 |
 |
 |
|
 |
The last point of discord as regards the Greek consonants concerns the so-called "μέσα" or "mediae"In contrast with the "aspiratae", the latin term "mediae" is a faithful translation of the greek term "μέσα", which does not prejudice the verdict on their nature. There is, therefore, no need to adhere to the greek terminology and the two terms will be used interchangeably. β, δ, γ, which are presently voiced continuants ("fricatives"):
|
ΒO: Β = [v] |
ΔO: Δ = [ð] |
ΓO: Γ = [ɣ] |
The values assigned to these letters by the Catholics are the corresponding voiced stops:
|
ΒC: Β = [b] |
ΔC: Δ = [d] |
ΓC: Γ = [g] |
Once again, the difficult sounds [ð] and (especially) [ɣ],[ɣ] is an extremely difficult sound to comprehend and reproduce for most western Europeans. This has led the Nobel-Prize winner Odysseas Elytis to make the following (almost chauvinistic) statement: "Γ. - Το πιο ελαφρύ· που η αδυναμία σου να το προφέρεις, δείχνει το βαθμό της βαρβαρότητας σου|Γ. - The lightest of all; wherein your inability to pronounce it, indicates the degree of your barbarism". Here, the term "βάρβαρος|barbarian" (whence the term "βαρβαρότητα|barbarism") should probably be construed in its ancient sense of "speaker of a foreign (i.e., non-Greek) language", not in its modern meaning of "uncultivated/uncivilised". which do not exist in the native languages of most "reconstructionists", have been brushed off and the more familiar "harsh" values [d], [g] have been adopted. The arguments used are mainly based on Latin transcriptions and focus on the labial series. Like the other disputed consonants, we will show that the catholic argumentation is based on unjustified or even wrong assumptions, that important evidence precluding the strict catholic specification have been silenced, overlooked or misrepresented, and that the solution to the "mediae riddle" is actually more compromising than both Orthodox and Catholic think.
Let us first review the rationale behind the adoption of the alternative values of the mediae. It is evident that both Orthodox and Catholic agree on the phonation (i.e., "voiced") and the place of articulation (i.e., "labial", "dental" and "velar/guttural")I will not consider the superficial argument that ΒO is not labial, but "labiodental". As discussed in the case of φ, such a distinction is meaningless. The labiodental fricatives [f] and [v] are the commonest and most stable fricative versions of the (undoubtedly labial) stops [p] and [b]. and disagree on the amount of stricture ("stops" vs "fricatives"). The place of articulation of β, δ, γ is clear from their grouping with π, τ, κ and the associated descriptions provided by the Grammarians. The other properties require some investigation.
It is strange that the Catholics have settled for the present ("modern") phonation of the mediae without feeling the need to provide substantive evidence: "There is no doubt that the sounds represented by β, δ, γ were voiced" (ALLE87, p. 29).
One of the arguments put forward by Allen is that "They do not combine in groups with voiceless sounds". However, this suggests that they were of the same type, but not what type that was (note that the aspiratae also exhibit this behaviour, e.g., *πθ→φθ, *κθ→χθ, without anyone claiming that it is confirmation of their present values). Furthermore, the argument relies on the axiom that all other sounds were "voiceless", i.e., on an a priori acceptance of the phonation of the greek sounds, in which case... we do not need to investigate.
The second (and more important) argument relies on their transliterations "by voiced sounds in other languages - e.g. Latin barbarus, draco, grammatica". Essentially, it relies on the correspondence of Greek and Latin:
This is yet another case of Roman infallibility, where the sound of Latin (as presently pronounced by the Germans) is the certain guide in any investigation. Just how precarious this conclusion is can be seen by the existence of several "colloquial" transliterations Γ→C, C→Γ, Δ→T, D→T, B→Π, P→Β, which have led Sturtevant to posit a non-perfect correspondence between the voiceless and voiced sounds of Greek and Latin. It is true that Allen speaks not only of Latin, but of "other languages" in general; yet, with the exception of a single example from "transcriptions of Greek names on Indian coins" (to argue the lack of "aspiration" in the greek mediae), no other examples are provided but from Latin; it is, therefore, not clear which "other languages" may serve to establish the voiced nature of the greek mediae. Taking Egyptian as an example, there are three major objections associated with such transliterations:
It is evident that the conclusions drawn from transcriptions into other languages are not as safe as advertised. Neither is a safe conclusion to be drawn from comparative linguistics, namely from the indo-european cognates. If we refer to the table of sound correspondences, we see that the associated sound in almost half the IE|Indo-European language families is a voiceless one (the "reconstruction" of the PIE sound as *d in the 19th century is evidently due to it being primarily based on the Greek, Latin and Sanskrit sounds, which agree on the presence of voice).
Whence comes, then, the certainty about the voiced nature of the greek mediae? I dare say that this feature owes its universal acceptance to it being the cross-section of the greek (favoured by the Orthodox) and the latin (favoured by the Catholics) traditions; otherwise, the question of phonation of the mediae has not been (or cannot be) sufficiently investigated. This does not mean that we should reject the voiced nature of the mediae, though. Consistent with my practice of asking for evidence of alternative pronunciations before considering them as candidates, I will not contest the premise that the mediae were voiced; we simply have no evidence to the contrary. The reason I played the devil's advocate was to indicate that we do not have incontrovertible evidence to the positive either.
We have seen that, in describing the ἄφωνα, the Grammarians refer to β δ, γ with the term "μέσα|middle, intermediate", whence the latin "mediae". In the specific case of Halicarnasseus, the μέσα are defined with respect to the ψιλά and the δασέα as: "μεταξὺ τούτων|between these", "κοινὰ δὲ ἀμφοῖν|common to both", "μέσον δὲ ἀμφοῖν τὸ β|β comes between the two", "μέσον δὲ καὶ ἐπίκοινον τὸ δ|δ [is] medial and common", "τὸ δὲ γ μετρίως καὶ μεταξὺ ἀμφοῖν|γ [is pronounced] moderately and between the two". All these expressions suggest that, whatever the ψιλά (π, τ, κ) and the δασέα (φ, θ, χ) are, the μέσα (β, δ, γ) come between them.
This terminology and specification has puzzled the Catholics, who have sought ways to reconcile this with their conviction that the δασέα were plosives "followed by an audible puff of breath" (STUR20, p. 174). The most naive interpretation is provided by Blass: "the name media denotes a half aspirated sound" (BLAS90, p. 108). Dawes (one of the last orthodox scholars) demonstrates the absurdity of this interpretation by means of the mathematical equations π=p, φ=p+h, β=b+½h (which follow from a literal interpretation of the Grammarians' testimony in light of the catholic translation of the word "πνεύμα") and objects that "it is difficult to conceive of a half-aspirated letter"; she also points out that (PIE) "sonant aspirates" ("bh", "dh", "gh") are supposed to have been converted to "voiceless aspirates" ("ph", "th", "kh") and rightly wonders "if this was the fate of the fully-aspirated sonant explosives, is it reasonable to suppose that a language which rejected them should for centuries have retained, if not actually introduced, half-aspirated sonant explosives?" (DAWE95, p. 40).
Sturtevant presents a fancy theory of Couch Linguistics about "how a mute may develop into a spirant" and, based on the observation that the mediae (allegedly) "have changed into the modern voiced spirants" like the aspiratae (which "have developed into the modern voiceless spirants through the intermediate stage of affricates") and unlike the tenues (which remained unchanged), "confirms" the statement of the Grammarians that "β, δ, and γ had more aspiration than π, τ, and κ" (STUR20, p. 183). Allen interprets this conclusion of Sturtevant's as a thesis that the greek mediae were "voiced aspirates, rather like the bh, dh, gh of Sanskrit", which he rejects based on lack of confirmatory evidence and on the fact that the Indians use their plain voiced stops ("b", "d", "g") to represent the greek mediae (ALLE87, p. 30).
The alternative explanation offered by Allen is that "the use of such terms as μέσα simply indicates the writers' perplexity when faced with phenomena which were not describable within their favourite binary framework", the "phenomena" being the presence of voice, the nature of which "European phonetics was slow to discover"; in other words, the term "μέσον" merely indicates something "different", which is "neither this, nor that". Interesting as this thesis may be, it does not bond well with the explicit statements "μεταξὺ τούτων|between these", "κοινὰ δὲ ἀμφοῖν|common to both", etc mentioned above, which rather clearly define the order (taking the labial series as an example):
| Π | Β | Φ |
| → δασύτης | ||
or rather (if the above-discussed feature of voice is taken into account):
|
(1) | ||
It is evident that the matter of determining the phonetic value of the mediae (as well as of the aspiratae) boils down to determining what the horizontal property ("δασύτης") stood for. Since we have two competing candidates, namely the catholic "aspiration" and the orthodox "stricture", we should investigate whether the model defined by (1) fits either of those.
The presence of any amount of "aspiration" in the greek mediae is practically precluded for the reasons provided by Dawes and Allen, as presented above. This means that the catholic phonological matrix for the (labial series of the) mutes has to be:
|
(2) | ||
which does not look like (1) above and any attempt to bring β towards the middle (i.e., more to the right) would mean that we have to acknowledge some amount of (yet, not full) aspiration in it, which is inconsistent with the above conclusion (and practically unattested in any language).
Although the verdict is not favourable for the Catholics, this does not necessarily mean that it is (fully) compliant with Orthodoxy. As is evident from ΠOC, ΦO and ΒO, modern β and φ are both (full) fricatives. In discussing the evolution of Greek phonology (always within the catholic framework), Browning points out (BROW83, p. 27) that the (alleged) fricativization of the aspiratae and the mediae produced the "triangular consonant system"
| p | ||
| f | v |
which he declares "an unusual one"; he subsequently wonders "whether b, d, g really existed in late Koine, giving a rectangular consonant system"
| p | f | |
| b | v |
which is essentially a slice of The Matrix (as is presently the case in modern-Greek phonology).Browning claims that "the question of the phonemic status of b, d, g is still not settled in regard to modern Greek". This assertion is not correct. The sounds [b], [d], [g] of modern Greek are not allophones of some other phoneme, but clearly independent phonemes, as is supported by the existence of minimal pairs like βαίνω|[ˈveno] I go/walk/μπαίνω|[ˈbeno] I enter, ένα δύο|[ˈenaˈðio] one two/έν' αντίο|[ˈenaˈdio] a farewell, γάμμα|[ˈɣama] gamma/γκάμα|[ˈgama] variety. Browning is probably the first Catholic to correctly identify one of the axes (the vertical in the "triangular" system and the horizontal in the "rectangular" one) with the feature "stricture" instead of "aspiration"; however, his depiction of the "triangular consonant system" is non-sensical, as a) [p] does not share a common feature (other than place of articulation) with both [f] and [v], i.e., does not come between them and b) the sound described by the Grammarians as "intermediate" is that of β(=[v]), not π(=[p]). The proper representation of the orthodox values of the three letters is
|
(3) | ||
which looks more like (a mirror image of) the right triangle of (2) rather than the isosceles triangle of (1) above. Strictly speaking, the orthodox model does not reflect the teaching of the Grammarians better than the catholic one does. Where is the truth then? Apparently somewhere in the middle. Prepare for the Judgment of Solomon.
