 |
 |
 |
|
 |
Ever since I started this investigation, this is the chapter I was looking forward to writing the most. The story of ζ's "reconstruction" is one of 19th-century phonological ignorance overtaking the center stage and, instead of being cast aside as unscientific, becoming the official narrative of Catholicism.
The present value of ζ is, as one would expect (from its reflections in English and French z):
ΖO: Ζ = [z]
This value was accepted by all until the mid-19th century, when the initiative passed to the german school. In the german phonological model, there was no room for an independent /z/ phoneme; [z] can only be an allophone of /s/. And this model was exactly the one promoted. We have seen how, despite the lack of serious evidence, the validity of (present-day) "voiced σ" was stretched to the classical times and beyond. For the letter ζ itself, it appears that the initial proposal was a voiced affricate ([d͡z]), in line with the Italian and (partly) German models; of course, the scholars of that era lacked the necessary linguistic sophistication (and tools) to denote the affricate as we do today and used the representation ds (more on the reasons later). In the face of contradicting evidence, their favourite value was replaced by a strange construct, the consonant cluster [zd]. Finally, the first inscriptional instances of ζ for "voiced σ" compelled them to consent to the "modern value" from the post-classical antiquity on, but without abandoning their previous theories. The official catholic point of view, at least as expressed by the 20th-century Popes (ALLE87, pp. 56-59; STUR20, pp. 189-191), appears to be a compromise of all three proposals:
ΖC: Ζ = [dz] ([d͡z]) → [zd] → [z]
The narrative accompanying this odd equation goes like this (quotes are from the above-cited passages of Allen and Sturtevant):
However, the time boundaries are more than hazy; the allegedly classical pronunciation [zd] is traced back to "primitive Greek" and even "Indo-European", but it was also referred to by the Grammarians in their descriptions, which "almost certainly reflect a grammatical tradition" or was "no doubt [...] for a time a peculiarity of the speech of the educated people" (note the certainty in both accounts); the affricate value [dz] (rather [d͡z]) "must have been current among the early Italian Greeks" and "has always survived on Italian soil", as it "seems also to have been known to late Latin speakers" and as it is used "in Italian in the value of dz (orzo) and ts (grazia, etc.)"; at the same time, it is considered that z was introduced in Latin with "the value of the Greek ζ [which] was a voiced fricative [z], as in English zeal", at "the time of this adoption (in the first century B.C.)", whereas "Before the adoption of the foreign sound and letter, the Greek ζ had been rendered by its nearest Latin equivalent, viz. by the voiceless s initially and ss medialy, e.g. sona (Plaut.), Setus (inscr.), massa = ζώνη, Ζῆθος, μάζα" (ALLE78, p. 46). In other words, the value of ζ is variably assumed to be any of the three proposed, either before or after the period of its alleged preponderance, as the Catholics see fit for substantiating their contradictory theories or for explaining away inconvenient evidence. This catholic monstrosity, which should have been rejected right away by any reasonable person, is allegedly "backed" by evidence that will be scrutinised herein below.
We will first examine the evidence that are used against the traditional value ΖO. I will show that, contrary to the catholic assertion, the present pronunciation (at least as far as its quality is concerned) is compatible therewith and victim of misinterpretation or misuse of the evidence put forward by the Catholics. Thus, in accordance with the Presumed-Innocent Principle, the orthodox value remains (potentially) valid and is not ruled out by the provided evidence.
I have divided the evidence in three categories, according to evidential value.
The main reasons why the Catholics sought a different value for ζ are the facts that, in most cases, ζ makes position in poetry (i.e., a preceding syllable comprising a single "short" vowel is treated as "long by position") and that the Grammarians define it as one of the "double consonants" and prescribe the existence of a σ and a δ in ζ. From the former they conclude that ζ was equivalent to consonant clusters, which also make position, and from the latter that it corresponds to the cluster σδ or δσ. This is a straightforward interpretation of the two facts, but after these are placed under scrutiny this conclusion turns out to be a rather naive one.
