencies with an ostentatious demonstration of sound change especially moderated by diphthongization.

There is a sizable degree of predictability in the mutation of vowels which shall only be seen to be followed by confusion in the divergence in Romance Languages which produces the similar yet vexingly variant vowel inventories seen today. This paper is written to (1) highlight the changes in the vowel system that coincided with the decline in use of spoken Latin and (2) to seek similarities between the eventual outcomes of the Latin “speech-diaspora” and sequential variation.

The Latin Writing System and Underlying Phones

The process of reconstructing the Latin vowel system, which must be our point of inception in this discussion, has been millennia in the making and the modern consensus has arisen only after copious examination of the testimonies of classical grammarians, inscriptionary evidence and later phonetic change. Latinists now, with a heavy degree of certainty, can say that the classical language bore 10 phonemically distinct vowels which were represented by 5 graphemes (this is ignoring the non-native phoneme /y/ and its grapheme <y>).

The five vowels, which are needless to say <a, e, i, o, u>, had each “long” and “short” values. In the old classical period this was certainly vowel length in the modern linguistic sense, as can be ascertained from Latin metric poetry; that is, length was distinct by the time spent enunciating each vowel. It is, nonetheless, just as clear that by the first century A.D., the short forms of all vowels except for <a> had acquired phonetic values distinct from the long forms and each had begun to evolve in dissimilar ways. In fact, in Vulgar Latin, phonemic vowel length totally gives way to differentiation based on vowel quality, and thus it is necessary to approximate such phonetic values in the course of tracing the evolution of Pan-Romance.

Allen 1978, whose work Vox Latina is an important milestone in the corpus of Latinism, provides the most succinct summary of the reconstruction of the Latin vowel system. Short vowels in the system employed in Vox Latina deviate further and further from the original long forms the higher in the mouth they are pronounced. The long and short <a>, being pronounced low in the mouth are equal in phonetic quality, only dissimilar by length. The short mid vowels <e, o> exhibit are realized in [ɛ,ɔ] respectively but their differences are not as great as those between the long high vowels, whose long and short realizations are [u, i] and [ʊ, ɪ]. Below shows these contrasts in minimal pairs, which due to the vowel deficiency of the Latin alphabet are identical orthographically.

Orthography	With Short Vowel	With Long Vowel
MALUM	[malʊm] “bad”	[maːlʊm] “apple”
UENIT	[wɛnɪt] “he comes”	[weːnɪt] “he came”
HIC	[hɪk] "this"	[hiːk] “here”
OS	[ɔs] “bone”	[oːs] "mouth"
DOMUS	[dɔmʊs] “house”	[dɔmuːs] “houses”

This reconstruction is, of course, classical in nature and there needs to be several further steps taken further before the vowel inventory of Vulgar Latin or Proto-Romance can be established. As mentioned above, the extra length of the long vowels evaporates in the first century, thus [aː, eː, iː, oː, uː] > [a, e, i, o, u]. This means that the “long” and “short” /a/ are homophonous and merge at the beginning of the post-classical era, thus lātus “wide” and lǎtus “side” are indistinguishable etcetera; accordingly, no Romance language maintains any difference in pronunciation in such forms.

The Point of Divergence

It was originally posited that all Romance languages had experienced a further change in which the phonetic value of the short high vowels lower themselves such that they merge with the long mid vowels; /ʊ/ and /o/ become one phoneme: /o/ and /ɪ/ and /e/ become one phoneme: /e/. These two parallel transformations are indeed readily apparent in most modern Romance languages:

(Hall 23, 31) *The prenasal transformation to [i]exists in some languages.

Still, although this generalization is succinct for most areas of speech, it contradicts the data from peripheral dialects from Sicily and Sardinia (Elcock 45); Sardinian, for example, merges /ʊ/ into /u/ and /ɔ/ into /o/, thus realizing [bukka muska pira limba] (Hall 23, 31). Nevertheless, the seven vowel system produced from the mergers above has formed the working point for many comparative analyses and instructives of Romance linguistics (Fradejas 2010, Azevedo 2009). This outcome has garnered the title Western or Italo-Western Romance for the dialects exhibiting the above vowel shifts above while two other groups encompassing dialects less numerous (Southern Romance in Sardinia and Corsica and Eastern Romance in Romania and South Italy) have been coined to explain other dissimilarities. The Western evolution will encompass the majority of the following analysis.