One characteristic of (2) and (3), as compared with (1), is that they are lopsided; in order to restore the symmetry, a value is missing at one of the corners or the value of the media (Β) has to be (more) centered. For (2), the Catholics would expect that the missing value is the "voiced aspirate" [bʱ]; we have seen that this notation is flawed (originating in the obscurity of the 19th century) and [b̤] is more appropriate. However, it is evident that [b̤] is not to [b] as [pʰ] is to [p]; they relate to different types of phonation and their depiction along the same axis is plainly wrong (if not impossible). Furthermore, if one is to convert (2) to something like (1), it is necessary to "push" Β=[b] to the right along the "aspiration" axis; disregarding the fact that one cannot add aspiration (but only breathy voice) to a voiced sound, it has already been explained that only zero and full "aspiration" is possible, while there is no such thing as "half aspiration" (which is a prerequisite for the skewed version of (2) to resemble (1)).
This is where the orthodox model is superior to the catholic one, as it affords intermediate values along the horizontal axis; while the distinction along axes representing phonation (voice, aspiration, breathy voice) appears to be only binary (presence or absence of voice, etc), it is not too difficult to envisage different degrees of "stricture", i.e., apertures allowing different amounts of airstream to escape. One would, therefore, be tempted to conclude that the mediae involved an opening halfway between that of the aspiratae (which are identified under the orthodox model as fricatives) and the full closure of the tenues. Not so fast! A fricative is essentially the corresponding stop when the obstruction is relaxed just barely for producing a turbulent airflow; if the aspiratae were fricatives (as we concluded in the previous chapter), the mediae should involve an even smaller aperture, which is not distinguishable by the naked ear. The assumption that the horizontal axis represents different apertures is, therefore, not tenable.
Nevertheless, it seems to be too much to expect of the ancient Greeks to be able to distinguish different degrees of a property, such as stricture (after all, they possessed of no spectrograms, palatograms and other similar contraptions that are indispensable to the modern phonetician). A theory like the classification of the mediae as intermediate between the tenues and the aspiratae must have rather been based on the feeling or impression produced by the pronunciation of the respective sounds. Surprisingly, the compliance of the orthodox phonological model with the description of the Grammarians is confirmed by none other than Blass: "It certainly may be maintained that the name mediae suits the present pronunciation also, in so far as the breathing in β v is really weaker than in φ f" (BLAS90, p. 108); this assertion is certainly not true, as both [v] and [f] involve the same amount of turbulent airflow, but it is certainly proof that Blass felt that this is the case.Maybe the presence of voice gives the impression that a voiced fricative is softer and, hence, weaker than the corresponding voiceless one (cf. the terms "sharp s" and "soft s" for [s] and [z], respectively in BLAS90, p. 91). It is true that Blass also feels that the orthodox values are not the only ones that fit the definition of "mediae" (as implied by the word "also"); however, this feeling is based on the conviction that "the name media denotes a half aspirated sound", which (contrary to [v] and [f]) he most certainly never heard (and its existence is as certain as that of Santa Claus). In other words, the feeling that the term "media" had something to do with "aspiration" comes from theoretical consideration (a.k.a. Couch Linguistics), whereas its association with "friction" comes from practical experience.
Armed with the consent of the Catholics and having demonstrated the absurdity of their various attempts to reconcile the terminology of the Grammarians with their dogma that δασύτης is (both "voiced" and "voiceless") "aspiration", it would be too easy to declare the orthodox value ΒO (as well as ΔO and ΓO) the only viable candidate. Yet, there is one fact that is still overlooked and does not allow us to triumphantly declare the victory of Orthodoxy. The triangular consonant system of (3) is indeed "an unusual one"; for one, it is not symmetrical. To restore the symmetry of (3) it is necessary to add a value with [voice=yes, stricture=full], i.e. [b], resulting in Browning's (and modern Greek's) rectangular consonant system; however, the absence of a b/v phonemic distinction in ancient Greek is evident from the lack of separate graphemes in any of the ancient alphabets.If we were still in the 19th century, some might argue that Ϝ (digamma) was the missing different grapheme for [v] (cf. CARA95, p. 171: "digamma, which corresponded to the Phoenician letter waw, and had the sound of v"). A modern scholar (which Caragounis is not!) would have no reason to believe Ϝ stood for anything but [w]. What happened, then? Did the model remain flawed (=non-symmetrical) for such a long time, until the upper left corner was filled with the missing value [b] (as the result of foreign loans or the voicing of post-nasal π)? I contend that the model was a compromise between the triangular and the rectangular consonant system.
|
(4) | ||
The meaning of (4) is that the phoneme represented by Β had no specific value, but spanned the entire upper row of [voice=yes]. In other words, the distinguishing feature of Β with respect to both Π and Φ was voice, stricture being non-distinctive.Allen contemplates similar theories: "the attempt has been made to justify the Greek terminology as meaning that the voiced series was 'indifferent' to the opposition of aspirate/non-aspirate found in the voiceless series", but immediately brushes them aside claiming that "this is probably to attribute too great a sophistication to Greek phonological theory" (ALLE87, p. 30). It is reasonable that he finds such theories unconvincing; trapped in their obsession with aspiration and not being able to see beyond their dogma, the Catholics wince at the possibility of normal voiced and "voiced aspirated" stops being represented by the same umbrella symbol. If only they could overcome their prejudice and consider the opposition stop/fricative instead!
|
Βalt: Β = [v/β~b] |
Δalt: Δ = [ð~d] |
Γalt: Γ = [ɣ~g] |
Phonemes that span a phonetic range are not unheard of; for instance, Greek has a single a-phoneme (<α>) and the lack of distinction among [a], [ɐ], [ɑ], [ä] and [ɒ] has turned the trapezoidal IPA vowel chart into a triangular one (which parallel the rectangular consonant system of modern Greek and the triangular consonant system of ancient Greek, respectively). In the specific case of the labial (as well as dental and velar) series, there exists a very enlightening modern parallel. All it takes is to glance at the consonant table of modern Castilian, where one finds the following subtable for the labial series (similar subtables are provided for the dental and the velar series):
| . | Labial | |
| Stop | p | b |
| Continuant | f | |
This is exactly the table of (4) turned clockwise by 90°. I am not sure what the reason behind the placement of /b/ between (or in both) the "stop" and "continuant" rows is. There are three possibilities that I can think of:
| Spanish g ([ɣ̞]) in gol|goal |
Any of the three cases would also apply to the labial (and dental and velar) series of ancient Greek. Castilian, therefore, provides a suitable model for the state of things in ancient-Greek mutes. This is not mere speculation or wishful thinking. It derives from the following considerations:
The alternative values Βalt, Δalt, Γalt are, therefore, the logical consequence of the above line of reasoning. The repercussions of their adoption are interesting. First of all, the conclusion of the previous chapter that the distinctive characteristic ("δασύτης") of the δασέα was friction is reinforced. Despite that, the Catholics seem to have (accidentally) got one of the possible realisations right, as both [b] and [v] (as well as [d] and [ð], [g] and [ɣ]) would be perceived by the Greek ear as equivalent. In other words, the debate about which ones of the catholic (ΒC, ΔC, ΓC) and orthodox (ΒO, ΔO, ΓO) values are correct is, therefore, moot; they are both right!If you think I sound like Nasreddin Hodja, then... you are quite right! ☺︎
While we have reached the conclusion that each media should be mapped to a range of (voiced) values, it is unquestionable that each time a media was uttered, it had a specific value (e.g., either [v] or [b], but not both together). This value might have depended on phonetic environment (e.g., post-nasal or intervocalic), speaker (e.g., upper vs lower class) or dialect (e.g., Attic vs Pamphylian). In this section I intend to demonstrate that the favoured value in Attic and other dialects at least since classical times was that of a continuant (be it fricative or approximant). In other words, the native speakers would pronounce the mediae like modern Spaniards do, while anyone who pronounced them as stops would sound like a German trying to speak Spanish (unfortunately, I doubt that the poor Germans realise the difference between their pronunciation and that of the natives).
The primary evidence that points to a continuant realisation of the mediae in classical Attica is the graphic representation of the velar nasal [ŋ], which is the value with which the nasal phoneme /N/ surfaces before velar consonants (Κ, Γ, Χ, Ξ). Originally, <Ν> was used in this place (like english kink and king), but was gradually (from V BC in Attica) replaced by <Γ> (STUR20, p. 168), which ended up being the orthographic norm (cf. Άγγελος vs angel).
The choice of <Γ> for representing a continuant would be odd if the normal value of Γ were [g]. Allen describes the paradox as: "There is nothing in the nature of a velar plosive that would account for the nasalization of a preceding plosive" (ALLE87, p. 35). This assessment is most certainly wrong, as it is not a case of "nasalization of a preceding plosive"; phonologically, the preceding phoneme is a nasal, i.e., already "nasalised" (e.g., εν+κλισις→ἔγκλισις); it is rather a case of better representation of a sound for which no particular symbol existed. What would make the Greeks prefer <Γ> over <Ν> for [ŋ]? If you have never pronounced a [ɣ] in your life, it will forever remain a mystery. If you have, you find it no wonder that a voiced velar continuant was considered closer to a velar nasal than a dental nasal was.
The use of "nasal Γ" has puzzled the Catholics so much that they have come up with ridiculous theories to deny it. Beza, for one, in "De germana pronunciatione Græc. Ling.", asserts "in a decided tone" that "the grammarians have been deceived in this matter" by copyists who used lowercase letters "lengthening and widening the ν before a γ, making it resemble the latter so much as to be mistaken for it" (PICK18, pp. 265-266)! Needless to say that this fanciful theory is the product of Beza's imagination and has absolutely no basis in actual facts. Moreover, Beza would certainly feel embarrassed about his Couch Linguistics, had he been informed that the tradition of writing <Γ> for the velar nasal is attested not only in the oldest surviving manuscripts (cf. PICK18, p. 266), but also in classical-time inscriptions (cf. STUR20, p. 168). It was on such daydreaming that Catholicism was based (for more catholic absurdities on the assimilation of Ν, see PICK18, p. 276).Taking this occasion, Pickering remarks (p. 265): "it will be a useful lesson to us, who are but foreigners as to this question, to recur to the theoretical arguments, which those eminent men and their followers have urged against it. It will teach us to be cautious, in questions of this nature, how we condemn the universal and very ancient usage of a whole people, whenever it happens to be repugnant to our own habits or prejudices" (emphasis in the original). As evident by the events of the next two centuries, not only did the Catholics pay no heed, but they have brought these quixotic theories to a whole new level.
The modern Catholics, having seen the manuscripts and the inscriptions, could obviously not follow this line of argument. They had to come up with more sophisticated explanations of the inconvenient orthography (ΒΔΓ3). The seeds were sown by the latinate scholarship of the 19th century, who were seeking to import to Greek features ascribed to Latin (evidently on account of peculiarities of Italian). Thus, at the end of the century, Blass reports that "Some would assume the guttural nasal, written γ, before μ and ν" and provides the first rudimentary evidence for a nasal pronunciation of Γ in the groups ΓΝ, ΓΜ (BLAS90, pp. 88-89). Sturtevant takes the opportunity to declare the emergence of this conditional pronunciation of Γ before nasals the possible "external cause" that triggered the "revolution in orthography" (<ΝΚ>,<ΝΧ>,<ΝΓ>→<ΓΚ>,<ΓΧ>,<ΓΓ>) and provides "three additional [to the fact that <GN> was probably [ŋn] in... Latin!] reasons for believing that the change actually occurred" (STUR20, pp. 168-169).