In order to make use of Ζ1, we need to first understand how ancient metre worked. However, every theory that tried to explain metre has been shown to entail some loose ends which are irreconcilable with the evidence. In view of this, Ζ1 is nothing more than yet another rule of the metrical Sudoku, a convention of unknown origin that had to be respected by the creators of metric puzzles. Nevertheless, the Catholics interpret Ζ1 as indicating that ζ was a consonant cluster, obviously because consonant clusters are known to render the preceding syllable "long by position". However, there are three main ways for a single "short" vowel to form a syllable "long by position", namely when it is followed by A) a consonant cluster, such as κτ, πρ; B) a doubled (presumably geminated) consonant, such as λλ, μμ; or C) a "double" consonant, i.e., ξ, ψ, ζ. Nothing necessitates that all three cases should involve the same kind of sounds; in fact, apart from comprising two (or more) graphemes, the first two types (A and B) are clearly different. The identification of the "doubles" with consonant clusters is, no doubt, based on the present values of ξ(=[ks]) and ψ(=[ps]), which are commonly accepted by Orthodox and Catholics as their perennial values; however, we have seen that not only is the possibility of a divergent development of the clusters /ks/ and /ps/ in the various greek dialects not to be excluded, but a monophthongal (or, at most, geminated) value answers the question of origin of the graphemes Ξ, Ψ more satisfactorily and renders the descriptions of the Grammarians more understandable.It may also be considered that the very name "διπλά" (doubles) speaks against a consonant-cluster value. As pointed out by Theodosius, "Εἰ γὰρ δῶμεν ἀληθῆ εἶναι τὸν λόγον τὸν φάσκοντα πρὸ τῆς εὑρέσεως αὐτῶν διὰ δύο συμφώνων ἀποτελεῖσθαι τὰς φωνὰς τούτων, ἐξ ὧν καὶ συνεστήκασιν, ἔδει σύνθετα ὀνομάζεσθαι καὶ οὐ διπλᾶ· ὧσπερ καὶ τὰ ἐκ δύο λέξεων συγκείμενα ὀνόματα σύνθετα καλοῦμεν|Because if we accept the opinion saying that, before these were discovered, their sounds comprised two consonants, of which they were composed, they should be called composites and not doubles; exactly as we name composites the nouns that are composed of two words". Had they been consonant clusters (or "consonant groups", as Allen calls them), the name "σύνθετα" (composites) would be more suitable. The applied term "διπλᾶ" (doubles), though, is rather suggestive of a letter with a uniform value and (considered) equivalent to two identical letters, either in pronunciation (i.e., geminated or long consonants, as suggested by Theodosius and the Scholiast of Thrax) or convention (in terms of metrical pattern). At any rate, even if two of the "doubles" were consonant clusters, it does not necessarily mean that the third one had to also be one; their grouping together was evidently incited by the fact they were all single graphemes that (metrically) behaved like the double graphemes (clusters or doubled). Even if one is convinced that only consonantal sequences can make position, one does not need to equate the "doubles" with category A above, but they could also correspond to category B, namely they could be geminated consonants;This is actually the (already mentioned) conclusion drawn by the Scholiast of Thrax (BEKK16, p. 823). Although his "proof" relies on arbitrary axioms (e.g., that the letters or "στοιχεῖα|elements" are indivisible), I do not see why it would be less rigorous than the assumption that only consonant clusters can make position. and even if we accept that the "doubles" ξ, ψ, ζ made position because they were equivalent to consonant clusters, this equivalency might have been merely a tradition relating to their origin from /ks/, /ps/ and *dj, *gj, etc, since we do not know whether the prosodic rules set out by the Grammarians related to their current pronunciation, an older one or an artificial one.For example, it is possible that, when the Homeric poems were composed, the related sound was still *dj or *gj (possibly reflecting or using an even older pronunciation), but by the time they were written down (presumably VI BC) these clusters had already been converted to ζ (whatever this stood for), thus creating the associated tradition. Such hypothesis might even explain the exceptions to Ζ1, such as Ζέλεια and Ζάκυνθος (BLAS90, p. 119), which being of foreign origin (the former a non-Greek city, the latter comprising the typical pre-Greek ending -νθος) could have comprised the sound (later) represented by ζ in their original form.
