:::: MENU ::::

Classical Latin’s Complicated Relationship with Aitch


Classical Latin’s Complicated Relationship with Aitch

Stone blocks with “H” shape


by Todd Clary, Ph.D.

Students of the Carmenta Spoken Latin Classes will no doubt have noticed that there are many aitches written at the beginning of Latin words that they are taught not to pronounce. The practice of not pronouncing these aitches reflects an actual language change known as psilosis that occurred in the history of spoken Latin. The term psilosis derives from the Greek adjective ψιλός ‘stripped bare, bald’. Greek grammarians used ψιλή, the feminine singular of this adjective, to refer to smooth breathing, or lack of aitch, at the beginning of words. So, when they noticed that some dialects of Greek did not have an aitch, or rough breathing, at the beginning of words which in most dialects had an aitch, as in Attic Greek ἕκαστος (hekastos) ‘each man’ vs East Ionic ἔκαστος (ekastos), they called that dialect psilotic, and psilosis became the word for h-dropping. For more on psilosis you can read its Wikipedia or Facebook page:


Latin scholars commonly assert that in Classical Latin initial aitches were pronounced, but that in Late Latin they were not. In other words, psilosis in Latin is generally dated to Late Latin, roughly 3rd to 6th centuries CE. A word often cited to support this argument is Modern English ‘orchard’ its initial element from Latin hortus, which must have been borrowed early from Latin into Germanic since it occurs in Gothic ‘aurti-gards’ and Old English ‘ort-geard’, both obviously without the aitch even though these Germanic languages had lots of words with initial aitch. In Romance languages Latin words with initial aitch, although they often retain the aitch in spelling, do not retain it in pronunciation: Latin hominem > Spanish hombre /ombre/, French home /omme/. The silent aitches in Spanish hombre and French homme bring up an interesting question about Latin itself: since we cannot listen to native speakers, why do we assume that aitch was pronounced in Classical times? As the Spanish and French examples show, arguing that it must have been pronounced based on orthography alone is extremely risky.

Modern bust of Catullus on the Piazza Carducci, Sirmione

In fact, there are other, more compelling reasons to think that initial aitch was pronounced in Classical Latin. The first is the fundamental premise of a poem by the Roman poet Catullus, who was writing around 60 BCE. Catullus mocks a man for starting words with aitches even where they don’t belong:

Catullus 84

Chommoda dicebat, si quando commoda vellet
Dicere, et insidias Arrius hinsidias,
Et tum mirifice sperabat se esse locutum,
Cum quantum poterat dixerat hinsidias.
Credo, sic mater, sic liber auunculus eius,
Sic maternus auus dixerat atque auia.
Hoc misso in Syriam requierant omnibus aures:
Audibant eadem haec leniter et leviter,
Nec sibi postilla metuebant talia uerba,
Cum subito affertur nuntius horribilis:
Ionios fluctus, postquam illuc Arrius isset,
Iam non Ionios esse sed Hionios.

Arrius would say happropriate whenever he wished
To say appropriate and hambush for ambush;
And he hoped that he had spoken marvelously
When, to best of his ability he had said hambush.
I believe that so his mother, so his free uncle,
So his grandpa and grandma on his mom’s side had spoken.
But when this guy was sent to Syria all our ears got a rest.
They heard the same words smoothly and lightly,
Nor did they fear such words in the future,
When all of a sudden some horrible news came:
The Ionian Sea, after Arrius had gone there,
Is no longer Ionian, but Hionian instead.

So it seems that Arrius and his family are suffering from a terrible case of hypercorrection for aitch. In his Introduction to Historical Linguistics Lyle Campbell gives a good definition of hypercorrection:

Hypercorrection involves awareness of different varieties of speech which are attributed different social status. An attempt to change a form in a less prestigious variety to make it conform with how it would be pronounced in a more prestigious variety sometime results in overshooting the target and coming up with an erroneous outcome (113).