ΓC/: Γ/_Μ = Γ/_Ν = [ŋ]
Allen ratifies Sturtevant's conclusion, but only as far as <ΓΜ> is concerned, as he remarks: "Surprisingly, however, there is no cogent evidence for γν = [ŋn], so that in this respect the Greek situation appears to be the reverse of the Latin" (ALLE87, p. 37). This is indeed surprising, as the reason why the Catholics looked for evidence of a diverging pronunciation was their conclusion that <GN> stood for [ŋn] in Latin. The development of events in the case of the clusters ΓΝ, ΓΜ can be seen to have taken an unexpected turn:
In other words, starting with wishful thinking and the wrong assumption ("Preconceived Theory"), the Catholics discovered evidence ("Evidence Selection") for reaching a precarious conclusion that was the necessary "scientific" basis ("Scientific Claim") for explaining away the inconvenient representation of the velar nasal by Γ. Does anyone still have doubts that mainstream scholars are the best implementors of The Recipe? I dread the moment a Khoisan, Hmong or Navajo scholar decides to follow in the footsteps on his western colleagues and try to impose features of his native language on Greek.
Even though the methodology is unscientific, the evidence circumstantial and the conclusion opportunistic, which should have led all decent scholars discard the whole theory as yet another (in Blass' words) "extraordinary piece of perversity", we are obliged to examine the evidence put forward, as the first half of ΓC/ has made it into the Bible of the latest Pope, who is adamant that "the pronunciation [ŋm] is recommended for γμ in all cases". The arguments presented by Sturtevant (and earlier by Blass) can be summarised (and commented upon) as follows:
The name agma for the velar nasal appears only once in Priscian, who quotes Varro, who reports the theory of some Ion (STUR20, p. 89). It is evident that "agma" is hearsay of hearsay and we cannot be sure whether the sound had a name, much less whether the original form of the name was "ἄγμα" (as the Catholics are eager to accept) or "αγμμα" (a "transposition of γάμμα" as Blass believes; BLAS90, p. 88) or "ἄγγμα" (as emended by others; BLAS90, p. 89). Thus, the "agma" story should be taken with a grain of salt. What is more important is that the examples provided by Varro are "aggulus aggem agguilla iggerunt" and "agceps agcora" (i.e., before g and c), but "neither he nor any other Grammarian says anything about the occurrence of the same sound before m n" (BLAS90, p. 88, n. 4).
Sturtevant concludes that if one media (Β) becomes a nasal before another nasal, so should the other media (Γ). This case of proof by analogy actually works contrary to intension, if all facts are considered, for besides Β and Γ there exists a third media (Δ), which is not converted to a nasal. This is already pointed out by Blass (BLAS90, p. 89), but Sturtevant tries to explain away the resistance of δν and δμ to a similar change as rare and Homeric, and to prove the reduction of some instances thereof to nasals by means of a speculative spree involving asterisks and metatheses (STUR20, p. 169). The official description of the behaviour of mutes before Μ is provided by his contemporary Smyth (SMYT20, p. 26, §§85-87): labials become Μ, velars become Γ, dentals become Σ.Smyth claims (SMYT20, p. 26, §87) that the development of dentals to Σ is probably analogical, judging from the occasional maintenance of dentals before Μ (πότμος, αριθμός). Nevertheless, it is strange that a similar analogical change is not claimed or at least investigated for the velars, given that they, too, occasionally resist change before Μ (ακμή, δραχμή; §85a). Since the exceptions do not seem to concern inflection (which is the main area the later-discussed argument of Allen's focuses on), I will adopt the general conclusion that, before Μ, labials→Μ, dentals→Σ, velars→Γ. If these three results have anything in common, it cannot be that they are nasals (which Σ is not), but that they are continuants! Thus, even if one believes that there is an analogy in the behaviour of mutes before Μ, ΓO provides a satisfactory association, whereas ΓC/ does not. For the behaviour before Ν, Allen's summary rejection of the [ŋn] value, as well as the lack of anything similar for the dental series suffice for the time being.
The former is explicitly considered "too isolated to be significant" by Allen (ALLE87, p. 37, n. 59). The latter concern γίγνομαι (see next piece of "evidence") and Ἀριάννη, which is identified as misspelling of Ἀριάγνη; it is not clear to me why the first ν cannot be considered to be due to reduction of δν of the more popular form "Ἀριάδνη" (cf. the french version Ariane). The spelling "φθέγγματα" may well be etymological from φθέγγομαι (cf. Allen's "contribution" below).
γίγνομαι→γίνομαι is singular in ancient Greek and no other instance of ΓΝ (e.g., γνώμη, ἄγνοια, etc) seems to have followed down the same path; Allen's comment in that matter (ALLE87, p. 37, n. 60) is rather obscure ("there may be special considerations connected with the preceding γ"). The conclusion drawn from πρᾶγμα→πράμα in the modern language is yet another illusion of preservation deserving separate treatment, as it has stealthily intruded in Allen's argumentation.
Indeed, it is not uncommon for the γ in modern πράγμα|thing to be omitted, so that it often rhymes with "γράμμα|letter"(=[ˈɣrama]). One also finds other similar omissions, such as σπουδά(γ)ματα (in a kid song where πράματα rhymes with γράμματα) and βρε(γ)μένος|wet, soaked (from βρέχω|I dampen, I rain). But this is as much as we can tell about this phenomenon. To paraphrase Blass: "To infer from this, as some have done, that the Greeks pronounced [γ as the velar nasal before μ], is an extraordinary piece of perversity". The bulk of words comprising the cluster γμ are pronounced with the normal [ɣm] and never with the reduced [m], e.g., δόγμα|dogma, ρήγμα|fault, divide, σίγμα|sigma, μίγμα|mixture, πλέγμα|grid, τάγμα|platoon, δείγμα|sample and the most popular of all παράδειγμα|example. One cannot explain all these instances away as "learned pronunciation" and claim that πράμα has more weight.
It is interesting to quote what Chatzidakis has to remark on such theories of preservation of a phantom pronunciation: "διδάσκεται ἀπὸ μακροῦ ὅτι τὸ γίγνομαι καὶ γιγνώσκω ἐγένοντο γῑνομαι καὶ γῑνώσκω [sic], ἐπειδὴ τὸ γ πρὸ τοῦ ν ἐξεφωνεῖτο ὡς ἔρρινον ṅ καὶ οὕτως ἐξεβλήθη μετ' ἀντεκτάσεως. Ἀλλ' ὅστις ἐνθυμεῖται ὅτι λέγεται παρ' ἡμῖν καὶ σήμερον ἔτι ἐν τῇ δημώδει γλώσσῃ [γν... γν... γν...] κτλ., ἔτι δὲ πολλαχοῦ τῆς Στερεᾶς Ἑλλάδος τύποι μετά τοῦ γμ, οἴον [...γμ...] κλπ., οὗτος δὲν θὰ πιστεύσῃ ποτὲ ὅτι ἐξεβλήθη τὸ γ πρὸ τοῦ ἔρρινου μ ἤ ν, ἀλλὰ θὰ σπεύσῃ νὰ ζητήσῃ ἄλλην ἐρμηνείαν|it is taught since long time that γίγνομαι and γιγνώσκω became γῑνομαι and γῑνώσκω [sic], because γ before ν was uttered as nasal ṅ and was, thus, discarded with compensatory lengthening. But everyone who remembers that in our popular language we pronounce [... list of examples starting with γν ...] etc., and in many places of central Greece forms comprising γμ, such as [... examples with γμ ...] etc., will never believe that γ was discarded before the nasal μ or ν, but he will rather seek a different explanation" (CHAT02, p. 323). The "different explanation" is (yet again) dissimilation (this is what Chatzidakis also claims at the end of the quoted text, but I have not been able to locate the chapter he refers to).It is funny how the Catholics often cancel each other, sometimes even without respecting the principle of... causality, i.e., recent ones cancelled by others who lived 10-100 years earlier! Consider the modern variants απόγε(υ)μα|afternoon=[aˈpoʝe(v)ma], θα(ύ)μα|miracle=[ˈθa(v)ma], cretan πε'μου=[ˈpemu] from πες μου|tell me=[ˈpezmu], idiomatic φκιάνω=[ˈfcano] from φτιάχνω|I make=[ˈftçaxno]. In all these developments, a fricative is dropped before a nasal without being a nasal (or being converted to one) itself; the suppression of the first continuant ([v], [z] and [x] in the examples) is rather the consequence of Greek's intolerance to sequences of sounds that share [+cont] (also [-cont]).The usual behaviour, when the second continuant is a mute (θ, χ), is for it to be turned to the corresponding plosive: φθ→φτ, χθ→χτ, σθ→στ, σχ→σκ. I suspect that, since in the case of the second continuant being a nasal there exists no "corresponding" plosive, the onus is on the first continuant to mutate. Why the mutation is a mere deletion, is something for which I have no further explanation. Nevertheless, the existence of the aforementioned examples of fricative deletion before Μ (which is, in any case, a rare phenomenon) teaches us that this is the preferred treatment of the cluster fricative+[m], whereas I have no examples of a fricative being converted to a stop in the same position. There is, therefore, no need to assume Γ=[ŋ], since Γ=[ɣ] is sufficient reason for the deletion of Γ before Μ. So, let's be... pragma-tic.
It is evident that the "evidence" produced in support of ΓC/ is circumstantial and superficial. This is probably the reason why Allen rejected the ΓΝ=[ŋn] thesis, but he could not help make his own "contribution" (plus he had to keep the general thesis alive for providing a half-credible explanation for the emergence of "nasal Γ" before velars): by comparing the perfect-passive forms λέ-λεγ-μαι (from λέγ-ομαι) and ἔ-φθεγ-μαι (from φθέγγ-ομαι), he concludes that in the second case the nasal of the stem (the first γ of φθέγγ-) appears to be lost; he then remarks that everything "would be explained, however, if the γ of γμ were pronounced [ŋ]", as the resulting γγμ=[ŋŋm] would be "phonetically simplified" to [ŋm]=γμ (ALLE87, p. 36). He devotes the rest of his arguments to fend off criticism of his thesis, occasionally using one of the above-mentioned arguments of his predecessors, and for expanding his theory into grey zones (such as the surprising suggestion that the preposition ἐκ was pronounced [eŋ] before μ, on account of its occasional appearance as εγ, such as εγ μακεδονιασ).