Thus, Ζ1 may mean that ζ was:
Of these three possibilities, the Catholics promote only the second, obviously because it corresponds to their metric theories for syllable division and it complies with their proposed values [dz] and [zd]. However, it should be noted that the first proposed value, being actually an affricate [d͡z] (if it comes, as is argued, from a "palatalisation" - or rather assibilation - of d and g in original *dj and *gj), cannot be considered a consonant cluster (this is evident from many languages that possess the affricate [d͡z] and represent it either with one letter or with a digraph considered as a single letter); in other words, once an affricate is "split" into a stop and a sibilant (in order to match the syllable-division rules set out by Allen & co.) and turned into a sequence of consonants, it ceases to be an affricate. The second value is clearly a consonant cluster, but fails to explain the necessity for or the reason behind the use of a single letter for the cluster σδ and not, e.g., for στ, σβ, σγ, etc; moreover, it is even more incomprehensible why the Greeks chose to graphically join these two letters, when they were metrically (and one would presume also phonologically) separated (in different syllables).
Hence, the first and third possibilities mentioned above, together with a uniform value for ζ, either simple (first case) or geminated (third case), are at least equally valid or even more likely, since they do not entail such contradictions or paradox.
This concerns comments by Archinus (via Theophrastus and Syrianus/Alexander), Aristotle, Thrax, Halicarnasseus and Sextus. None of them describes ζ as equal to σδ or δσ, in the sense that one might have been used in lieu of the other or that one has replaced the other or that ζ is just a shorthand used for σδ or δσ. It is important to note their exact wording.
The most laconic and probably less descriptive account comes from Thrax: "διπλᾶ δὲ εἴρηται ὅτι ἕν ἕκαστον αὐτῶν ἐκ δύο συμφώνων σύγκειται, τὸ μὲν ζ ἐκ τοῦ σ καὶ δ|they are called double because each one of them is composed of two consonants, ζ of σ and δ [Stu's translation]" (STUR20, p. 188). Faithful to their principles of lingo-math, some Catholics interpret this passage as describing the equation ζ=σ+δ (e.g., STUR20, p. 189: "the passages cited above from Dionysius Thrax and Dionysius of Halicarnassus declare that ζ had the sound of σ+δ (i.e., zd)"). However, ζ is merely described as being composed of a σ and a δ without defining what this "composition" involves. Thus, strictly speaking, Thrax only tells us that ζ=f(σ,δ), f(·,·) being a non-defined function.
The Catholics draw the same conclusion (cf. the preceding quote from Sturtevant) from the description of Halicarnasseus: "διπλᾶ δὲ λέγουσιν αὐτὰ ἤτοι διὰ τὸ σύνθετα εἶναι, τὸ μὲν ζ διὰ τοῦ σ και δ, τὸ δὲ ξ διὰ τοῦ κ και σ, τὸ δὲ ψ διὰ τοῦ π και σ συνεφθαρμένων ἀλλήλοις ἰδίαν φωνὴν λαμβάνοντα, ἤ διὰ τὸ χώραν ἐπέχειν δυεῖν γραμμάτων ἐν ταῖς συλλαβαῖς παραλαμβανόμενον ἕκαστον|they are called double either because they are composite, receiving a distinctive sound through the coalescence respectively of σ and δ into ζ, of κ and σ into ξ, and of π and σ into ψ; or because they occupy the room of two letters in the syllables where they are found [Stu's translation]" (STUR20, p. 188). The passage would possibly afford such a conclusion, had it ended right after "τὸ μὲν ζ διὰ τοῦ σ και δ|ζ through σ and δ" (or rather after "τὸ δὲ ψ διὰ τοῦ π και σ|ψ through π and σ"); however, reading the entire passage reveals that the author's intention is to describe a phenomenon different from mere juxtaposition, in particular that the two sounds (of σ and δ for ζ, of κ and σ for ξ, of π and σ for ψ) are "worn together in each other" (this is the most accurate translation of the verb συμφθείρω that I can come up with) or, in the words of Sturtevant, coalesce to produce their own sound, evidently a distinct one, as Sturtevant translates. A value ζ=σδ or ζ=δσ is not compliant with this definition. What kind of value would comply? I can think of no better parallel than my favourite High-German [ʃ], which was produced by coalescence of the two constituents (s and k) of West-Germanic /sk/; taking the analogy of this example to the case of ζ originating from σδ (if one accepts the etymologies ὄζος from ὄσδος, ἵζω from ἵσδω, βύζην from βύσδην and even Ἀθήναζε from Ἀθήνασδε, etc; cf. BLAS90, pp. 118-119), the description of Halicarnasseus might as well refer to σ and δ of the "original" σδ (but not δσ, since Greek was intolerant to dental/sibilant sequences) coalescing to, e.g., the present sound [z]. The description of Halicarnasseus is, thus, not only against the catholic proposal, but it is probably suggestive of a uniform sound for ζ (also for ξ and ψ, as we have seen).