The Evolution of New Diphthongs

The vocalic changes that occurred from the prescribed Classical to Vulgar Latin all share the common denominator that they had decreased the vowel inventory of the tongue. Classical Latin bore 13 possible syllable nuclei composed of either 10 simple vowels and 3 diphthongs; but in the course of only several hundred years, those numbers had been reduced to 7 and 0 respectively. However, phonetic changes in the medieval period began to take place which would not only maintain the distinction between remaining vowels, but would enforce extending variation; this mostly occurred by the addition of applicable diphthongs to the phonological tool set.

The most prevalent locales for this diphthongization were the “short” mid vowels; that is /ɛ/ becomes slowly /iɛ/ and /ɔ/ becomes /wɔ/. The evolution of diphthongs in Romance languages is however an area of sizable contention as to when precisely such variation occurred. Indeed, even the vocalic values which demonstrate the greatest amount of diphthongization [ɛ,ɔ] find not solely different realizations, but some fail to diphthongize at all. Spanish and Italian realize piedra and pietra for the Latin petra, while Catalan and Portuguese demonstrate undiphthongized forms in pedra. The historical issue this brings shall be discussed later, after a brief overview of diphthongization as realized in most languages.

Mobile Diphthongs

The Italian grammarian Benedetto Buommattei was the first to coin the term “mobile diphthong” as opposed to “firm diphthongs.” This terminology has gained minimal penetration outside of the specific study of the Italian language, but it does represent changes which occur in most of the Romosphere. Buommattei noted that some diphthongs in Italian never vary in quality regardless of the stress of the word; thus the diphthong in piani has the same phonetic value as pianissimi and thus is called firm. This stands in contrast with the mobile diphthongs in movimento versus muovo in which a change in the location in stress results in the diphthongization of <ɔ> into <uɔ>, the vocalic change we are tracing.

The distribution of mobile and firm diphthongs is naturally regular in terms of etymology; firm diphthongs result from the vocalization of Latin consonants: piano < planu; chiave < clave etc. Mobile diphthongs contrarily are the fruits of the evolution of Latin's short mid vowels /ɛ,ɔ/ which are only diphthongized in stressed positions ['muɔvo] < ['mɔweo].

Italian, although the area of the first explicit observation, is obviously not the sole exhibitor of “mobile diphthongs.” Often called “stem-changing verbs” in pedagogy, they are ubiquitous either as a result of phonemic change or euphony. Spanish is a particularly liberal language in terms of its employment of diphthongization. Italian, over the ages, has maintained the 7 vowel system of PR, while Spanish has in fact, lost /ɛ,ɔ/ to their constant diphthongization and, in unstressed positions, their merger with /e, o/. While mobile diphthongs in Italian and other languages exist in open syllables, Spanish wields them in closed syllables; however diphthongization does not precede palatized consonants, thus Latin /pɛctu/ yields Spanish /petʃo/ and not /pietʃo/.

However most confounding is the actual distribution of such diphthongization of /ɛ, ɔ/ as illustrated below: (Examples of diphthongization in bold.)

Lat.	[kɛlu]	[mɛle]	[dɛke]	[ɔwu]	[mɔla]	[mɔrte]
Eng.	“heaven”	“honey”	“ten”	“egg”	“grindstone”	“death”
Ital.	[tʃɛlo]1	[mjɛle]	[djɛtʃi]	[uɔvo]	[muɔla]	[mɔrte]
OSFr.	[tsɛl]	[mɛl]	[dɛts]	[wɔw]2	[mɔla]	[mɔrt]
ONFr.	[tsjɛl]	[mjɛl]	[dis]1	[wɛf]	[muɛlə]	[mɔrt]
Cat. (Arc)	[tsel]	[mel]	[dɛu]	[ɔw]	[mɔlə]	[mɔrt]
Spn. (Arc)	[tsjɛlo]	[mjɛl]	[djɛts]	[wɛβo]	[muɛla]	[muɛrte]
Prt. (Arc)	[tsɛo]	[mɛl]	[dɛts]	[ɔvo]	[mɔ]	[mɔrte]

(Hall 38, 47) Note archaism in consonant structure.

¹Lack of diphthongization may be attributed to assimilation with consonants.