Allen's argument apparently relies on the need to deny the loss of a stem-internal nasal. However, admitting such a loss may not be so dramatic an option. After all, the nasal is absorbed or deleted in the other (non-velar) cases of nasal+mute, such as πέμπω→(πε-πεμπ-μαι→)πέπεμμαι (cf. SMYT20, p. 26, §85b), even κάμπτω→(*κε-καμπτ-μαι→)κέκαμμαι and, most importantly, κυλίνδω→(*κε-κυλινδ-μαι→)κεκύλισμαι. As to the reason why the nasal is deleted in the case of γγμ (the deleted nasal being the first γ), it is enough to point out that (to the best of my knowledge) a consonant is never sandwiched between two nasals in greek morphemes;Clusters NCN (N:nasal, C:any consonant) may be produced in composition, such as εν+πνέω→ἐμπνέω, but there special considerations apply for the ancient language (cf., e.g., ALLE87, p. 46, n. 81). therefore, something has to give and it could be either the first γ (nasal) or the second γ (velar); loss of the velar γ would result in a [ŋm] cluster, the survival of which is "improbable in view of the elimination of [nm]" (ALLE87, p. 37, quoting Lupaş);The counterargument of Allen's that "a difference of treatment" is not to be excluded a priori is correct (although the example used is not from a spoken living language, but from a never-heard written one), but... irrelevant. First of all, the combination γμ (which should be [ŋm] according to this theory) is not as rare as Allen appears to be suggesting (just think of the words ending in -γμα and -γμενος). Most importantly, the objection is one of "improbability" (based on the avoidance of the sequence νμ=[nm]) not "impossibility". we are, thus, left with the only option of eliminating the nasal, the sequence ΓΜ (wherein Γ is the voiced velar) being perfectly acceptable, e.g., at least as the voice-assimilated version of ΚΜ and ΧΜ (as in "ἀκμή", "δραχμή"). So far, no specific velar value has been assumed for (normal) Γ, but if the continuant value ΓO is assumed, the vulnerability of the cluster [ŋɣm](=γγμ) will be even more evident. Allen's "contribution" to the destruction of yet another sound (that of Γ) receives a satisfactory explanation without having to assume an unattested exotic pronunciation.
Indeed, for the pronunciation ΓC/ there is absolutely no direct evidence. The only "evidence" cited is indirect, i.e., aberrant behaviour that has led some scholars to infer Γ=[ŋ] before Μ or Ν, but no direct evidence of such a value: the Grammarians that mention agma do not even hint at it being Γ's value before Μ; there exist no transcriptions into languages that possessed characters for [ŋ] (Sanskrit?); there is not the faintest trace of a velar nasal before Μ in any modern dialect of Greek.
Furthermore, the argument that <Γ> was used for the nasal before velar mutes because it had developed a nasal allophone before Μ is rather naive. When a letter develops an allophone, i.e., a different value in a certain context, it does not make sense to use the letter out of context for denoting the same sound. The usual application involves the transfer of the environment together with the letter: "English, for example, writes n for [ŋ] before velars and ng [not merely n] elsewhere" (ALLE87, p. 39); french G has the "normal" value [g], but it has also developed a post-alveolar allophone [ʒ] before front vowels, so that when the latter value needs to be indicated before back vowels, <ge> is used (cf., e.g., plonger|to dive→nous plongeons|we dive[FRA] vs. passer|to pass→nous passons|we pass[FRA]); [b] has emerged in Greek as a post-nasal allophone of Π, so that when a representation of the sound is needed, the nasal (always Μ) is used together with Π (e.g., Μπους|Bush=[bus]); I can think of no example where a letter is used with a conditional value without also using the environment where this value appears. Thus, if Γ had a nasal allophone before Μ and the ancient Athenians wanted to indicate the same value of the nasal phoneme /N/ before velars, the most reasonable choice would be the digraph ΓΜ (e.g., †ἄγμχος, †ἄγμκυρα) and not Γ alone.
All in all, the Catholics have made a big (and unjustified) mess in their attempt to find suitable explanations of a phenomenon (use of <Γ> for [ŋ]) otherwise unexplainable under their plosive model (ΓC). In other words, instead of acknowledging the superiority of the explanation offered by the orthodox model (ΓO), the Catholics have tried to justify their wrong assumption (ΓC) by engaging in far wronger speculation (ΓC/). It is time for the Catholics to accept the orthodox values of the mediae, try to familiarise themselves with them and stop chasing chimeras.
There are many indications of lesser (compared to the "nasal Γ") importance that point to a fricative (or at least non-plosive) pronunciation of the mediae. In accordance with the Aggregation Principle, their significance increases when considered together.
The first witness we may summon is none other than Chatzidakis. We have seen that, despite being a devout Catholic, he condescends to grant the orthodox pronunciation of the aspiratae several more centuries of validity by producing supportive evidence (greek transcriptions of persian and italian fricatives) carefully chosen to be barely post-classical (from IV BC). He uses the same approach in the case of the mediae. After having subserviently interpreted the teaching of the Grammarians (of I-III AD and even the Scholiast of Thrax!) in favour of the Catholics (basically relying on the already discussed prejudices about the labial series), he declares their teaching inconsistent with the pronunciation of their time, claiming that it was valid in V BC (for obvious reasons), but not later, because from IV BC the foreign [b] is variably rendered as β, ββ, μβ; the main example he cites is the inscriptional Ἀρύββας (attic 343 BC), which is found as Ἀρύμβας in Demosthenes (cf. the various spellings of the name) and which he juxtaposes to the later Ἄθαμβος, Ἄθαββος, Ἄθαβος in Delphi (II BC) and Ἀμβακούμ of the Septuagint (normative Ἀββακούμ). Before rushing to celebrate on Chatzidakis' admission, one must wonder how he knew that the original sound rendered as β, ββ, μβ was [b]; after all, we do not know the languages where the first two names belonged to, much less their phonologies, whereas the last one is of unknown etymology. The words he uses are: "Οἱ ποικίλοι οὗτοι τρόποι β, μβ, ββ τῆς παραστάσεως τοῦ ξένου φθόγγου b διδάσκουσι μεγαλοφώνως ὅτι πάντως τὸ Ἑλλ. β δὲν ἐξεφωνεῖτο τότε ὡς b|The various ways β, μβ, ββ of representing the foreign sound b teach loudly that in any case greek β was not pronounced as b at that time" (CHAT02, p. 438); indeed, the variety of representation makes us suspect that the represented sound could not be uniquely or adequately represented with the available greek letters, but it is not clear whether the sound was [b] and β's normal value was [v]/[β] or vice versa. Chatzidakis' thesis appears to rely on the a priori admission that post-nasal mediae were (always) stops (cf. pp. 440-441), so that μβ could not relate to the fricative sound. As we do not want to base the proof on assumptions, we will make a note of the evidence and the... general mood of Chatzidakis, but we will not take the fact as decisive.Another example of the spelling ββ for representing a foreign sound is the word Σάββατον|Sabbath, which derives from שבת (masoretic שַׁבָּת). Here, the double β cannot be explained as an attempt to reflect the foreign spelling, because there is no doubling in the original Hebrew (unlike, e.g., the name Ἰωάννης|Johannes (John), where the double ν is a faithful representation of the hebrew spelling יוחנן). If the masoretic version (where the dagesh suggests that the beth stands for the stop [b]) is to be trusted, then this well-known spelling corroborates the thesis that ββ was a (post-classical) graphemic convention for rendering a sound ([b]), which did not exist in Greek.
Since we have mentioned the spelling ββ, it is worth mentioning that the mediae are never doubled as such. Graphical doubling of Γ indicates a mute (the second Γ) preceded by a nasal (the first Γ);There is a single case where the spelling ΓΓ is considered to involve two identical Γs, namely (in Allen's spelling) εγγονοσ (the other two cases of "εγγραψασθαι" and "κακ γονυ" appear to be attempts to stretch isolated misspellings to verifications of the thesis that γγ sometimes stood for [gg]). This is a peculiar, singular and probably aberrant case. Allen assigns to it the original meaning "offspring, descendant" and derives it from ἐκ+γονος, also mentioning the variant spelling ἔκγονος in manuscripts (note the difference in accent: ancient antepenult vs. modern ultima). To conclude from the frequent spelling εγγονοσ that the first γ was a voiced κ is probably precarious. In general, the κ of ἐκ does not assimilate to a following voiced mute in composition (SMYT20, p. 26, §82 N.2), such as ἐκβάλλω, ἐκδίδωμι (the inscriptional assimilation to εγ before β and δ mentioned by Allen must relate to the preposition ἐκ, which was separate from the main word and not a part thereof, probably a "trendy" spelling), so that the derivation from ἐκ+γονος is suspect. The spelling εγγονοσ might well be the result of false etymology (e.g., assuming it derives from ἐν+γονος, whence the "ἐγγενής 'innate, native, kinred'" mentioned by Allen), which cannot be a priori excluded. ΒΒ probably had a special meaning in words of foreign origin (as explained above) and ΔΔ only appears in attempts to represent dialectal (probably mutated) Ζ. It is, therefore safe to state (SMYT20, p. 25, §81) that
What would account for such a lack of gemination, if β, δ, γ were stops (as in present Italian)? Would ΒΔΓ4 make more sense if β, δ, γ were continuants? A reasonable answer would probably be that, if β, δ, γ were fricatives, their geminated versions would not be acoustically so much different from their singletons, so that gemination would not add an audible distinguishing feature. On the contrary, the maintenance of the occlusion and the delayed release of a geminated stop (such as [pː]) would be phonetically more noticeable. Under this consideration, the ability of the (undoubtedly stops) π, τ, κ to geminate and the lack of such ability for the (assumed fricative) β, δ, γ would not be so surprising. However, there is some further explaining to do before considering this answer satisfactory: why would this inability to geminate not apply to the other continuants of ancient Greek? This is a tricky question. The "other continuants" would comprise μ, ν, λ, ρ, σ and possibly φ, θ, χ. Of these, the first four (μ, ν, λ, ρ) are sonorants, i.e., not exactly the fricatives we assume β, δ, γ to be; nevertheless, it would be difficult to imagine why, e.g., [mː] would be acoustically more salient than [m], but [vː] and [v] would not be sufficiently distinct. A possible explanation would be to identify "gemination" with the maintenance of an occlusion, as done above for the voiceless stops; then the nasals (μ, ν) would provide the necessary (oral) occlusion that could be sustained; the lateral λ certainly involves some contact and a near occlusion (the lateral passageways being barely noticeable); ρρ (ῤῥ) is a special case, for which there are more important questions to be answered (like what exactly it stood for); similarly, σσ is of peculiar origin (*κι̯, *χι̯, *τι̯, *θι̯, *τϝ) and certainly non-Attic; φ, θ, χ are "geminated" as πφ, τθ, κχ in Attic (SMYT20, p. 25, §81), which evidently comprise an occlusion (that of π, τ, κ). Extrapolating from the latter, we may conjecture that an attempt to geminate a true fricative involves the creation and maintenance of an occlusion at the same place of articulation, hence the forms πφ, τθ, κχ.I understand that a collateral consequence of this conjecture (gemination=sustained occlusion) is a possible confirmation of those who identify σσ with [t͡s]. Indeed, σ(=[s]) is a fricative and the associated occlusion would be an alveolar [t] (not necessarily identical with dental τ=[t̪]? and, thus, not so liable to be reflected in a spelling τσ, as argued by Allen). As I have demonstrated, I have no prejudice against alternative values; I actually do propose some myself. If the evidence suggests an affricate value for σσ, so be it. At any rate, such a conclusion would not have any repercussion on the subject of our investigation, Attic and its offshoot. As mentioned, the corresponding attic form is ττ, which does not need to be assumed to stand for anything other than what it appears (i.e., [tː]), affricates being alien to Attic. The adoption of the (ionic) spelling σσ in the Koine does not necessarily imply a corresponding adoption of a posited ionic pronunciation (i.e., [t͡s] or [tːs]), but it could well have involved a "spelling pronunciation" (which would also explain the present pronunciation of σσ as [s]). The lack of affricates in Attic and the inability to distinguish geminated fricatives from their singletons would also explain the lack of genuine gemination of σ (to [sː]) and the simplification of genuine σσ to σ (SMYT20, p. 29, §107). An analogous treatment of (fricative) β, δ, γ would necessitate the creation of a similar occlusion, i.e., clusters [bv] (or [bβ]), [dð], [gɣ]; however, the first and the second constituents of these clusters would be perceived as equivalent or even identical (cf. Βalt, Δalt, Γalt) and only a trained ear would be able to discern these clusters from the single consonants β, δ, γ. Certainly, this explanation entails some speculation (gemination=maintenance of occlusion) that needs to be verified before the theory bears any weight; however, for the time being it appears to be the only theory that together with the orthodox (fricative) values of the mediae would explain ΒΔΓ4; on the other hand, there seems to be no way to reconcile ΒΔΓ4 with the catholic (plosive) values. Incidentally, this theory would also explain the choice of ββ for the (foreign) sound [b] mentioned above, which would only occur in the geminated form of the letter (and possibly also after μ, whence the convention μβ).