There is also another part in the description of Halicarnasseus that deserves attention: besides the "coalescence" theory, an alternative explanation for the term "doubles" is also put forward, which merely refers to fact Ζ1, namely that they occupy the place of two letters. The first significance of this explanation is that the consideration of ζ, ξ, ψ as "doubles" may be a mere technicality. The second is the apparent confusion or at least uncertainty of the ancients with respect to the true nature of the "doubles" and particularly ζ; for, if it were clear that ζ=σδ or ζ=δσ, no further explanation would be necessary or even justified. It, therefore, appears that not even the Grammarians were certain about the theory that the "doubles" are composite (be they either sequences or mixtures of) letters.
This uncertainty is even clearer in the accounts of Aristotle: "οἱ μὲν γὰρ τὸ ζα ἐκ τοῦ σ καὶ δ καὶ α φασὶν εἶναι, οἱ δέ τινες ἕτερον φθόγγον φασὶν εἶναι καὶ οὐθένα τῶν γνωρίμων|for some say that ζα is composed of σ, δ and α, while others say that it is a distinct sound and not any one of those which are familiar to us [online translation]" (Metaphysics, 993a)Allen and Sturtevant allude (ALLE87, p. 164; STUR20, p. 191) that the manuscripts have "σμα, μ" and that the quotation has been "Restored after the commentary of Alexander Aphrodisiensis"; in that case, it appears that the quote is a conjectured emendation and not a fact. and of Sextus: "λέγουσι δέ τινα τῶν συμφώνων καὶ διπλᾶ, καθάπερ τὸ ζ, ξ, ψ· συνεστηκέναι γάρ φασι τὸ μὲν ζ ἐκ τοῦ σ καὶ δ, τὸ δὲ ξ ἐκ τοῦ κ καὶ σ, τὸ δὲ ψ ἐκ τοῦ π καὶ σ|they also call some of the consonants doubles, like ζ, ξ, ψ; because they say that ζ is composed of σ and δ, ξ of κ and σ and ψ of π and σ". Note that they both use the term "φασί(ν)" (they say/it is said), implying that they do not themselves share this point of view. In other words, from the Classical age (Aristotle) to the late Roman age (Sextus), the view that there is a σ and a δ in ζ is treated as a mere theory, rumour or allegation. Had ζ been pronounced the same as σδ (as the Catholics want us to believe), there should have been no doubt about this theory, as the two elements would be readily identifiable in such a consonant cluster. It is also improbable that the confusion related to two different pronunciations (the "reconstructed" [zd] and the "modern" [z]) used by different parts of the population, as implied by Allen and Sturtevant (ALLE87, p. 58; STUR20, pp. 190-191); in that case, it would be reported that some people pronounce ζ differently, not that they say/claim that it is composed of σ and δ.
