Monday March 20, 2006

More Latin Orthography

A few months ago, I wrote about the Claudian letters, which were proposed by Emperor Claudius to augment the Latin alphabet to handle some sounds it lacked.  More recently, I mentioned that I read W. Sidney Allen's Vox Latina: The Pronunciation of Classical Latin.  Allen refers to the Claudian letters at a couple of points; what's more, he also discusses some other little-known innovations and edge cases in the Latin writing system.  The innovations dealt primarily with two problems: how to write geminates, and how to differentiate the vowels i and u from the semivowels written the same way.

In classical Latin, geminate (or "double") vowels and consonants were both phonemic.  It's not uncommon in modern Latin references, such as the textbook I used in high school, for the distinction to be made explicitly by writing double consonants twice and long vowels with a macron (e.g. ō), but that wasn't always the case.  Regarding double consonants, Allen writes:

Before considering the individual sounds in detail, it is important to note that wherever a double consonant is written in Latin it stands for a correspondingly lengthened sound...  [I]t is necessary to observe this in pronunciation, since otherwise no distinction will be made between such pairs as ager and agger, anus and annus.  English speakers need to pay special attention to this point, since double consonants are so pronounced in English only where they belong to separate elements of a compound word—as in rat-tail, hop-pole, bus-service, unnamed, etc.; otherwise the written double consonants of English (e.g. as in bitter, happy, running) have the function only of indicating that the preceding vowel is short.  The English compounds in fact provide a useful model for the correct pronunciation of the Latin double (or 'long') consonants.

In early systems of Latin spelling, double consonants were written single; the double writing does not appear in inscriptions until the beginning of the second century B.C.  Ennius is said to have introduced the new spelling..., but in an inscription of 117 B.C. the old spelling is still more common that the new.  [Footnote: Another device, mentioned by the grammarians and occasionally found in Augustan inscriptions, is to place the sign 'sicilicum' over the letter to indicate doubling (in the manner of the Arabic 'shadda')—thus, for example, os̉a = ossa.]  The single spelling in such cases does not of course indicate single pronunciation, any more than the normal single writing of long vowels indicates a short pronunciation. (p. 11)

After this early experimentation, the writing of Latin geminate consonants as double letters became very well established.  Double vowels were another story: no method of writing them ever really caught on, although Allen describes a couple of attempts at creating one:

The standard Latin orthography does not distinguish between short vowels and long.  This inadequacy was not unnoticed in ancient times, and various attempts were made to render the writing more representatitve of speech.  The first such device was to write long vowels (like long consonants) double.  The institution of this as standard practice is attributed to Accius, who had presumably adopted it from Oscan, where it is common...[however,] the practice does not long survive the death of Accius.

...Nor does ii occur for long ī, but we know that in this case Accius recommended the writing of this time the original diphthong ei had come to be identical in sound with long ī...  This spelling continued into imperial times; but from the time of Sulla there also appears for the long ī the 'I longa', rising above the line of other letters, e.g. felIcI (later, however, the use of this symbol became much extended).

At the end of the republic a new device makes its appearance—the so-called 'apex' placed above the vowel symbol; this, however, does not appear on the vowel i until the second century A.D.  The shape of the symbol varies, but a mark like the acute accent (ˊ) is characteristic of the empire, and ʾ or ʼ of the republic.  (pp. 64-5)

[Note:  On these devices see R. P. Oliver, 'Apex and Sicilicus', Amer. J. of Philology 87 (1966), pp. 129 ff., where it is suggested that both the apex and the sicilicus...are simply variant forms of a 'geminationis nota', and that the 'I longa' derives from a short I with the form of this mark superscript.]  (p. 121)

You can see an example of an inscription with the I longa in this image in the French version of Wikipedia.  It strikes me as an elegant and iconic solution to the problem of writing long i—to write a long I you simply write I a little longer—but the same method obviously isn't easily extended to the other vowels, and in any case distinguishing such a slight difference in size was probably not practical in any but the most careful writing.  Even in the inscription above, for example, the A in the lower word looks a little bigger than the other letters.  As a reader, I'd hate to have to puzzle out whether it was intended to be long or short.

Another well-known innovation in the Latin alphabet was the later distinction between I and J and between U and V.  This innovation was very late, though, well after Latin no longer had any native speakers:

The distinction of writing i, u for vowels and j, v for consonants is of relatively recent origin, beginning no earlier than the fifteenth century.  Latin inscriptions have used I, V for both (though the 'I longa' was sometimes used for the i-consonant in imperial inscriptions, and Claudius tried to introduce a special sign Ⅎ for the u-consonant); the forms U and J were of cursive origin.  In the middle ages v and j tended to be used as initial variants; but the suggestion of a vowel/consonant distinction is first mentioned by Leonbattista Alberti in 1465, and first used by Antonio Nebrija in his Gramática Castellana of 1492.  The distinction was subsequently proposed by G. G. Trissino in his Epistola de le lettere nuovamente aggiunte ne la lingua italiana (1524); its definitive adoption for Latin dates from Pierre la Ramée's Scholae Grammatiae (1559)—whence the new letters are sometimes known as 'lettres Ramistes'.  For French it was taken up by such reformers as Ronsard, and was crystallized by the practice of Dutch printers, who were responsible for much printing of French books during the sixteenth and seventeenth centuries.  (pp. 37-8)