Chatzidakis also lists (CHAT02, p. 435) several instances of Β for "original" Ϝ (digamma, most certainly /w/) of non-Attic dialects (but one of them reported in Xenophon's Anabasis) from IV BC. It is evident that [b] for [w] is inconceivable, whereas [v/β] for [w] is a coarse (yet acceptable) approximation. It is also evident that, at the time of the inscriptions, there are the following possibilities: a) Ϝ maintained its original sound [w] or b) Ϝ had undergone fricativization (and de-velarisation) to [v] (as per modern Tsakonian); A) Β had the predominant value [v/β] or B) Β had the value [b]. These define four possible cases (aA, aB, bA, bB) of which the two "orthodox" cases (aA, bA) always make sense (in aA, [v/β] is a good approximation for [w]; in bA, the two sounds are identical), one "catholic" case (aB) never makes sense ([b] is a very bad "approximation" for [w], particularly when <ΟΥ> or even <Υ> was available) and the other "catholic" case (bB) only makes sense under special circumstances: "ἐλλείποντος του Ϝ ἐν τῷ νεωτέρῳ Ἰωνικῷ ἀλφαβήτῳ δι' οὐδενὸς ἄλλου συμφώνου ἠδύναντο νὰ παραστήσωσιν ὁπωσδήποτε τὸν πρότερον διὰ τοῦ Ϝ παριστώμενον φθόγγον ἤ διὰ τοῦ β· θὰ κατέφευγον ἄρα εἰς τὴν γραφὴν ταύτην καὶ ἂν τὸ β ἔτι ὡς b ἤθελεν ἐκφωνεῖσθαι|having Ϝ been left out of the new Ionic alphabet, they would certainly not be able to render the sound earlier represented by Ϝ but by means of β; hence, they would resort to this spelling even if β were pronounced as b". Here, Chatzidakis seems to forget his previous corollary that Β was not [b] in IV BC, speculate about the circumstances of the reported instances (e.g., that the non-Attic spellings were written in the ionic alphabet), leave the question of "τὸν πρότερον διὰ τοῦ Ϝ παριστώμενον φθόγγον|the sound earlier represented by Ϝ" open (was it still [w] or had it already changed to [v]?) and engage in the typical catholic argumentation used to explain away inconvenient evidence for justifying some imaginary pronunciation (we have seen this practice in the case of Φ for F). The evidence is certainly not incontrovertible, but the orthodox explanation definitely involves the least (if any) assumptions.
Of greater importance is certainly the frequent omission or even wrong insertion of Γ (in egyptian papyri, it bridges the gap between vowels, which is awfully reminiscent of modern forms like αγέρας from αέρας|air, wind in Greek and oigo|I hear[SPA] from oyo in Castilian; cf. LLOY87, pp. 353-354), which leads Blass to state that "the sound was so undefined and weak, that it was thrust in and left out at will"" (BLAS90, p. 110). The phenomenon of omission is not confined in peripheral dialects or forms of the language, but it "is occasionally found in Attic from the late 4 c. B.C" (ALLE87, p. 32). An attempt to explain this away is made by Chatzidakis, who asserts (CHAT02, p. 439) that one instance is analogical, the other is dissimilatory and the third one is false etymology. Even though analogy was the favourite toy of the linguists of the 19th century, modern linguistics is not so fascinated by it (cf. Lloyd's statement: "simply because one can formulate a particular proportion there is no guarantee that an analogical formation will necessarily result") and it appears that no future scholar has followed Chatzidakis down that path. Instead, Sturtevant is convinced that the phenomenon constitutes "The earliest clear indication of the spirant pronunciation of the voiced mutes" (STUR20, p. 184), a conclusion that Allen (who, as we have seen, badly misanalyses the phenomenon) does not contest; instead, he opts for Blass's approach ("The phenomenon was however in any case strange to the standard Attic"; BLAS90, p. 110) of brushing the evidence aside as rustic, with reference to Herodian, who "specifically states that Plato Comicus treated it as a barbarism in attributing it to the demagogue Hyperbolus" (ALLE87, p. 32). Allen (probably intentionally) refrains from specifying what "it" is, so that the reader may be misled to believe that the rejected feature is the "spirant pronunciation" of Γ; however, it only takes a careful study of the relevant passage to realise that Plato was criticising not any particular pronunciation of γ, but "τὴν ἄνευ τοῦ γ χρῆσιν ὡς βάρβαρον|the use without the γ as barbaric". Indeed, even today the omission of letters (not only in modern Greek) is considered a sign of lack of education or sophistication; but, in order for a consonant like Γ to be prone to omission, it is necessary to assume that its pronunciation was laxer than that of a voiced plosive. Thus, cultivated ancient Athenians, just like the modern ones, did not "slur" their Γs, but the reported omission (remember that Plato Comicus was senior to our familiar Plato and lived in V BC) would only be possible if Γ was "officially" pronounced as the fricative of modern Greek (cf. τρώγω|I eat→τρώω) or, at most, the approximant of late Latin assumed as the reason behind reductions like magister[LAT]→(maistre[FRA]→)master.
As for Δ, abundant evidence for its non-plosive nature (or its compliance with the corollary that Ζ=[z]) has been presented in the chapter on Ζ: only Δ=[ð] (together with Ζ=[z]) explains the statement of Archinus that Ζ is generated "ὥσπερ τὸ δ|just like δ"; Ζ appears in lieu of Δ (or ΔΔ) in dialects that are known to have assibilated Δ or otherwise brought the two phonemes closer together (CHAT02, pp. 441, 444; also BLAS90, p. 113); limited occurrences of such variants are also reported for Attic (THRE80, p. 550: "rare cases of Δ for Ζ on graffiti from the Agora"); the ease with which the suffix -ζε (not always originating in -σδε) was established as a variant of -δε in Attic is more understandable if they were both fricatives; the orthodox value of Δ (and all other involved letters) adds point to the pun "ὦ Βδεῦ Δέσποτα" far better than the catholic one does (if at all); the phonological origin of Ζ not only in δj (and γj), but also in plain *j is reasonable for fricative value of Δ (and Γ), particularly taking into consideration the similar development (assibilation) of western-Romance dj (and gj), which shared the fate of consonantal I=/j/ (LLOY87, pp. 133, 162, 247, 296).The claim that σ changes in "aeolic and ionic into δ, as ὀδμή ἴδμεν for ὀσμή ἴσμεν" may imply an affinity between the sound of voiced Σ (assumed [z]) and Δ (indeed it is found in Homer's Iliad and Odyssey); however, Smyth considers the latter to be analogical and the former to stand for ὀδ-σμη.
All of the above receive a satisfactory explanation under Βalt, Δalt, Γalt with predominant values ΒO, ΔO and ΓO, rather than under the strict catholic thesis, which identifies the mediae as voiced stops (ΒC, ΔC, ΓC).
The aforementioned phenomena, which receive a superior explanation under the orthodox model, have not been unknown to the Catholics. However, something prevented them from seeing the simple answer lying in front of their eyes and made them develop exotic theories for explaining the evidence away. That something was obviously their conviction about the correctness of the catholic values, which was not only an act of faith, but was based on the interpretation of further evidence. Some of the arguments have already been discussed in the previous chapter, as they are identical to those used for the δασέα (classification with the ἄφωνα, nasals do not disappear before the mediae, a vowel followed by β/δ/γ+ρ defines a "common" syllable, etc). Others are trivial and little-thought (e.g., the "πίνειν καὶ βινεῖν|drink and screw [Dillon's translation]" of Aristophanes Frogs 740 is as much alliterative as "tucking and fucking" would be in English; "βροντὴ καὶ πορδὴ ὁμοίω|Trump and Thunder, are similar to each other [Hickie's translation]" of Clouds 394 is as justified under any pronunciation as would be to liken "brat" to "fart"). My intention here is to demonstrate that even the most important ones are the results of misunderstanding or narrow-mindedness.
There can be no doubt that the primary reason why the Catholics sneered at the traditional values ΒO, ΔO, ΓO and proposed [b], [d], [g] instead was ΒΔΓ1 (e.g., BLAS90, p. 108: "Latin b g d and Greek β γ δ correspond to one another with perfect regularity"). This fact alone does not warranty the identity of the corresponding Greek and Latin sounds; for example, Spaniards and Brits use each other's words without converting the bdg's (e.g., colorado is not spelled coloratho in English), which they pronounce in accordance with their own habits (e.g., líder|leader[SPA]=[ˈliðer]). There exists, however, a related argument that was introduced by Erasmus (cf. PICK18, pp. 250-251) and hints at the similarity of sound: in Fam. ix 22.3, Cicero "identifies the pronunciation of βινεῖ with that of the Latin bini" (ALLE87, p. 31); in actual fact, Cicero only states that "bini" sounds obscene to the Greeks, without mentioning the word "βινεῖ" or its meaning (nevertheless, we can think of no other explanation for this statement of his). Based on these facts, which suggest a correspondence of β δ γ and b d g, Blass' conclusion that, since "the value of the Latin mediae is certainly identical with that of the present Romance and German, the pronunciation of Greek β γ δ must have been approximately the same as that of our mediae" seems reasonable.