Aristotle also gives us another account, which (in true... Aristotelian fashion) is so... metaphysical that no solid conclusions can be drawn: "ἐπεὶ καὶ τὸ ΞΨΖ συμφωνίας φασὶν εἶναι, καὶ ὅτι ἐκεῖναι τρεῖς, καὶ ταῦτα τρία: ὅτι δὲ μυρία ἂν εἴη τοιαῦτα, οὐθὲν μέλει (τῷ γὰρ Γ καὶ Ρ εἴη ἂν ἓν σημεῖον): εἰ δ᾽ ὅτι διπλάσιον τῶν ἄλλων ἕκαστον, ἄλλο δ᾽ οὔ, αἴτιον δ᾽ ὅτι τριῶν ὄντων τόπων ἓν ἐφ᾽ ἑκάστου ἐπιφέρεται τῷ σίγμα, διὰ τοῦτο τρία μόνον ἐστὶν ἀλλ᾽ οὐχ ὅτι αἱ συμφωνίαι τρεῖς, ἐπεὶ πλείους γε αἱ συμφωνίαι, ἐνταῦθα δ᾽ οὐκέτι δύναται|Indeed, they assert also that Ξ, Ψ and Ζ are concords, and that because there are three concords, there are three double consonants. They ignore the fact that there might be thousands of double consonants—because there might be one symbol for ΓΡ. But if they say that each of these letters is double any of the others, whereas no other is, and that the reason is that there are three regions of the mouth, and that one consonant is combined with σ in each region, it is for this reason that there are only three double consonants, and not because there are three concords—because there are really more than three; but there cannot be more than three double consonants [online translation]" (Metaphysics, 1093a). The example of "τῷ γὰρ Γ καὶ Ρ εἴη ἂν ἓν σημεῖον" (lit.: "because to Γ and Ρ may be one sign/point") might be construed as indicating that Ξ, Ψ and Ζ are similar to ΓΡ, namely consonant clusters. However, the wording is, to say the least, weird: why "Γ καὶ Ρ" and not "ΓΡ"? why only one example? why this particular example? how does it relate to ΞΨΖ? why "σημεῖον" (sign/point) and not the usual "στοιχεῖον" (element/letter)? does the strange syntax with dative mean that a single sign/symbol could be used for "Γ καὶ Ρ"/"ΓΡ" or that a single point (which is another meaning of "σημεῖον") may represent Γ and Ρ, in the same way that Ζ represents Σ and Δ (see below)? Although the latter explanation appears more reasonable to me,Note that there is a connection between greek γ ([ɣ], as now pronounced and as appears possible for the Antiquity) and ρ ([r]
It appears that Aristotle's Scholiast generally adopts this interpretation and cites Archinus to his defence ("ἀλλὰ καὶ Ἀρχῖνος ταύτῃ τῇ ἀποδόσει ἐχρῆτο|but Archinus, too, was making use of this same explanation"): "ἔλεγε γὰρ Ἀρχῖνος ἤ ἔξω τι παρὰ τὴν μύσιν τῶν χειλέων ἐκφωνεῖσθαι, ὧσπερ το π, καὶ διὰ τοῦτο τὸ ψ προς τῷ ἄκρῷ γεννᾶσθαι τῆς γλώττης ὡς ἐκ τοῦ π καὶ σ συγκείμενον· ἤ τῷ πλάτει τῆς γλώττης παρὰ τοὺς ὀδόντας, ὥσπερ τὸ δ, καὶ διὰ τοῦτο τὸ ζ κατὰ ταύτην γεννᾶσθαι τὴν χώραν, ἤ τῷ κυρτῷ [τῆς γλώττης] πιεζομένῳ ἐκ τοῦ ἐσχάτου, ὥσπερ το κ, ὅθεν το ξ προϊέναι|because Archinus was saying that either a sound is pronounced somewhat outside near the closing of the lips, just like π, and for this reason ψ is produced near the tip of the tongue as being comprised of π and σ; or with the blade [lit. breadth] of the tongue near the teeth, just like δ, and for this reason ζ is produced in this region; or with the arched tongue pressed upon from the back [of the mouth], just like κ, whence comes ξ [based on Stu's translation of the similar text of Syrianus]". This account that emanates from the Classical Age (Archinus lived in V BC, Theophrastus that reports it lived in IV BC) leaves no room for interpretation of ζ as any sequence of σ and δ (i.e., neither σδ nor δσ), but rather clearly states that it is a sound produced near the teeth, using the whole width ("πλάτει") of the tongue, like δ (not after or before δ). Such a description is not at all appropriate for the catholic [zd] or [dz] and rather characteristic of the orthodox [z], particularly if one is not prejudiced by the catholic assignment δ=[d], but considers the orthodox value δ=[ð]; for only [z] sounds like [ð] ("ὥσπερ τὸ δ").To illustrate this, I have devised the following imaginary dialogue between a French (F) and a German (G) scholar:
F: Zis guy says zat our z sounds like ze greek delta. What do you sink?
G: I sink zat zere is absolutely no connection between your z and zeir delta or, as zey say, "zelta". This last remark may be the key to understanding the whole theory behind the association of ζ with σ and δ.