Other difficulties with the Latin writing system arose as a result of as sound changes that eventually led to the modern Romance languages.  For instance, in discussing m, Allen notes that it often wasn't pronounced fully in syllable final positions and that at least one Roman proposed a new letter to show this:

...In general [m] seems to have been reduced (like the n before a fricative internally) to a mere nasalization of the preceding vowel—in the imprecise terminology of the grammarians it is 'almost a foreign letter' (Velius Longus, K. vii, 54), or 'obscurum in extremitate dictionum sonat' (Priscian, K. ii, 29); and in early inscriptions one often finds the final m omitted, e.g. in the third-century opetaph of L. Corn. Scipio:

honc oino ploirume cosentiont...
duonoro optumo fuise uiro

(= hunc unum plurimi consentiunt...bonorum optimum fuisse uirum).  In the course of the second century official spelling established the writing of final m; but forms without m continued occasionally to be found.

...For the m in this position, when followed by an initial vowel, Verrius Flaccus is said to hav favoured writing a half-m (Λ) only (Velius Longus, K. vii, 80); Quintilian...describes it as hardly pronounced; and later grammarians refer to it as being completely lost.  (pp. 30-1)

The Latin h sound was also on its way out at this point, although it apparently hung on doggedly for a couple of centuries, leading to the same kind of mismatch between spelling and pronunciation that we have in English with words like honor and herb (depending on your dialect):

By the classical period in fact knowledge of where to pronounce an h had become a privilege of the educated classes; and attempts at correctness by other speakers were only too liable to lead to 'hypercorrect' misapplications.  The point is amusingly made by Catullus in his poem about Arrius, with his 'hinsidias' and 'Hionios'...  The situation sometimes gave rise to uncertainty even in the orthography; umerus, for example, tended to acquire an unetymological h (cf. Sanskrit aṃsas), similarly (h)umor, (h)umidus; and there was controversy about (h)arena, (h)arundo, the favoured forms being apparently harena, arundo...  So far as intervocalic h is concerned, even the grammarians recognize such forms as uemens, prendere for uehemens, prehendere (indeed prensare is general at all times).  (p. 44)

I've heard it asserted on several occasions that the Latin alphabet, when used to write the classical language, was very phonemic, and that is was only later developments that led to rather convoluted orthographies like those of English and French.  I think the quotes above from Allen show that the Latin writing system had its own set of problems.  Leaving aside that business with QU (what was that about?), the alphabet borrowed from Greek did not distinguish several of the important differences in the Latin sound system, and so Latin writers had to come up with the same sorts of innovations—digraphs, letter variants, and diacritics—that modern languages written with the Latin alphabet still resort to.

I am The Tensor, and I approve this post.
01:00 AM in Linguistics | Submit: | Links:


TrackBack URL for this entry:

Listed below are links to weblogs that reference More Latin Orthography:


One subtelty on the overproduction of H's by some Latin speakers- I remember from when we read "chommoda dicebat" in Latin that in Pompeii graffiti have been found giving such forms as "marithus" (husbhand). Surely nobody hypercorrects graffiti?

Posted by: Pete Bleackley at Mar 20, 2006 2:53:20 AM

I saw "V for Vendetta" this weekend, and it preserves something I'd assumed was an error in the graphic novel: the phrase "vi veni veniversum vivus vici". When I read the novel, I assumed that the "e" in "veniversum" had been added by someone unfamiliar with the use of "v" as a vowel -- that the word was really "vniversvm". But judging by a bit of googling that does turn up a few non-"V"-related pages, it seems that if it was an error Alan Moore wasn't responsible for it.

Posted by: KCinDC at Mar 20, 2006 8:32:12 AM

By the time of Pompeii many Greek borrowings with ch, ph, th would have been well established, but ordinary people probably pronounced them with the Latin sounds [k p t], so (especially as independent h was also silent), the cluster symbols would have been just graphic variants of the plain stops.

The clusters had also been creeping into native words, especially (as with _pulcher_) in the vicinity of a liquid. If this was an allophone, it wouldn't be hypercorrection to use it in _marithus_, but extension of the rule.

Posted by: aput at Mar 21, 2006 5:21:19 AM

Latin (and its Italic relatives) were not the only ancient IE languages in which gemination and vowel length were both distinctive. Old Norse had similar rules; I'd include a few minimal pairs, if I could find my textbook (oops!) However, the continental Scandinavian languages preserve those distinctions in modern times--

Danish: værre [worse]/være [to be] (vowel length)

Swedish: lössen [the lice]/lösen [ransom] (vowel length *and* gemination)

Norwegian: hakke [to chop]/hake [chin] (vowel length *and* gemination)

Danish no longer has geminated consonants, although a word containing a short vowel often keeps double letters in its spelling. Norwegian and Swedish, however, have developed a general rule: stressed short vowels must be followed by two or more consonants (and a geminated one counts as two).

Posted by: Ingeborg S. Nordén at Jun 9, 2006 5:15:30 PM