This is the final part of a four-part series. It offers Tegos Skrībbātous’ suggested prosodic systems for ancient Celtic languages, based on the historical evidence and theories.
See Part 1 for an introduction to what prosody is and why it matters.
See Part 2 for an examination of the attested evidence of Celtic prosody.
See Part 3 for analysis of the evidence and theory about ancient Celtic prosody.
Note: Unless stated otherwise, all Proto-Celtic roots can be sourced in Matasović (2009).1
Finally, we have reached the end of the series. Here, we will take the research and discussion from the previous posts and use it to establish practical, evidence-based solutions to the problem of old Celtic accentuation.
I say “solutions,” plural, because as noted before, prosody can sometimes change, either through natural evolution of the language or as a result of prolonged contact with foreign speaking communities. Additionally, changes and developments may not impact all regions, or may not impact them all evenly.
It must also be noted that there is a degree of uncertainty to these historical reconstructions, and it’s possible that there may have been more complexities or variants that simply aren’t evident in the historical record. These are simply the best solutions the Tegos has thus far devised based on this study of the topic. This outcome is subject to change upon any new discoveries in the field, if needed.
Now without further ado, the following are Tegos Skrībbātous’ proposals for old Celtic accentuation:
Díkantlom Kentukéltīkom (‘First’ or ‘Proto-Celtic Accent’)
For Proto-Celtic’s earliest accent, we will assume it retained some form of the Proto-Indo-European mobile pitch. This means certain syllables are accented with a higher pitch than others, but stress is otherwise applied evenly.
The pitch can be rising, falling, or both. Which syllable the pitch is placed on is irregular and variable, so it’s necessary to compare Celtic words to cognates in Proto-Germanic, Ancient Greek, and Sanskrit, to determine its type and placement. As you can see from the headline, diacritics (accent marks) are used to show which syllables have the pitch accent.
This period is *kentu- ‘first’, because it represents Celtic at its earliest phase after breaking away from the mother tongue. Since this is the earliest phase of Celtic, we will assume that it retains all of the most archaic features, including the *ϕ/f phoneme and non-initial s-clusters (e.g. *bēsman > *bēmman).
This phase of the language would be most suited towards quantitative and syllable-counting poetic styles, like those known from Ancient Greek and Sanskrit traditions. It may not have developed or widely adopted the practice of alliteration yet.
Bēmman Kitukeltīkom (‘Common Celtic Stress Accent’)
This accent is regarded as *kitu- ‘common’ (cf. W cyd)2 Celtic, because probably every daughter language had this accent type for some time, at least in their early stages (although in the Goidelic branch it has survived in varying forms to the present day). It is described as bēmman lit. ‘striking’ (cf. Ir béim), because it represents a stress accent that was applied to the beginning of each word and which replaced the PIE mobile pitch. It is postulated that this accent was acquired after contact with Vasconic (Proto-Basque) or other non-Indo-European speakers.
This fixed, initial stress can be termed towissubēmman (cf. Ir tús, W tywys ‘leading’) to contrast it with other stress accent systems that do not fall on the start of each word. While not all of its precise details are provable, we can consider the commonalities between Old Irish and Proto-Germanic initial stress.3 Following that existing schema would seem both practical and sufficient.
That is, a primary stress falls on the first syllable of all adjectives and nouns. A secondary stress falls on the first syllable of the subsequent member of a compound (e.g., *senabenā /ˈse.na.ˌbe.naː/).
For verbs, the first syllable of the verbal root takes the stress, but single prefixes do not (e.g. *to-reteti ‘s/he approaches’ /to.ˈre.tet.i/, cf. OIr. do·reith). However, the stress cannot fall later than the second member of the compound. So if more than one prefix is added to a verb, only the first prefix skips the stress (e.g., *(f)are-kom-wedeti ‘s/he prevents’ /ar.e.ˈkom.wed.et.i/, cf. OIr. ar·coat).
Unstressed words and clitics are known as dīaltā (*dī + *(f)alto- ‘joint’ > OIr. dealt).
Because the Vasconic speakers are thought to have been unable to pronounce the *ϕ/f phoneme, it is lost in this phase. Furthermore, we can assume that other known Proto-Celtic sound changes occurred here, such as the assimilation of non-initial s-clusters (e.g. *bēsman > *bēmman).
Speakers may optionally choose to shorten vowels that are in unstressed positions as a stabilizing strategy and as occurred in the development of Goidelc (see Part 2 for discussion).
Under this prosodic system, we would expect poetry to commonly feature alliteration, or owxsman (cf. OIr úaimm from PIE *pewǵ-+s-mn̥ > *(f)owx-s-man4). While the quantitative and syllable-counting meters from the Kentukéltīkom period could still be used, bēmman is more naturally suited for accentual or stress-counted metrics. Inspiration for poetic methods could be drawn especially from Old Irish and Germanic sources, since they shared this type of accentuation.