I hope I have trained you enough to spot the missing link in the above reasoning: how do we know that the pronunciation of b, d and g at the time of these transcriptions and statements was "certainly identical with that of the present Romance and German"? As this line of reasoning merely involves kicking the can down the road, let us review how we know that their values in classical Latin were [b], [d], [g]. Here is the list of arguments from Allen's Vox Latina (ALLE78, pp. 20-23):
Do not adjust your browsers! The list is intentionally left blank, because there is no single argument provided about the plosive nature of latin b, d, g, but the values [b], [d], [g] are assumed, i.e., they are a priori dogma, clearly because no German, Brit, Italian or Frenchman (the ranks of which provided the vast majority of western scholars) could think of a different pronunciation, thus giving the illusion of a unanimous tradition amongst "the present Romance and German". Nevertheless, the german pronunciation cannot be considered to constitute part of the latin tradition, while unanimity amongst Romance has not prevented mainstream scholars to propose diverging (from tradition) values for latin C, H and (particularly) consonantal V. As evidenced by modern Spanish (as well as Portuguese, Catalan, Occitan, etc), Romance tradition of the mediae is not as uniform as we are led to believe, certainly less uniform than the tradition of, e.g., consonantal V, which is uniformly [v], but considered to have been /w/ (e.g., ALLE78, p. 42: "there is no such evidence for any such development [=fricativization of consonantal V] before the first century A.D., and the [w] value of consonantal u must be assumed for the classical period"). Ironically, it is thanks to the latter that we know that the Spanish pronunciation was the norm rather than the exception in Latin of the christian era, as acknowledged by none other than Blass: "The Latin b too in many places had a similar development [i.e., coincidence with v], being pronounced in the same way that survives at the present day among the Spaniards and many of the French of the south, whose vivere is according to the well known witticism bibere" (BLAS90, p. 109). The reason is that "from the first century A.D. onward, we find confusion of the letters V and B in many written documents" and this is not confined in word-internal positions, but "incorrect use of the lettres B and V(U) is found initially also" (LLOY87, p. 132; further examples in PICK18, p. 254 and STUR20, p. 43). But this is not all; greek Β also stands for latin V in post-christian transcriptions: "β for v is common from the first century on," (GRAN07, p. 135).Isolated instances are reported since much earlier: "τὸ Λατ. v παριστᾶτο [...] απὸ δὲ τοῦ 180 π.Χ. καὶ διὰ τοῦ β (πρβλ. Λίβιος, Βίβιος ἐκ Δήλου)|latin v was represented [...] since 180 BC also by β (cf. Λίβιος, Βίβιος from Delos)" (CHAT02, p. 436); another "two exx. from the time of the republic" are mentioned by Blass (BLAS90, p. 109, n. 3). As already pointed out, the story of confusion of latin V and greek Β parallels that of latin F and Greek Φ, wherein sporadic instances of correspondence of the two letters come into sight in II BC and become more common from I AD. The entire process of convergence must have taken in both cases more than two centuries to complete; the continuation of spellings with ΟΥ for latin V until V AD are probably due to a scholarly or conservative pronunciation of latin V by the upper or educated classes (yielding inconsistent or even double representations of the V of latin names in Greek as ΟΥ or Β); naturally, there was no way to represent a corresponding conservative pronunciation of latin F (proposed as Falt), because, contrary to [w] which could be approximated by ΟΥ, there was no appropriate approximation for (the assumed) [ʍ] and Φ=[f] was the only candidate. There are, therefore, the following important facts that characterise the relation between greek β and latin b and (consonantal) v:
The interpretation of βbv1-βbv4 by the Catholics is supernatural: they claim that initially β=b=[b] (and v=[w]), then around I AD or later β→[v], b→[v], v→[v] (e.g., STUR20, p. 112: "Before there could be confusion between b and v both must have become spirants")! It is as if a black hole was established at [v] (a sound hitherto allegedly unexpressed in both languages) and sucked all voiced labials of the empire. It appears that, in their desperation to justify the "german pronunciation of Greek", the Catholics are eager to deny what is in front of their eyes. A reasonable person can but arrive at one conclusion:
β and b coincided at [v] (or [β]) before I AD and v met them there (on its way from [w]) afterwards
In other words, the argument that b and β were identical works against the catholic thesis and the practice that should be amended is not the modern pronunciation of β by the Greeks, but the modern pronunciation of b of western GermanicI am not sure whether the culprit should be identified as Western Germanic in general or High German in particular. It is believed that the voiced "mutes" of Proto-Germanic were also largely fricatives and only marginally stops, with the allophonic distribution generally agreeing "with the patterns of voiced obstruent allophones in languages such as Spanish". It is probably worth noting that the principal view used to be that they "were originally fricatives and later 'hardened' into stops in some circumstances", but now this view is considered (?) obsolete, because "it is equally possible that the allophony was present from the beginning"! Whatever the case may have been, the official thesis is that the voiced "mutes" of Germanic were at least predominantly fricatives and developed into stops after the second germanic consonant shift. It is interesting to point out that, if the catholic thesis were correct, it would have the surreal consequence that voiced "mutes" evolved along opposite directions in Greek (stops→fricatives) and Germanic (fricatives→stops) at roughly the same time! and of the germanised Romance. This had already been proposed a couple of centuries ago by Pickering, who was ahead of his time (PICK18, p. 253: "very strong evidence, that the Romans did for a long period pronounce their B so nearly like a V, as in general not to be distinguishable from it, at least by the ears of foreigners, just as is the case with the Spaniards at the present day"He adds as a note: "I have often thought it probable, that the Spaniards and Portuguese, who were a Roman colony, may have retained more of the ancient masculine [sic] pronunciation of the Romans, though tinctured perhaps with a provincial rusticity, than is to be found even in Italy itself; for the Latin language throughout Italy, which was the great theatre of the operations of their enemies, the Goths, must have been much corrupted by the constant and immediate influence of the language of those invaders; while the provinces of Spain and Portugal, being remote from that influence, might preserve the language, which they originally received from their Roman masters, in greater purity than the people of Italy". He is perhaps the first scholar to point at the probability of germanic influence, although the Goths were almost certainly not the culprits; for one, there were Goths in Iberia, too, and stayed there for a longer time; furthermore, the gothic voiced "mutes" of IV AD were also largely fricative, at least with a distribution that resembled the one reported for modern Spanish (cf., e.g., STUR20, p. 185). If there was indeed a germanic influence (which is not at all unlikely), the responsible should be sought among the ranks of western Germans, i.e., the Franks and Burgundians of Gaul, the Lombards of Italy and even the French-speaking Normans of Sicily.), but his wisdom was unfortunately overrun by the prejudices of the german scholars of the later half of the 19th century.
Just how well established and perhaps advanced (i.e., "modern") the voiced fricatives were in I AD can be seen from their use to represent foreign sounds that were clearly not stops.
Thus, contrary to the prevailing dogma, not only are the catholic values of Β, Δ, Γ not confirmed by their correspondence with latin B, D, G (ΒΔΓ1), but, if all evidence is considered, a non-plosive realisation of both greek and latin mediae has to be concluded.
One of the oldest arguments in Catholicism, which even predates Erasmus (according to ALLE87, p. 142, it is fathered by Aldus Manutius), relies on the transcription of the sheep's cry. Although we have pointed out the difficulties in accurately rendering the cries of animals, demonstrated the inconsistent manner in which different languages render the same cry and concluded their inappropriateness for determining human pronunciation, the argument is a persistently recurring one, brought up by Catholics more often than the Pope recites Hail Marys. The argument goes like this:
The ancients tell us that the sheep bleat "βῆ βῆ", which could not be the "modern" [vi vi], but could only stand for [bɛː bɛː], ergo β=[b], η=[ɛː]
I am not stating the first part (about the ancient testimony) as a fact, because it is inaccurate, if not misleading. To begin with, "βῆ βῆ" for the cry of sheep is not the direct testimony of the "ancients", but actually comes from the "moderns"; indeed, a verse from Dionysalexandros of Cratinus ("ὁ δ' ἠλίθιος ὥσπερ πρόβατον βῆ βῆ λέγων βαδίζει") is quoted by three sources (cf. JANN95, p. 46, n. 1), which date between X AD and XIII AD (the earlier manuscripts being "of a still later date (13th-15th century)") and have obviously copied from each other (the apparent order being Suidas→Zonaras→Eustathios).There is a second "fragment" attributed to Aristophanes that allegedly verifies the bleat as "βῆ" ("Θύειν με μέλλει καὶ κελεύει βῆ λέγειν"). This fragment is mentioned by Jannaris (JANN95, p. 46), but it is curiously not at all commented upon; the argumentation focuses exclusively on the passage by Cratinus. It does not feature in any of the sources quoted by Jannaris (but briefly mentioned by Meineke in connection with the excerpt from Dionysalexandros) and I have not been able to spot a Catholic who has used it instead of or in addition to Cratinus' verse. I recall having read a comment that "βῆ" is actually the imperative "go!", which would make sense in the passage by Aristophanes and might add to the uncertainty about whether that "βῆ" truly stood for the sheep's cry. Perhaps this is the reason no-one quotes it anymore. At any rate, anything we say on the "βῆ βῆ" of Cratinus also applies to the "βῆ" of Aristophanes, if it refers to a bleat. Neither the context of the verse (i.e., the rest of the comedy or at least an extended excerpt) nor any other information is provided, except for the comment accompanying it: "Βῆ: το μιμητικόν τῆς τῶν προβάτων φωνῆς, οὐχὶ βαὶ λέγεται Ἀττικῶς|Βῆ: the imitative of the voice of the sheep; in Attic, it is not βαὶ". There are a number of further issues that result from this clarification:
In view of the above, it is more correct to state as a fact that
Now, let us see whether we can, nevertheless, make use of this information. We will leave the examination of the vowel part to the corresponding chapter and concentrate on the consonant part. The belief that the sheep bleat "μπε μπε" ([be be]) is so firmly established in modern Greek that most people refuse to consider any alternative, such as the [me] of Hebrew and even the μἐ(ε) of some Greeks (cf. PICK18, p. 257). This is yet another facet of the catholic dogma; one only has to listen to the testimony of the sheep themselves to realise that the sound is neither [be] nor any other human-reproducible sound and to agree with Aristotle that "καὶ γὰρ τῶν θηρίων εἰσὶν ἀδιαίρετοι φωναί, ὧν οὐδεμίαν λέγω στοιχεῖον|Animals utter indivisible sounds but none that I should call a letter [Fyfe’s translation]" (Poetics 1456β).
The above objection notwithstanding, let us assume that the sound Cratinus "heard" was [be(ː) be(ː)]. How would he render the consonantal part of the sound? Clearly by Β under the catholic model (ΒC). However, what the Catholics fail to apprehend is that both the orthodox (ΒO) and the alternative (Βalt) models lack a specific representation for the sound [b] (the modern rendering <ΜΠ> having been established in post-christian times), so that an approximation or compromise would be necessary; there appears to be no other reasonable possibility than to use <Β>, even if that letter normally stood for a voiced labial fricative.The situation is comparable to the representation of the sound of hushing in English (shh!) and Greek (σσ! or σστ!). The Greeks produce the same sound ([ʃʷː]), but have no way to render it as accurately as the Brits, for Greek lacks post-alveolar (and rounded) sibilants; thus, σ is the best available approximation. Indeed, the same approximation seems to have been used in the medieval(=orthodox) βελάζω|I bleat, which never appears as †μπελάζω (if the [b] of [be be] were to be more accurately represented).
But most importantly, the Catholics who are parroting this argument for establishing the sound of Β as [b] ignore the very context in which this testimony has been transmitted to us. All sources emphasise that the oddity of the attic form with respect to their accustomed one lied with the representation of the vowel part, i.e., η instead of αι ("Βῆ [...] οὐχὶ βαὶ|Βῆ [...] not βαὶ"). As we know that the pronunciation of X AD was already "modern" in all respects (specifically for Β, this is proven beyond doubt by the values В=[v], Б=[b] of the cyrillic script established in IX AD), their contemporary rendering of the bleat ("βαὶ") would be pronounced [ve]; but nowhere is it to be discerned that the β of either form struck them as odd. If β was considered satisfactory representation of [b] (or the sheep's cry, in general) at a time where a graphemic convention (<ΜΠ>) was available for rendering the sound [b], why should it be an unreasonable choice for Cratinus, who did not even have that orthographic option?