²Abnormal diphthongization may be due to initial location.

Questions of Origin

The chart above brings difficult questions into the open. All linguists would clearly classify, for example, Spanish as an Ibero-Romantic language along with Portuguese and Catalan, springing from a more recent common ancestor than Italian and French. However in terms of diphthongization alone, Spanish is clearly phonetically more similar to northern dialects of Old French or Italian. This signifies that the Proto-Ibero-Romance would have to have a mechanism by which some of its descendents could develop parallel phonetic change with outside tongues.

Perhaps the most likely possibility is that diphthongization in a limited degree evolved as a euphonic rule in some stressed positions, and some languages began to generalize its occurrence to other positions: Romanian to long vowels, Spanish to closed syllables and similar changes in other stressed verbs etc. This mechanism, despite whatever kind of fancifulness it may evoke, is defensible in that there have been occurrences in living memory of similar evolution: dialectal Spanish, in example, does demonstrate a concurrent example of the overgeneralization of diphthongs in the Chicano speech of the southwestern United States in which the diphthongized stems are generalized to all verb forms (puedemos > podemos etc.) (Acevedo 352).

Final Questions

The inquiry of the Romance diphthongs brings us numerous questions of the nature of language and speakers’ perceptions thereof. Historically speaking, if indeed diphthongs evolved in individual languages of their own accord, is it that the phonological inventory of Latin was innately predisposed to such changes? Perhaps equally it could be posited the extent to which feedback amongst the Latin sprachbund have encouraged such parallel evolution.

On a more piercing mental level, does the unconscious parole of individual speakers reflect original non-diphthongized forms? That is, do speakers of Spanish unwittingly perceive puede as an irregular verb form or an implicit realization of the underlying form of pode? Similar questions have been asked of English, perhaps most notably in Chomsky and Hale’s Sound Patterns of English, which made attempts to explicate predictable bizarrities of English spellings as reflections of archaic, underlying forms.

Whether or not such an analysis be true, it would nonetheless be interesting to observe if children learners do in fact make generalizations off of diphthongized forms. In the form of a “wug” test, it could be trailed if any young speakers of Italian would render the 3^rd singular form of a hypothetical [fɔgare] as [fuɔga] etc.; similar studies have been initiated in the context of Spanish (Bybee 2007) in which participants are given a diphthongizted verb form and some indeed reply with a non-diphthongized infinitive when cued. Clearly however, there may as well be a conceptual shelflife; few Spanish speakers may actually perceive fuego and focal as morphologically related although in an etymological sense, they certainly are.

Thus we are left with more questions than at our inception in this discourse, which in a sense in desirable. The link between Proto-Romance vowels and its offsprings has been traced meticulously, yet not only specific historical details remain, but cognitive implications and the intuition patterns of individual speakers. The specific phonetic transformations that have occurred over the past 2000 years are, relatively speaking, well-documented yet further research begs further questions as to the causality of changes, the development of parallel forms and the underlying mental perceptions of native speakers’ speech.

Selected Sources

Allen, W.S. (1978) Vox Latina. Cambridge University Press.

Azevedo, M.M. (2009) Introdución a la lingüística española. Upper Saddle River, New Jersey: Prentice Hall.

Buommattei, Benedetto. (1643). Grammatica della lingua toscana.

Bybee, Joan. (2007). Frequency and Use and the Organization of Language. Oxford University Press.

Chomsky, Noam and Morris Hale. (1968). The Sound Patterns of English. New York: Harper and Row Publishers, Inc.

Elcock, W.D. (1960). The Romance Languages. London: Faber & Faber Limited.

Fradejas, J.M. (2010). Las lenguas románticas. Madrid: ARCO/LIBROS, S.L.

Hall, R.A. (1976). Proto-Romance Phonology. New York: American Elsevier Publishing Company, Inc.

Lathrop, T.A. (1989). Curso de gramática histórica española. Barcelona: Editorial Ariel, S.A.

Machonis, P.A. (1990). Histoire de la langue, du latin a l’ancien français. Lanham, Maryland: University Press of America, Inc.

Mateus, M.H. and E. d’Andrade. (2000). The Phonology of Portuguese. Oxford University Press.

Miret, F.S. (1998). La diptongación en las lenguas románticas. München: LINCOM EUROPA.