Wokʷennatus (‘Penultimate Stress’)
Penultimate stress (stress that falls on the second-to-last syllable) is best attested in Brittonic, where it likely became fully established sometime after the 1st or 2nd Centuries CE. It is also ascertained in Gaulish, based on the evolution of Gaulish toponyms. Its appearance in Gaul presumably more-or-less predated Britain, but the date and origin ultimately remain unknown.5
Based on Schrijver’s research as well as the reasoning put forth in Part 3 of this series, Tegos Skrībbātous is working under the assumption that all the Celtic languages originally had towissubēmman (fixed initial stress), but that wokʷennatus (vn. of *u(f)o + *kʷenn-a- > W. gobennu, cf. OIr. díchennaid) later appeared somewhere on the continent and eventually diffused. It likely became the predominant or sole stress among Celtic speakers in Northern Italy, all throughout Gaul, eventually spreading North to all Brittonic speakers, and there is also reason to believe (as discussed in Part 2) that it had made incursions into Iberia. Only the Goidelic branch (and possibly obscure branches such as Pictish) were unaffected.
As discussed in Part 3, wokʷennatus may have evolved out of the secondary stress found in Bēmman Kitukeltīkom, or it may have grown out of penultimate long vowels attracting stress away from initial syllables. It’s also possible that Classical Latin’s movable stress (which often fell on the penult) could have had an influence.
The application of this prosodic system is very simple and regular. Stress falls on the penult (second-to-last syllable) of each word. The only exceptions would be dīaltā, same as mentioned in the previous section. Wokʷennatus is not a movable stress as seen in Classical Latin, and suffixes are included in a word’s syllable count.
Modern Welsh does have some words where the stress falls on either the final or a pre-penultimate syllable. The former situation is a remnant of a phase where apocopation made penultimate stress oxytonic (final syllable), before it shifted back to penultimate. The latter is almost exclusively foreign loanwords. Thus, neither situation points to a real exception to the wokʷennatus rule as far as our usage is concerned, although speakers may optionally stress foreign loanwords according to the prosodic rules of their origin language.
Due to the stress falling closer to the ends of the words, this could perhaps be a phase where rhyming became more significant to poetry. Inspiration for poetic forms could especially be drawn from Welsh, which shares this accent type.
And as with the towissubēmman, speakers could optionally shorten unstressed vowels and/or lengthen stressed ones to decrease tension or instability in the prosodic system.
Labrātus Trebākos (‘Metropolitan Speech’)
This prosodic system refers to the phenomenon observed by Falc’hun, where certain Gaulish and British toponyms evolved as if they had been stressed on the antepenult (third-to-last syllable). This largely developed after wokʷennatus, in major urban centers and coastal areas that would have had greater traffic and trade with foreign speakers, thus making it ‘metropolitan speech’. Its onset in Britain could have been anywhere from the 2nd to 5th Centuries CE. It possibly appeared at an earlier date in Gaul than it did in Britain.
In Part 3, I discussed how Falc’hun’s observation may have been a red herring, and could instead represent a point in time after final syllables had been lost (apocopation), and thus would still be an example of wokʷennatus. However, as that is my own speculation, we cannot dismiss the possibility that an antepenultimate stress did exist in metropolitan areas. Antepenultimate stress never survived in the Brittonic branch, as one would expect if influential and metropolitan areas, with their higher populations, did indeed feature this stress. However, we will continue to work with the assumption that it may have existed as a dialectal variant at some point.
Thus, labrātus trebākos can either be apocopated (dīkʷennatos, cf. OIr. díchennaid) speech spoken with wokʷennatū. Or, it can retain the final syllables of words while accenting words with arewokʷennatū (antepenultimate stress). The latter system would be identical to the wokʷennatus discussed above, except that the stress falls one syllable sooner.
This series, long though it may have been, is just the start of using prosody in old Celtic languages. From here, speakers should now be able to determine which prosodic system is suitable to them, and begin to make decisions of how to write poetry and song. Hopefully, this will form a foundation to help support the creation of new schools and traditions of Celtic pagan art and devotion.
- Matasović, R. (2009). Etymological dictionary of Proto-Celtic. Leiden: Brill. (back)
- Cyd. (2006). GPC Online. (back)
- Salmons, J. (2003). Accentual similarities between Germanic and Celtic. In Accentual change and language contact: Comparative survey and a case study of early Northern Europe (p. 160). London: Routledge. (back)
- Mcbain, A. (1911). ‘Fuaigh’. In An etymological dictionary of the Gaelic language (p. 182). Stirling: Eneas Mackay. (back)
- Schrijver, P. (1995). II The British accent. In Studies in British Celtic historical phonology (pp. 16 – 22). Amsterdam: Rodopi. (back)