This had already been substantiated at the beginning of the 19th century by Pickering, who wonders why everyone down to XVI AD "should have transmitted this word from age to age, without ever intimating, that it was a poor imitation, because it began with a letter, which sounded like V and not like B" (PICK18, p. 260); yet, the catholics kept reproducing the naive argumentation, which developed into a popular (or populistic) "proof" of the value of Β. Certainly, blindly accepting the identity βῆ=μπέε and making the corresponding associations without a bleat involves less cognitive burden than making all the above considerations. However, if we allow ourselves to put some of our grey matter to work, we will soon realise that <Β> was the only available option under any of the three (catholic, orthodox or alternative) models.
It is time for this jurassic argument to find its place next to the association θύειν|to kill↔tuer|to kill[FRA] of Erasmus and the diphthongal pronunciation of ου as [ow] and of ει as [ej] by Lebrixa and Cheke (cf. ALLE87, pp. 141-145) in the dustbin of history.
Before moving on and since we are discussing the testimony of the ancients, it might be worth dedicating a few lines to a cryptic passage of Plato's cited by both sides as proof of each one's thesis: in Cratylus 427α and β, "Socrates" explains how the giver of names (allegedly) uses in each word certain letters, whose properties are appropriate for the intended meaning; he, thus, states: "τῆς δ᾽ αὖ τοῦ δέλτα συμπιέσεως καὶ τοῦ ταῦ καὶ ἀπερείσεως τῆς γλώττης τὴν δύναμιν χρήσιμον φαίνεται ἡγήσασθαι πρὸς τὴν μίμησιν τοῦ 'δεσμοῦ' καὶ τῆς 'στάσεως'|And again he appears to have thought that the compression and pressure of the tongue in the pronunciation of delta and tau was naturally fitted to imitate the notion of binding and rest [Fowler's translation]". Allen sees a confirmation of the plosive nature of Δ: "Plato specifically refers to the 'constriction' and 'pressure' of the tongue in pronouncing δ as well as τ" (ALLE87, p. 31). Jannaris, on the other hand, considers the passage a "tolerably clear testimony" of "the existence of interdental ð" (JANN97, p. 60, §61). Chatzidakis is less certain: "εἶναι φανερὸν ὅτι διὰ τούτων δὲν περιγράφεται ὁ τρόπος σχηματισμοῦ τοῦ τ καὶ δ καὶ ἡ φύσις αὐτῶν, ἀλλ' ἁπλῶς τὸ αἴσθημα ὃ ἐκ τῆς ἐκφωνἠσεως τούτων ἐγεννᾶτο ἐν τῷ Πλάτωνι, ὅθεν ἡμεῖς οὐδὲν σαφὲς μανθάνομεν|it is evident that this [description] does not describe the manner of articulation of τ and δ and their character, but merely the sensation caused to Plato by its pronunciation, whence we learn nothing unambiguous" (CHAT02, pp. 342-343). The truth is that... they are all right! The reason is that they rely on different versions of the text. According to the official translation of the aforementioned passage in English, the terms "compression" (not "constriction", as reported by Allen) and "pressure" ("συμπιέσεως ... καὶ ἀπερείσεως") refer to both δ and τ, so that it may appear that they are properties of both. However, an exact translation of the text as published (by the modern editors) is not easy. Its syntax appears to be messed up, as the two properties and the two letters are intertwined; only if the two properties were mentioned together after the two letters (i.e., "τῆς δ᾽ αὖ τοῦ δέλτα συμπιέσεως καὶ τοῦ ταῦ συμπιέσεως καὶ ἀπερείσεως τῆς γλώττης") would the translation be justified. On the contrary, the separation of the two properties and the placement of each next to one of the letters suggests that "compression" characterises δ and "pressure" (or possibly "leaning upon", i.e., "contact") characterises τ; this is consistent with the rest of the text, which associates each different letter with a different property (or many letters with one property, as in the case of the "φυσώδη" φ, ψ, σ, ζ), but never many letters with many properties. The syntax of the sentence and the context of the passage, therefore, suggest that one property ("compression") refers to δ and a different property ("pressure" or "contact") to τ, so that "Socrates" appears to treat τ and δ as if they were letters of different nature (and the difference could not be one of voice).This is also the interpretation adopted in the (modern-)Greek translation: "με το δέλτα η γλώσσα πιέζεται ενώ με το ταυ αντιστέκεται|the tongue is compressed with δ, whereas with τ it resists" (PLAT94, p. 193). The distinction between δ and τ is clearer in this translation. Jannaris appears to base his argument on the aforementioned reasoning, as he relies on the emended form "καὶ τῆς τοῦ τ καὶ ἀπερείσεως" ("especially if we read καὶ τῆς τοῦ τ ἀπερείσεως"). Chatzidakis does not even mention that the passage was almost certainly not transmitted to us sound, but directly uses the emended form, as suggested by Jannaris; his observation that we are not informed exactly how δ and τ are pronounced is correct, but he disregards the fact that the passage (particularly in the emended form he adopts) describes τ and δ as substantially different. All in all, the conclusion that Plato confirms the catholic value of δ is not justified, as the passage is ambiguous and almost certainly in need of emendation, the most reasonable one being that suggested by Jannaris, which is rather compliant with the orthodox model.Another point that certainly needs to be emended is the word "στάσεως". The entire argumentation of "Socrates" relies on the association of each discussed letter with one or more words beginning with this letter and having a meaning that "Socrates" associates with the letter's sound (or the sensation caused by its pronunciation, as submitted by Chatzidakis). Thus, the provided examples are all words that start with the associated letter: ι-ἰέναι/ἵεσθαι, ψ-ψυχρὸν, ζ-ζέον, σ-σείεσθαι/σεισμὸν, γ-γλίσχρον/γλυκὺ/γλοιῶδες. Only λ escapes this rule in two of the four provided examples (λεῖα, ὀλισθάνειν, λιπαρὸν, κολλῶδες), but it is discussed alone, not in association with another letter. Now, for δ and τ, two examples are provided: "τοῦ 'δεσμοῦ' καὶ τῆς 'στάσεως'"; the first does start with δ, so one would expect that the second would similarly start with τ. It, therefore, follows that the original wording must have been "τοῦ 'δεσμοῦ' καὶ τῆς 'στάσεως'", τάσις ("tension") making much more sense than στάσις ("rest" or "standing"), the acoustic identity of "τῆς τάσεως" and "τῆς στάσεως" being sufficient to confuse a copyist who works from dictation. That [t](=τ) relates to a "standing" does not need to be explained. Whether [d] (catholic δ) or [ð] (orthodox δ) better imitates a binding is something that can be subjectively answered either way, but objectively the catholic explanation is not very probable.
Furthermore, it would be wrong to stop reading at that point. The following sentence in 427β describes the slippery or gliding λ and the next one argues that γ partakes in or takes over the gliding effect. Curiously, I have not seen this passage mentioned by any Orthodox, even though it provides a further clue about the nature of γ, but it must have been mentioned at least once, for Chatzidakis feels the need to comment on it: "εἴτε γλυκὐ ἐξεφωνεῖτο εἴτε glukú (ὅπως γλιουκού ἔλεγον οἱ Βοιωτοί), πάντοτε σύνδεσις ἀφώνου τινὸς ὑπερωϊκοῦ μετὰ τοῦ λ γίνεται|if it were pronounced either as γλυκὐ or as glukú (like the γλιουκού spoken by the Boeotians), it is always a coupling of a velar mute with λ" (CHAT02, p. 343). Here Chatzidakis is disregarding the fact that "Socrates" is not speaking about velar mutes in general (there is no mentioning of κ or χ), but about γ in particular (note also how Chatzidakis considers both [ɣ] and [g] to be ἄφωνα). Chatzidakis must have been severely brainwashed to fail to realise that there is nothing in the nature of a plosive ([g]) that would complement, further or enhance the gliding nature of λ and that only the "modern" pronunciation ([ɣ]) renders the cluster γλ indicative of... slippery slime. All in all (cf. also the discussion of the "πνευματώδη" or "φυσῶδη" letters), the etymologies of Cratylus owe their verisimilitude to an orthodox pronunciation of the used examples.
We have already established as a Principle that it is very difficult to confirm the preservation of an (assumed) "ancient" pronunciation in the modern language. The mere fact that a modern feature coincides with (what we think to have been) a feature of the ancient language is not enough proof that the feature has been preserved throughout the entire history of the language; it may well be that the feature (re-)developed due to an irrelevant sound law (cf. the story of the "preservation" of the "e-value" of η). Such superficial similarities have been exploited to support preposterous claims, such as the greek provenance of Finnish or even Hebrew and some Polynesian (e.g., DORI00, pp. 26-27) languages.
Now, the same people who have lambasted such theories are declaring: "Το ίδιο συμπέρασμα αντλούμε από την εξελιγμένη [sic] προφορά του συμπλέγματος -νδ- ως [nd], με αποτέλεσμα να γραφτεί -ντ- προκειμένου να σημανθεί η διαφορά: π.χ. άντρας (< ἄνδρα, αιτιατική του αρχ. ἀνήρ), ντύνω (< ελνστ. ἐνδύνω < αρχ. ἐνδύω), παντρεύομαι (< αρχ. ὑπανδρεύω < φρ. ὕπανδρος γυνή). Ομοίως εξηγείται η συμπεριφορά του συμπλέγματος -μβ-, που προφέρθηκε [mb], με αποτέλεσμα να γραφτεί -μπ- προκειμένου να σημανθεί η διαφορά: π.χ. μπαίνω < αρχ. ἐμβαίνω|We draw the same conclusion from the evolved[!?] pronunciation of the cluster -νδ- as [nd], with the consequence that it was spelled -ντ- in order to indicate the difference: e.g., άντρας (< ἄνδρα, accusative of anc. ἀνήρ), ντύνω (< hellen. ἐνδύνω < anc. ἐνδύω), παντρεύομαι (< anc. ὑπανδρεύω < phr. ὕπανδρος γυνή). Similarly is explained the behaviour of the cluster -μβ-, which was pronounced [mb], having as a result its spelling -μπ- in order to indicate the difference: e.g. μπαίνω < anc. ἐμβαίνω" (WIANPR). The passage is preceded by a first part (which I consider pointless to reproduce here) that declares the (present) pronunciation [ɲɟ] of -γγ- before front vowels unexplainable under the present value of γ (evidently, they would expect the cluster to be pronounced [ɲʝ], since γ is pronounced [ʝ] before front vowels) and an indication of a "preserved" ancient value; the "evidence" is then augmented with similar observations on the (related nasal+media) clusters -νδ- and -μβ-, as quoted above. The reasoning of this passage is messed up (it rivals the fictitious interpretation of Halicarnasseus' description of the velars as evidence of absence of palatalisation before front vowels). First of all, the (occasional, but not universal) pronunciation of -νδ-, -μβ- (and -γγ-) as [nd], [mb] (and [ŋg] or [ɲɟ]) is described as "evolved"; however, if it is "evolved", it cannot be "preserved" (i.e., "original")! Furthermore, the change of orthography from -νδ- to -ντ- does not imply anything on the preservation of the pronunciation as [nd]; even though change in orthography is usually the consequence of a change in pronunciation, the change involved does not have to be the universal sound of δ in all other positions, as implied by the Catholics, but it may very well be the change of -νδ- from [nð] to [nd], i.e., the conditional pronunciation of δ: ð→d/[n]_. Thus, the occasional pronunciation of -νδ- as [nd] is evidence of a "preserved" ancient pronunciation of δ as [d], as much as the forms δείχνω|[ˈðixno]←δεικνύω|I show/point out and οχτώ|[oˈxto]←ὀκτώ|eight are evidence of an ancient pronunciation of κ as [x] or as much as ντόπιος ([ˈdopços]) from ἐντόπιος|native, ντρέπομαι ([ˈdrepome]) from ἐντρέπομαι|I am ashamed and μπλέκω ([ˈbleko]) from ἐμπλέκω|I entangle, prove the "preserved" voiced pronunciation of τ and π as [d] and [b], respectively! Hence, not only is this whole statement incoherent rambling, it has no scientific basis at all.
