This is the first part of a four-part series. It offers 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.
  • See Part 4 for our suggested ancient Celtic prosodic systems, based on the historical evidence and theories.

In the times of our pre-literate pagan ancestors, the spoken word reigned supreme. Their rich culture and traditions were in many ways shared and perpetuated through the memorable and pleasant art forms of poetry and song. Modern pagans may likewise wish to use the ancient language of their ancestors to express themselves in verse, either for liturgy or just general creativity. But in order to do so, the poet must find the right way to string the words together — to make them shine in people’s memory, to roll off the tongue. The words must propel themselves forward with the beat of a natural — nay, an inexorable — rhythm. But this can only be done when the poet knows the nature of the words he or she speaks. The poet must know their prosody.

But what exactly is Proto-Celtic’s ‘prosody’?

Defining ‘Prosody’

While we often think of the pronunciation of words in terms of what vowel and consonant sounds they have, there is yet another layer to how words are pronounced. This layer involves such elements as the weight and length of each syllable, as well as its pitch and volume. The spoken word is really not unlike a form of music, with vital patterns of rhythm, tone, and dynamics, which can vary by language and dialect. This musical side of pronunciation has fittingly been called προσῳδία, prosōidía (root meaning ‘song’), in Greek1, from which we borrowed our English word ‘prosody’.

A main aspect of prosody is a language’s accenutation. This more specifically refers to how a language stresses or accents certain syllables in a word or phrase. Typically, accentuation is divided into two types:

Tone or pitch accent, where certain syllables may receive higher, lower, rising, or falling pitches in contrast to the other syllables. These syllables are otherwise not treated differently, so tonal or pitch accents tend to have a relatively even rhythm.

Stress accent, where certain syllables are punctuated or attacked with more force than others. This results in the “stressed” syllable being more-or-less longer than unstressed ones, and typically louder too. Stress accents create an unevenness between syllables, giving the language a distinct rhythm.

Pearls visualizing difference in stressed and unstressed accents.
Pearl necklaces visualizing the difference between unstressed and stressed accents.

Dividing between “stress” accents and “tone” or “pitch” is somewhat misleading, however, since stress accents often also have a pitch component. In English, for example, stressed syllables are also spoken at a higher pitch than unstressed ones are.2

Additionally, languages may have multiple accents. In another example of how stress and tonal accents are not mutually exclusive, modern Welsh features both a penultimate (second-to-last syllable) stress as well as an oxytonic (last syllable) pitch accent.3

Accent Placement

Languages have certain patterns or rules for where the accent falls in a word. Some languages may have fixed accents, where the placement is generally always the same, in contrast to mobile accents where the placement can vary based on certain factors (or even seemingly randomly). Accents also may or may not be phonemic (change the meaning or function of a word). An example in English of a mobile, phonemic accentuation is:

insight (/ˈɪn.saɪt/) vs. incite (/ɪn.ˈsaɪt/)

Although the spelling and context distinguishes them too, notice how the only spoken difference is where the accent is placed.

While this alone demonstrates how accent is important in speech, let us also consider poetry for a moment. Poetry can be written in free verse, but very often poems and songs feature rhythms and patterns to the words. In pitch accent languages, these patterns may be based off of vowel length or syllable weight (morae), as seen in dactylic hexameter, the exemplary meter of Classical epic poetry:

Arma virumque canō, Trōiae quī prīmus ab ōrīs
“I sing of arms and the man, who first from the shores of Troy...”
— Opening line of Virgil’s Aenid

(Note that, although Latin had a stress accent, it had adopted Greek verse styles.)

In stress accent languages, stressed and unstressed syllables often form patterns like the iambic or trochaic meters that are so well-known:

The curfew tolls the knell of parting day The lowing herd wind slowly o’er the lea
— Elegy Written in a Country Churchyard, by Thomas Gray
Once upon a midnight dreary, while I pondered, weak and weary
— The Raven, by Edgar Allan Poe

We have some more complex examples of stress patterns from the Celtic sphere, such as the Welsh Cynghanedd (see also) or the Old Irish deibidhe as exemplified in that most famous poem, Pangur Ban:

Messe ocus Pangur Bán cechtar nathar fria saindán bíth a menma-sam fri seilgg mu menma céin im saincheirdd

I and white Pangur practise each of us his special art his mind is set on hunting, my mind on my special craft.

The pattern here is rhyming each line’s final, stressed syllable.

Examples like these show how poetry and lyrics often could not be crafted without a knowledge of where the accent falls in the language. Along with that old chestnut, “You put the emphasis on the wrong syllable!” these illustrate how crucial accentuation is to spoken language. Even a slight change can alter meaning, violate poetic rules, or just sound intolerable to native speakers.

But in a dead, unattested language such as Proto-Celtic, we have no native speakers to listen to or consult for help on how to accent the spoken word. So in order to get a better perspective of how the accent might have worked, we will need to turn to the knowledge available to us from historical linguistics.

