Synizesis, Diaeresis, and Consonantal I

Several of Latin verse's most puzzling scansions arise before the meter is applied: two written vowels may occupy one syllable, or an i that looks vocalic may function as consonantal /j/. Synizesis joins adjacent vowels in one syllable; vocalic diaeresis keeps them in separate syllables; consonantal i is an onset rather than a nucleus. These are phonological analyses, not emergency marks that a reader may invent whenever a line does not fit.

This page uses diaeresis in the vocalic sense “separate pronunciation.” Do not confuse it with a metrical diaeresis, a word end at a foot boundary, treated on Diaeresis and the Bucolic Diaeresis. Macrons below are pedagogical. Attributed verses retain their source spelling.

Begin with nuclei, not letters

The decisive question is how many syllabic nuclei a sequence contains. Ordinary ae, au, and oe are diphthongs and normally form one heavy nucleus. Hiatus gives each vowel its own nucleus. Synizesis instead draws vowels that would otherwise be separate into one syllable (literary/poetic). A consonantal i never forms a nucleus in the first place.

English spelling habits are dangerous here. English readers often count vowel letters mechanically, but Latin syllabification follows sounds. The macron in poēta warns that oē is not the diphthong oe: the division is po-ē-ta. By contrast, no number of accents or stressed beats can split the ordinary diphthong in caelum.

Synizesis: two vowels in one syllable

In synizesis, adjacent vowels that are not an ordinary lexical diphthong are realized together. The traditional term describes the metrical result more securely than an exact phonetic recipe: different sequences may have involved a glide, contraction, or unusually close transition. Do not automatically write a macron over either vowel. Vowel length and the weight of the resulting syllable are separate questions.

Vergil has dehinc as one syllable in Eurum ad se Zephyrumque vocat, dehinc talia fatur (Aeneid 1.131). Because h does not block vowel contact strongly in Classical scansion, de-hinc can contract to a single heavy realization (literary/poetic). At Aeneid 6.412, line-final alveo has its final e-o compressed: deturbat laxatque foros; simul accipit alveo. The contracted final syllable fills verse-final anceps. That position alone would not establish vowel quantity; morphology independently identifies the ablative ending -ō as long.

The genitive Oilei at the end of Vergil's unius ob noxam et furias Aiacis Oilei? (Aeneid 1.41) likewise compresses final e-i. Such evidence licenses the attested form in its poem; it does not make every ei a diphthong. The productive default remains to identify the dictionary form and normal division first, then test whether author, genre, and meter support synizesis.

Vocalic diaeresis and deliberate separation

The opposite operation is often called diaeresis, dieresis, or dialysis: vowels are heard separately rather than combined. Sometimes that separation is simply lexical, as in poēta. Sometimes an archaic or Greek-shaped ending supplies an extra syllable. The old poetic genitive aulāī divides au-lā-ī, whereas regular aulae divides au-lae. The form is real morphology, not a license to split ordinary ae wherever convenient (archaic/literary).

Modern editors may print a diaeresis mark, as in aëre, to warn readers that adjacent letters do not make a diphthong. That sign is editorial help, not normal ancient spelling. An unmarked edition therefore requires lexical knowledge. Compare the separate problem of hiatus between words, where a final vowel escapes expected elision; see Elision, Prodelision, and Hiatus.

Consonantal i

Initial i before a vowel commonly represents /j/: Iūlius, Iūnō, iam. This guide writes i, never j. Intervocalic i can likewise have consonantal value, often with strengthening, and poetry sometimes extends that treatment after a consonant. In abiete at Aeneid 2.16, i is consonantal: approximately ab-ye-te. That makes ab- heavy through closure. The spelling does not authorize four vowel nuclei.

The first example contains consonantal I- in Iūlia, i- in iam, and i- in iānuam. In prose these are ordinary pronunciations, not poetic license. The second uses teaching-form abietem in normal prose; whether a poet gives its medial i vocalic or consonantal realization must be established from actual metrical evidence.

A line worked syllable by syllable

Vergil's second line provides the classic diagnosis:

Italiam fato profugus Laviniaque venit (Aeneid 1.2; source spelling retained).

With analytical quantities and connected division:

Ī-ta-li | am-fā | tō-pro-fu | gus-Lā | vī-nya-que | vē-nit
– ⏑ ⏑ | – – | – ⏑ ⏑ | – – | – ⏑ ⏑ | – x

The opening I- of Italiam is lexically short; Vergil exceptionally lets it occupy a long position at the beginning of the line (poetic lengthening). That license must be identified rather than disguised as ordinary vowel quantity. Later in the line, the tempting division Lā-vī-ni-a-que supplies four syllables after Lā- and makes the fifth foot impossible. The corrected Lā-vīn-ya-que is not arbitrary deletion: i has become /j/, so it is an onset and no longer a syllabic nucleus.

Two consonant letters are not one rule

After finding nuclei, count sounds, not letters. Qu represents one labialized consonant /kʷ/; ch, ph, and th in Greek loans represent single aspirated stops in careful reconstructed Classical pronunciation. A stop plus liquid can form one onset, so patris may divide pa-tris. Conversely, one letter x represents /ks/ and normally makes position.

Common Mistakes

1. Counting every written vowel as a syllable

2. Treating every ei as a normal diphthong

3. Adding vowel length because a syllable is heavy

4. Dividing every pair of consonant letters

Key Takeaways

Synizesis joins vowel material into one syllable; vocalic diaeresis or hiatus keeps nuclei separate; consonantal i supplies an onset rather than a nucleus. None is a visual trick. Establish the word's morphology and phonology, account for every syllable, and only then use the metrical scheme as confirmation.