The authors of the above sentence are not to blame though, as they were not the first to see a confirmation of the catholic thesis in the aberrant pronunciation of the mediae after nasals. Chatzidakis makes essentially the same case albeit with superior wording; he asserts that the ancient pronunciation of the mediae has survived until today in the clusters ν+δ, μ+β, γ+γ; he points out that these clusters were in ancient times distinct from ντ, μπ, γκ; he finally claims that the change of orthography was due to the fact that we deem μπ, ντ more suitable representations of the sounds [mb], [nd] than the original μβ, νδ, because the pronunciation of β, δ in all other conditions was not [b], [d] (CHAT02, pp. 440-441). As far as the distinction between μβ, νδ, γγ and μπ, ντ, γκ is concerned, its cause is uncertain: the pairs [mv]/[mβ]-[mp], [nð]-[nt], [ŋɣ]-[ŋk], corresponding to the orthodox pronunciation of the individual letters of the former and the latter clusters, are sufficiently distinct to justify the lack of confusion, as much as the catholic or any other explanation. Furthermore, the reason why we have chosen μπ, ντ to represent the clusters [mb], [nd] cannot be that β, δ are fricatives in all other positions; π and τ also have a different pronunciation ([p], [t]) in all other places, but this has not prevented us from considering μπ and ντ suitable for representing [(m)b] and [(n)d]; perhaps the reason for the orthographic convention is that the voicing of π, τ after nasals is universal, whereas β and δ are not always plosives in the same position (e.g., σύμβολο|symbol=[ˈsimvolo], never †σύμπολο; συνδυασμός|combination=[sinðiaˈsmos], never †συντυασμός).
Chatzidakis, in turn, was merely trying to formalise the Master's observation that "even at the present day β is an explosive sound when following a nasal", claiming that "komvos could not have produced kombos, but an original v would have done away with the nasal" (BLAS90, pp. 109-110). To the fact that there is no linguistic necessity for the nasal to disappear before [v] bear witness forms like "environment" and Umwelt|environment[GER]; that the modern language is less tolerable to [mv] relates to changed phonotactics rather than changed pronunciation. To end this discussion, contrary to the catholic assertion that it is unexplainable under the orthodox model, the occasional pronunciation of μβ, νδ, γγ as [mb], [nd], [ŋg] can be readily explained under the orthodox (ΒO, ΔO, ΓO) or the alternative (Βalt, Δalt, Γalt) models as yet another case of dissimilation: the continuity present in a cluster comprising a nasal (which is inherently continuant) and a fricative or approximant is broken by converting the second one into the corresponding plosive; this simple explanation (backed by the parallel cases of dissimilation, φθ→φτ, χθ→χτ, σθ→στ, σχ→σκ, αυσ→αψ, ευσ→εψ, in modern Greek) is enough to end all illusions of unquestionable "preservation".
Having clarified that, it is interesting how this Linguistic Standard Delusion (LSD) came about and was propagated by the Catholics. Blass, who appears to be the first to mention this "evidence", based his conclusion on an observation made by Psichari (BLAS90, p. 110, n. 1), i.e., a Greek who, being a proponent of the vernacular language, was evidently trying to establish that the "purist" scholars had imposed an artificial form on the language (cf. also BLAS90, p. 85, n. 1).It is not clear whether the extent of the phenomena reported by Psichari were universal, i.e., whether μβ, νδ, γγ were always pronounced [mb], [nd], [ŋg] in all vernacular forms of the language. The current language comprises abundant exceptions (like the already mentioned σύμβολο, συνδυασμός, etc) that cannot all be attributed to scholarly influence or "spelling pronunciation". The ease with which this precarious conclusion of Blass' was adopted and augmented by other Greeks (Chatzidakis, etc) must be attributed to their eager anxiety to identify surviving ancient elements in the modern language (i.e., the opposite of Psichari's intention). To use the words of Jannaris (albeit on a different occasion), they "make it a patriotic or pleasant duty to bear testimony to the 'preservation' of [the ]sound in [Neohellenic or modern Greek], since they are told that it is 'classical'" (JANN97, p. 49). This is a recurrent theme, where a coincidence of sound with an (assumed) ancient model is hastily and without further confirmation declared a "preserved" pronunciation, as was the case in the "preservation" of the "e-sound" of η and in the "survival" of the affricate sound of ζ in Italian (even though it was not "classical"). Such sloppy assertions can only harm the reputation of those who make them and undermine their argumentation.
Having disqualified the occasional plosive pronunciation of post-nasal mediae in the modern language as evidence of a generalised stop nature of the mediae (irrespective of context) in ancient times, it should be noted that this does not necessarily prove the ancient pronunciation of μβ, νδ, γγ as [mv/mβ], [nð], [ŋɣ]. One cannot exclude the possibility that these clusters were pronounced as [mb], [nd], [ŋg] after nasals, at least dialectally, but there is no good evidence to support this thesis. For example, it is true that in most varieties of Spanish the mediae have plosive allophones after nasals, but to conclude that the same allophony characterised ancient Greek would be proof by analogy. Perhaps the only indication that μβ might have been pronounced [mb] (or [b]?) is the aforementioned use of the digraph among the alternative spellings of [b], as assumed by Chatzidakis. In any event, there appears to be no evidence that the plosive realisation of some post-nasal mediae in the modern language is a continuation of such a conditional pronunciation (instead of, e.g., another facet of dissimilation that is responsible for many post-classical innovations).
With the pronunciation of the mediae, we conclude the investigation of the consonants of "ancient Greek".There is still one open issue that I have not touched, namely gemination. Despite the uselessness of Allen's assertive proof on this matter, I do not wish to argue that doubly written consonants were a mere orthographic convention, e.g., for restoring metric regularity or for indicating an assimilated consonant. It seems that the ancient language had a different sense of rhythm as far as letter count is concerned, as suggested i.a. by compensatory lengthening (in the modern language, a deleted consonant leaves no trace in speech), so that a "lengthened" pronunciation of the graphically repeated consonants would be in line with this observation. In fact, I have often relied on the identification of gemination with lengthened pronunciation in my argumentation, e.g., for the pronunciation of ῤῥ and the effect of ζ on metre. The proposed (and substantiated) values (Βalt, Δalt, Γalt), together with the orthodox values of the aspiratae (ΘC, ΦC, ΧC), restore the phonological symmetry of the consonants of ancient Greek, a feature that is not (and cannot) be provided by the hodgepodge of the catholic model. Thus, the only important difference with the modern phonology seems to be the lack of distinction in terms of stricture ([+/-cont]) at the level of voice in the ancient language.As must have been evident from the discussion of the previous chapter, the orthodox model fails in another aspect of the consonants, namely the value of the δασεῖα. However, since I consider the most likely significance of the δασεῖα to be the presence of a property (like breathy voice or murmur) and not a consonant, I do not deem it a deficiency of the orthodox consonantal model. In keeping with the wording of the Grammarians, it is more correct to regard it as a difference at the prosodic level. The descriptions and terminology of the Grammarians make it sufficiently clear that the three-way distinction of each class of mutes (labials, dentals, velars) as ψιλά, μέσα and δασέα involved the gradual increase of "πνεῦμα", a term already demonstrated to refer to the airstream. This cannot be accommodated in the dubious catholic model that essentially follows the Armenian paradigm (characterised by the Voice-Onset Time feature that suggests a different ordering: voiced, tenuis, aspirated or μέσα, ψιλά, δασέα); instead, the teaching of the Grammarians matches nearly perfectly the triangular model of modern Spanish, which is based on stricture. At the same time, there is evidence of a preferred fricative pronunciation of the mediae, notably the use of <Γ> for representing [ŋ], the omission or wrong insertion of γ and the interchange of Δ and Ζ. As for the meagre arguments used in favour of the plosive pronunciation of the mediae, these are based on dogma, misrepresentation and hallucination.
The story of the voiced mutes in Greek may, thus, be "reconstructed" as follows:
Before leaving the... aphony of the consonants for the euphony of the vowels, it might be interesting to outline some thoughts on the third item above.
As already noted, the presence of a nasal before a voiceless plosive [k], [p], [t] induced voice into the latter, turning it into [g], [b], [d], respectively. It is not clear to what extend the nasal stood its ground; in the modern language its presence is phonologically irrelevant (as voiced plosives are to be found only in post-nasal position in the native vocabulary) and many speakers often omit it (making it possible that its retainment be the result of a "spelling pronunciation"); the fact that the digraphs <ΜΠ>, <ΝΤ>, <ΓΚ> were fairly early established as graphical representations of the voiced plosives (which were not part of the native phonology) may suggest that the same was the case in Koine or medieval Greek. This matter requires further investigation, but irrespective of the fate of the nasal the voicing of post-nasal plosives is not as uncommon as one might think, as it is also attested in other languages or dialects:
The first three cases relate to (post-classical) Latin and Romance. Where these independent developments or was there a latin influence on greek phonology (a greek influence on Latin can be probably ruled out, as it would not explain the regional developments)?
As far as the coptic evidence of the last case are concerned, it should first be emphasised that coptic <Β>, <Δ>, <Γ> most likely did not stand for voiced sounds: "It is a well-known fact that the explosives in Coptic are neither voiced nor aspirated. Therefore a Coptic ear was not accustomed to catch the difference between β and π, γ and κ, δ and τ. We may be certain that, in the general Coptic pronunciation of Greek words, there was no difference between β and π, γ and κ, δ and τ." (as stated by Girgis). Thus, coptic <Β> and <Π>, <Δ> and <Τ>, <Γ> and <Κ> should be regarded as equivalent when used in greek words. As a consequence, when it was desired to more precisely indicate the voiced nature of greek β, δ, γ (a feature absent from Egyptian), the combination of a nasal with a class-1 or class-2 stop was employed. Evidently, the reason is that post-nasal stops were voiced, but was this a feature of Greek adopted as a convention in Egyptian or a characteristic of the latter that spread into Greek?
I do not know how much we can rely on such evidence to draw conclusions about the origins or spread of post-nasal voicing. I believe, however, that such observations may serve as the trigger for a systematic investigation of the influences of one language's spelling conventions on the other's. We already know that the representation of the velar nasal in Gothic was also through G, not N (ALLE87, p. 35, n. 54) and Armenian and Coptic used the digraphs ու and ΟΥ for the sound [u]. In those cases, it is clear that the direction of transmission of the orthographic convention was from Greek to the other languages. However, in the case of nasal+voiceless stop for representing voiced stops, it is not clear whether its manifestations in the various languages are the result of any interaction, much less which the instigator language was. The issue is in need of further investigation and would make an interesting doctorate thesis.
|
|
|
|
|
