A number of loan words and New Latin words derived from Ancient Greek have word initial clusters of a plosive+nasal or dissimilar phonemes like /ps/ or /pt/. I cannot avoid aspirating or inserting a schwa when trying to pronounce these syllable initially, though syllable finally or word medially I can pronounce just fine.

Were these pronounced syllable initially in Ancient Greek? If so, how am I unable to?

Were these pronounced only when word medial? If so, what phonological process lead to this?

  • 7
    These clusters remained when the words were loaned, even to languages with different writing systems, implying that they were indeed pronounced.
    – Draconis
    Commented Mar 8, 2017 at 14:43
  • Can you say "epsilon" without aspirating or inserting a schwa? Commented Mar 9, 2017 at 10:03
  • 1
    @DavidRicherby: I pronounce it as separate syllables "ep·si·lon."
    – Anonymous
    Commented Mar 9, 2017 at 13:35

2 Answers 2


There is very little doubt they were pronounced: they are still pronounced in many languages other than English where they were loaned, and crucially in modern Greek; they were also spelled with those clusters when coeval Latin borrowed them.

The fact you cannot pronounce them doesn't mean they cannot be pronounced... No offense, but everyone's pronunciation abilities tend to be limited by their native language's phonology (which sounds are found and distinguished in the language) and phonotactics (which combinations of sounds are allowable in which situations). Foreign sounds, or combinations of sounds, can be hard to produce without training.

To hear how these sounds are pronounced in practice, you can just open Google Translate and click on the speaker icon to make it pronounce words of this sort in modern Greek, Italian (this one has an obvious schwa in "xenofobia", but I can testify that's a glitch in the speech synthesizer, as in standard Italian there is no schwa there), Spanish, Swedish or German.

I have also recorded myself pronouncing some words starting with this kind of clusters in Italian: psicologo, pseudonimo, pneumatico, pneumotorace, xilofono, xenofobo, just to have them from an actual human speaker too.

  • Well, modern Greek does have its share of spelling pronunciations, doesn't it? As do other languages with loans from Classical Greek. I think evidence from modern languages is significant, but it would be even more convincing if you gave examples showing these sounds were preserved in demotic, and not just in learned forms. Commented Mar 8, 2017 at 18:02
  • 2
    My answer's main concern was, honestly, to dispel the idea that someone finding certain clusters "impossible" to pronounce generally shows anything but limitations due to their native language's phonotactics. I tried to address that by showing many modern languages do not have a problem with those clusters. The fact it's possible being clarified, I think evidence abounds that that Ancient Greek had those initial clusters: aside from Demotic, coeval Latin borrowed words with them. About your mention of πταίω > φταίω: did you notice how, interestingly, Google effectively pronounced ψ as φσ?
    – LjL
    Commented Mar 9, 2017 at 1:42
  • I hadn't listened to Google. Do you mean /p^hs/ or /fs/? If I remember correctly, phi-sigma was used in some Greek texts instead of psi. The reason I think Latin is less compelling evidence than Demotic is because Latin is another classical language that is not spoken nowadays, so what we know of its pronunciation is reconstructed. Also, loans can have spelling pronunciations. Japanese loans from English for example may adapt schwa differently depending on if it is spelled with "e" or "o." Commented Mar 9, 2017 at 2:04
  • There is never really any single piece of definite evidence when reconstructing pronunciation of dead languages. Even Demotic itself could be tainted by spelling pronunciations (to make a parallel with English, today "often" is sometimes pronounced with a /t/ by some, but that's in fact a spelling pronunciation, because the /t/ had been previously lost as I understand it, despite being there even earlier). But I just think there are enough "obvious enough" little bits of evidence here to make it a no-brainer. As to Google, I meant /fs/, but that's just what I hear anyway, I could be wrong.
    – LjL
    Commented Mar 10, 2017 at 2:57

I'll assume you're a native English speaker. Since English doesn't have these clusters, it's difficult for an English speaker to hear or produce them correctly. But it is not impossible, and there is no reason to think Ancient Greek speakers did not pronounce them.

Word initial plosive-plosive clusters are definitely possible since they occur in modern languages

It is quite possible to have word-initial clusters of plosives without an intervening epenthetic vowel. Polish is an example of a language where this occurs; the word for bird is "ptak." Here is the Forvo page: https://forvo.com/word/ptak/

To me, it sounds like the female speaker is saying "tak," and the male speaker is saying "puh-tak." But this is probably not what they are actually doing.

  • I would guess that in fact, she is using a pronunciation with less audible release of the initial /p/, while he is giving it more audible release. It's true that it's hard to hear a stop with little audible release when there is no preceding vowel. But that doesn't necessarily mean the speaker is not articulating them. Aside from being easily to hear word-medially, as you mention, it would likely be more audible when it comes at the start of a word, and the preceding word ends in a vowel. This is a common situation in actual speech; speakers don't often say words in isolation.

  • While audible release sounds like a schwa in this position to an English speaker, phonetically there is a difference between audible release, and a vowel such as schwa. Hearing a schwa is an auditory illusion, like how Japanese speakers hear an /u/ vowel between the consonants of word-initial consonant clusters in English, such as the "s" and "m" of "smile." (In fact, there are good phonological bases for both of these illusions: Japanese speakers commonly realize /u/ as zero in some positions, such as between two voiceless consonants, and English speakers may realize schwa as zero in some positions, such as between two voiceless consonants as in "suppose" or even in other contexts as in "police". But these are language-specific processes.)

I don't know enough phonetics to be able to say how easy it is to distinguish aspiration word-initially before other obstruents. In fact, though, I don't think Ancient Greek has such a contrast; there might be exceptions that I can't bring to mind, but I think before unaspirated plosives we only see π τ κ, and before aspirated plosives we only see φ θ χ.

Demotic Greek shows reflexes of the first consonant in such clusters; this indicates that there was also a pronounced consonant in the earlier forms of Greek

LjL's answer makes a good point about these clusters being present in modern Greek and in loanwords from Greek in other languages. I left a comment where I wondered if these could be spelling pronunciations, but then I realized I could probably look that up myself to some degree. Wikipedia indicates that Demotic Greek, which is the form of Greek that evolved mostly on the spoken level (so it was probably not influenced by the spelling) does have reflexes of these sounds. For example, "φταίω" is the demotic pronunciation corresponding to the learned πταίω.

So in fact, we could say that even the Greeks somehow "preferred" other types of clusters over plosive-plosive clusters, and in unlearned pronunciation the first plosive naturally evolved to an aspirate or fricative, but the fact that this happened is evidence that the initial plosive was never lost altogether.

(This just applied to plosive + plosive clusters — unaspirated stops were unchanged in demotic Greek before nasals like /n/ or fricatives like /s/).

I think this, along with the fact that the Ancient Greeks must have had some reason for writing these clusters of consonant letters in the first place, makes it pretty clear that the first plosive of word-initial clusters was pronounced in Ancient Greek.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.