Let’s take a look at some examples found in Proto-Celtic’s Indo-European relatives and neighbors. To start with, we’ll look at Proto-Indo-European itself:

The Proto-Indo-European accent is generally agreed to have been a mobile pitch accent. It is reconstructed based on comparative research of daughter languages that have similar accentuations, though no language has preserved Proto-Indo-European’s original accentuation in its entirety. Examples include:4

  • Hittite
  • Vedic Sanskrit
  • Ancient Greek
  • Tocharian
  • Balto-Slavic
  • Early Proto-Germanic 2,5

However, other daughter languages are known to have featured some form of stress accent, where its position could vary (sometimes based on syllable weight):

  • Classical Latin
  • Later Koine Greek
  • Hindustani
  • Russian

While others featured fixed initial (first-syllable) stress:

  • Proto-Italic and Old Latin 2,6
  • Proto-Insular-Celtic (or earlier)2,7
  • Later Proto-Germanic 2,5
  • Czech

And others a fixed stress on the penultimate or antepenultimate:

  • Later Brittonic and Gaulish 7
  • Polish
  • Macedonian

(This is an ad hoc collection of Indo-European examples, across a very wide span of time, and not at all a comprehensive overview.)

You may have noticed already that some language branches had different phases of accentual styles. Accentuation is clearly something that can be subject to change over time. Linguists have observed that accentuation can even be sensitive to neighboring influence when there is intimate or sustained contact with another language. So speakers may adopt the accent of even unrelated languages that are nearby.8 These are both key things we must consider when determining how to accent our speech in old Celtic languages.

Now that we have established some definition and context for what ‘prosody’ is and why it matters, the Tegos’ next blog posts will dive deeper into the evidence we have for Celtic accentuation and ultimately how Celtic prosody may be harnessed to craft elegant or delightful verse.

To Part 2.

Translation of ‘prosody’

Before ending this post, I would like to give an addendum explaining how I express the concept of ‘prosody’ in Proto-Celtic. Since we do not have a reconstructible term for it, I have coined one instead: Dīkantlom ‘prosody, rhythm’.

Dīkantlom is inspired by the Welsh erddigan of the same meaning (besides the additional meanings of ‘harmony, melody, song’). Before we analyze this compound word and then calque it into Proto-Celtic, a few considerations must be made:

In later Celtic periods, the formation of complex compound words tends to be more liberal and promiscuous than it was in earlier times when affixes and roots in the language usually had more syllables. So if we take a modern Welsh compound word and make a 1:1 borrowing of it with Proto-Celtic forms, the result can prove lengthy and unwieldy. Also, when multiple affixes are compounded in the daughter languages, they oftentimes seem semantically empty or redundant. So for that reason, it’s not necessarily sensible to create exact calques from modern Celtic compound words.

If we look at erddigan, first documented in the 14th Century9, it appears to be composed of:

fare-, di-, and the root word cân ‘song’, of unknown etymology (though I assume a borrowing of some form of the Latin cantus, ‘song’). If we replace cân with a known Proto-Celtic synonym, the result would be *(f)are-di-kantlom.

It’s actually quite a nice sounding word, in spite of its length. But let’s analyze its construction more deeply:

Welsh ar-, er- (*(f)are- ‘in front of’) has long been seen as a cognate or equivalent to either English arch- or Latin per-, intensifying prefixes.10,11 While Matasović has shown this to be a misconception,12 it may have been how the prefix was understood at the time the word erddigan was formed.

The next component, *di-, means ‘of, from’.12 It often seems to imply an end result or derivative of the root it’s attached to. Since ‘prosody’ is the musical quality of speech, it could be seen as deriving from the act of singing. Thus, attaching this prefix to a word for ‘song’ should suffice for the meaning of ‘prosody’ without the need of any additional (let alone mistaken or redundant) affixes.

So we arrive at the final, aforementioned result:



  1. Retrieved on August 16, 2019 from Wiktionary, the free dictionary. (back)
  2. Sihler, AL. (2008). New comparative grammar of Greek and Latin. Oxford University Press. (back)
  3. Willis, D. (n.d.). Old and Middle Welsh. University of Cambridge. (back)
  4. Fortson, B.W. (2010). Proto-Indo-European phonlogy: Phonological rules (3.30). In Indo-European language and culture : An introduction. Malden: Wiley-Blackwell. (back)
  5. Schrijver, P. (2013). 3.2 Consonant gradation and Verner’s Law. In Language contact and the origins of the Germanic languages. London: Routledge. (back)
  6. Vine, B. (2009). PIE mobile accent in Italic: further evidence. In Roots of Europe – language, culture, and migrations (pp. 58 - 59). University of Copenhagen. (back)
  7. Schrijver, P. (1995). II The British accent. In Studies in British Celtic historical phonology (pp. 16 - 22). Amsterdam: Rodopi. (back)
  8. Campbell, L. (2013). 3.7.5 Borrowed rules. In Historical linguistics: An introduction. Edinburgh: Edinburgh University Press. (back)
  9. Erddigan. (1966). GPC Online. (back)
  10. Ar-. (2006). GPC Online. (back)
  11. Per-. (n.d.). Retrieved on August 16, 2019 from Wiktionary, the free dictionary. (back)
  12. Matasović, R. (2009). Etymological dictionary of Proto-Celtic. Leiden: Brill. (back)

Next Post