But people don't just do random stuff in their morphology, even on slang.gr. There is a reason why the coiner of the word chose to use that odd stem. To explain it, I'm going to write about the verb–noun and verb–verb compounds of Modern Greek.
Language advisory still applies.
I've written about verb–verb compounds elsewhere with Brian Joseph, in what may be my last linguistics paper; it has a response by Paul Kiparsky which makes historical sense of why this linguistic innovation happened, and (I've just googled) seems to have provoked another paper by Angela Ralli. The verb–verb compounds are dvandvas; the V1V2 compound can be paraphrased as "V1 and V2".
The phenomenon needs to be explained, because verb dvandvas, such as μπαινοβγαίνω "enter–exit = go in and out" are very unusual by European standards; in fact, Modern Greek's the only place they turn up. English arguably has them as well, as in freeze–dry; then again, typologically English is a lot closer to Chinese anyway...
I've written about the history of the construction there, and Kiparsky has written more, but for this post, we're going to concentrate on two features of the dvandvas: the tense they use, and the form of the first stem.
Verb dvandvas are often antonyms, and they refer to the alternation of V1 and V2: μπαινοβγαίνω "go in and go out", αναβοσβήνω "light up and extinguish = flicker". Even when they don't, they still refer to repetitive action: στριφογυρίζω "twist and turn". That means they always turn up in imperfective aspect—corresponding to the present stem: the V1 will never appear in the aorist stem, which has perfective aspect. So αναβοσβήνω "I am lighting up and extinguishing", not *αναψοσβήνω "I have lit up and am existinguishing" or αναψοέσβησα "I have lit up and exstinguished". (George Chortatzis uses εμπαινοβγήκα in the aorist for "I went in and out"; that's not allowed in the contemporary standard, and its V1 stem is still in the present tense form.) The one exception I could find was λυσοδένω "untie–and–tie", rather than λυνοδένω; and that was clearly influenced by the similar αλυσοδένω "bind in chains".
This is in contrast with Verb–Noun compounds, which are a bit older in Greek, and are a particular feature of Greek slang. In those compounds, the verb is typically in the perfective (aorist), ever since Demosthenes' φυγόδικος "flee-trial = fugitive". Thus, σπασαρχίδης "ball-buster", κλαψομούρης "cry-face". It is in the perfective even when the verb clearly refers to habitual activity, as in χασοδίκης "lawyer who keeps losing trials", αλλαξοκωλιά "ongoing exchange of arses = being a switch sexually; intimate friendship". The present stem can be used, but is rarer: σπαζοκεφαλιά "bust-head-ness = brain-teaser".
However, both VN and V1V2 compounds are unusual to being with, because they start with a verb. Compounding in Greek, as in most languages, normally starts with a noun. As a result, there is a tendency in V1V2 and VN compounds to make those V1 verbs look more like nouns. The first way of doing that is stripping tense suffixes from the verb, if you can get away with it. What's left behind will look like a noun—and often enough actually is a noun, if noun and verb share the same root.
You can get away with stripping the tense suffix if the tense suffix has a vowel in it. If the tense is formed by just appending a consonant, or by changing the root vowel, you leave it alone. So:
- Present μπαιν-, Aorist μπηκ-: nothing to strip, so μπαινοβγαίνω "go in and out"
- Present παιζ- (*παιγ-j-), Aorist παιξ- (*παιγ-σ-): no vowel in suffix, so παιζογελώ "play and laugh"
- Present χαν- , Aorist χασ-: no vowel in suffix, so χασοδίκης "lose-trial, incompetent lawyer"
- Present γελ-(άω), Aorist γελ-ασ-: no vowel in present suffix (we're discounting the γαμαοδέρνουλας form), so γελοκλαίω "laugh and cry"
- Present σκοτ-ων- (*σκοτ-ο-), Aorist σκοτ-ωσ-, Noun σκοτ-: σκοτοψώμης "killbread, one so hungry he hunts down and kills loaves of bread"
- Present ζυμ-ων-, Aorist ζυμ-ωσ-, Noun ζυμ-: ζυμοφουρνίζω "knead and bake"
- Present ανεβ-αιν-, Aorist ανεβ-ηκ-: ανεβοκατεβαίνω "go up and down"
Even more compelling, some verbs have been distorted to forms they never have in isolation, to look more like nouns:
- στρίβω + γυρίζω: στριφογυρίζω "twist and turn", cf. noun στροφή "turn"
- κλέβω + κότα: κλεφτοκοτάς "thieve-chicken, chicken thief", cf. noun κλέφτης "thief"
- σκάβω + κλαδεύω: σκαφοκλαδεύω "dig and prune", cf. noun σκαφή "digging"
(However, στρίβω does have a variant στρέφω, and κλεφτοκοτάς does alternate with κοτοκλέφτης, so these examples may not be that compelling.)