A great example of hypercorrection occurs in English dialects in the southern U.S. where some speakers pronounce umbrella as umbrellow. The form umbrellow is based on an analogy with pairs like fellow/fella, yellow/yella where the forms with –ow pronunciation are considered more formal. Hence, the speaker creates an analogy along the lines of fella is to fellow as umbrella is to umbrellow. Arrius in the Catullus poem appears to have hypercorrected for aitch. He has heard Latin words like, hinc ‘hence’, hirsutus ‘hairy’ pronounced with initial aitch and is erroneously putting aitches on words like insidiae ‘ambush’. According to the above definition of hypercorrection Arrius’ mistake presupposes two things: that initial aitch was pronounced in a prestige dialect in the Roman Republic (remember that Catullus was writing ca. 60 BCE); and that there were other dialects, of which Arrius was aware, that did not pronounce these same initial aitches. In other words hypercorrection presupposes both sides of the analogy leading to it, so that next to the forms hinc and hirsutus you need the forms inc and irsutus. That way Arrius can set up his analogy along the lines of fella/fellow > umbrella/umbrellow, that is inc/ hinc > insidiae/hinsidiae.

The coexistence of aitched and aitchless forms for the same word in Latin is validated in texts other than Catullus’ poem. For instance, graffiti at Pompeii attests IC for hic ‘this man’ (CIL 4.1321) and ABETO for habeto ‘let him possess’ (CIL 4.2013.10), while in other inscriptions we find forms like HOSSA for ossa ‘bones’ (CIL 6.13657) and HEIUS for eius ‘his’ (CIL 3.3917) (For more details see Weiss, 152-154). On occasion hypercorrect forms became generalized as the main form of a word, as in Latin haurio ‘I draw (water), which is actually cognate with Old Norse ausa, and Greek αὔω ‘I kindle a fire’ neither with any evidence for initial ‘h’.

Now, revisiting the claim that psilosis in Latin can be dated to Late Latin, we will need to qualify that, while psilosis had been going on for several centuries, it was not until Late Latin that the change fully diffused. In other words, sometime in the rather broad period of time between the 3rd and 6th centuries CE initial aitches were finally lost in all dialects of Latin. We are clearly looking at a language change that took centuries to diffuse fully, which brings up the question: when did dialects of Latin first become psilotic?

To answer this question we can rely on comparative evidence. Most Latin initial aitches trace back to the Proto-Indo-European (PIE) phoneme *gh palatal or non-palatal. The Latin word for swan, anser, appears in texts as early as Cato the Elder and Plautus:

ut de frumento anseres ‘just like swans from the tall grass…’ (Plautus Truculentus 252)

Latin anser ‘swan’ derives from a PIE word applying generally to water-fowl, and surfacing in English as goose, Greek as χήν /khḗn/ ‘goose’ and Sanskrit as hamsa—‘swan, goose or flamingo’. Hence, it must have had an initial aitch that dropped out altogether even before old Latin. So Cato the Elder and Plautus (ca. 200 BCE) give us evidence that psilosis in Latin was prevalent enough before 200 BCE to have eliminated initial aitch completely from at least one word. In the end, we must conclude that h-dropping aka psilosis in Latin was happening from at least the 3rd century BCE to the 3rd century CE. It took a millennium to take everybody’s breath away.


Abbreviations and References:

CIL = Corpus Inscriptionum Latinarum (multiple volumes)

Campbell, Lyle. 1999. Historical Linguistics: An Introduction 2nd edition. Cambridge, Mass.

Weiss, Michael. 2009. Outline of the Historical and Comparative Grammar of Latin. Ann Arbor / New York.


todd clary
Dr. Todd Clary, a specialist in the Latin, Greek, and Sanskrit languages, holds a Ph.D. in Classics and Indo-European Linguistics from Cornell University, where he now works as a professor in the Classics Department. He has taught “Homeric Philology”, “Latin Comparative Grammar”, “Archaic Latin and the Italic Dialects”, as well as all levels of Greek, Latin, and Sanskrit. Dr. Clary currently teaches the “Greek I” class and tutors Latin, Ancient Greek, and Sanskrit for the Carmenta Online Latin School.

Click here to see Dr. Clary’s full profile on the Carmenta Faculty Page.



Comments are closed.