Equipped with all that, we come back to γαμαοδέρνουλας. The coinage is V+V+noun suffix: it first coins a verb dvandva, γαμαοδέρνω "fuck and bash", summarising the phrase γαμάω και δέρνω, and then it attaches the augmentative ουλας, "someone big who fucks and bashes". The coinage is *not* VN, like αλλαξοκωλιά or χασοδίκης. So the V1 could not appear in its aorist form, as *γαμησοδέρνουλας. It couldn't appear like that anyway, since the aorist suffix -ησ- contains a vowel, and so is too obtrusively verblike.
Which means the dvandva should be the imperfective γαμοδέρνω, just like other dvandvas whose V1 is an -αω verb: γελοκλαίω, γεννοσπέρνω, γεννοβολώ, τσιμποφιλώ. The V1 should look like a noun.
The thing is, it does look like a noun, because the related noun already exists: γάμος "wedding". (Γαμέω "marry" was just the denominative verb derived from γάμος.) But unusually, the noun and verb have diverged appreciably in meaning, so the verb looking like the noun can be confusing. We could conclude that γαμαοδέρνω was coined instead, to avoid a confusing ambiguity with "wedding-bashing", especially when NV compounds are much more prevalent than VV compounds.
We could say that; after all the compound was coined by a single person, and they may well have consciously thought: "γαμοδέρνω. Hm. Sounds like weddings. Better drag it out to γαμαοδέρνω just to make sure I'm understood." If I say αυτός γαμοδέρνει "he XXX", it does indeed sound like it shouldn't be referring to coitus.
But ambiguity is not as effective a driver of language phenomena as we might expect. There is a much more real ambiguity in γαμοτράγουδα between "weddings songs" and its actual meaning, "songs about fucking". But that hasn't prevented γαμο- being understood as the verb in the compound: noone has felt the need to use *γαμαοτράγουδα.
There is a second reason why you would say γαμαοδέρνω: γαμαο- /ɣama-o-/ sounds identical to γαμάω /ɣama-o/, the inflected 1st singular form in γαμάω και δέρνω. We wouldn't expect the first half of a compound to be inflected, of course; in fact Hatzidakis expended some ink in refuting spellings like μπαινωβγαίνω, and arguing that this really was just a compound with a linking /-o-/ —the second person is μπαινοβγαίνεις and not *μπαινεισβγαίνεις, so the vowel isn't doing any inflecting.
We wouldn't want to go back to the bad old days of 1840s spelling with *γαμαωδέρνω; but the colloquial flavour of γαμάω would have encouraged the coiner to echo it in γαμαοδέρνω. The compound isn't actually quoting the original phrase literally: it isn't «γαμάω και δέρνω»-ουλας /ɣamaokeðernoulas/ "Lord Master 'I fuck and bash'", and the connecting "and" and the second verb inflection have been dropped. But the phonology of the verb in the phrase would have influenced the coiner's phonology in the compound, especially if they hesitated over the ambiguity of γαμοδέρνω.
Those of you that have made it that far may well be rolling your eyes by now. "Surely people don't go through this amount of geometry whenever they make up some crappy little joke compound." They don't do it consciously and deliberatively, certainly; the whole thing may have been a second of hesitation, and driven by intuition rather than ratiocination. But the intuition itself is based on rules and structures: the understanding of how Modern Greek compound works is not divine revelation, it's picked up from patterns, and it's followed quite diligently.
In fact, that's the real reason why γαμαοδέρνουλας sounds odd. Not because the analysis of /ɣama/ as a stem is impossible: people wouldn't be saying γαμάω if it was. But because stems ending in /a/ haven't made it to compounds yet: people's linguistic intuition is based on precedent, and without precedent for stems ending in /a/, they've never seen an /ao/ cluster in a compound before. So the /ao/ cluster in γαμαοδέρνουλας is jarring.
Which may well be the point of the compound. If you know the rules (intuitively), you also know when to break them, to greater effect. And γαμαοδέρνουλας is certainly effective.
It got two blog posts of morphological argumentation out of me, for one...