From the evidence examined so far, no Grammarian describes ζ as shorthand for σδ (or δσ). There is a single comment that almost clearly refers to ζ as a sequence of σ and δ (i.e., σδ); Blass alleges (BLAS90, p. 115, n. 5) that it comes from Sextus, but it is not to be found in the cited page(s) nor in any other part of that author's work.Chatzidakis also refers (CHAT02, p. 446) to the same passage citing page 622 of Sextus, but since no such statement can be found on the cited (or any other) page, he obviously copied from Kühner-Blass. It is also worth noting that, in his own book (BLAS90, 116), Blass cites page 662 of Sextus; but neither on that page is the statement to be found. I have only been able to find it in Bekker's endnotes: "Ἄξιον δὲ ζητῆσαι διὰ τί τριῶν ὄντων διπλῶν, τῶν ζ, ξ, ψ φημί, τὸ μὲν ξ καὶ ψ τελικὰ τῶν ἀρσενικῶν ὀνομάτων εἰσί, τὸ δὲ ζ οὐδαμῶς. καὶ ἔστιν εἰπεῖν ὄτι διὰ τοῦτο τὸ ζ οὐκ ἔστιν τελικὸν τῶν ἀρσενικῶν ὀνομάτων, διότι ἐκ τοῦ σ καὶ δ δοκεῖ συγκεῖσθαι, οὐδέποτε δὲ λέξις Ἑλληνικὴ εἰς ἄφωνον καταλήγει|It is worth examining for which reason, while there are three doubles (I am talking about ζ, ξ, ψ), ξ and ψ are finals in the masculine nouns, but ζ not at all; and this is the reason why ζ is not final in masculine nouns, because it is thought to be composed of σ and δ and never does a greek word end in a mute". Here, too, the testimony that ζ is composed of σ and δ, with the mute (i.e., δ) at the end, is provided by means of the impersonal verb "δοκεῖ", which contains an element of subjectivity ("it is thought") or uncertainty ("it seems"). While cute, this explanation is a theoretical attempt (by the mysterious author) to provide an answer to an artificial problem, namely why ζ does not serve as word final, while the other doubles, ξ and ψ, do, by means of an irrelevant fact, namely that greek nouns do not end in a mute.The problem is artificial, because nothing necessitates that all doubles should serve as noun finals. As pointed out by Theodosius, the letters of the same kind do not need to share the same properties, e.g., of the "ἀμετάβολα|invariable", only ν and ρ may be finals, while μ and λ never are. This theory, despite being improbable (for we know from etymology that, in the vast majority of cases, the origin of ζ is not σδ, but voiced mute+/j/ or only /j/), might bear some significance, if the authority that put it forward is sufficiently old, namely earlier than the alleged date of "transition to the modern pronunciation" in late IV BC. However, even Blass (who is the only Catholic mentioning it) ascribes it to Sextus of late II AD, something that is additionally not verifiable.
The only similar passage I have been able to locate, comes from Theodosius, an even later source (III AD), who submits with confidence that "Τὸ δὲ Ζ ἐν τῇ ἐκφωνήσει αὐτοῦ οὐ τὸ Σ ἔχει τελευταῖον ἀλλὰ τὸ Δ. Ἐπειδὴ δὲ οὐδέποτε ἑλληνικὴ λέξις εἰς ἄφωνον καταλήγει· διὰ τοῦτο ἀδύνατον ἐγένετο εὑρεθῆναι τὸ Ζ ληκτικὸν ὀνόματος|On the other hand, Ζ in its pronunciation does not have Σ as last, but Δ. And a greek word never ends in a mute, for this reason it became impossible to find Ζ as final of a noun". However, this work of Theodosius is indicative of the confusion of the Grammarians (particularly of late Antiquity) with respect to the nature and value of Ζ.Note that the pronunciation ζ=[z] can be proven with near certainty to have been established long (at least six centuries) before the time of Theodosius in III AD, by means of transliterations in early Latin, Indian, Gothic (for later times), etc, which never use anything similar to zd for transcribing the greek ζ (STUR20, p. 191). Indeed, having assertively proclaimed that δ is heard last when pronouncing ζ, he later considers that σμύρνα, σμίλιον should be spelt with σ, not ζ; had ζ been σδ, as he claimed before, there would be no reason to use it in these cases, inserting a δ where there is none.The real reason is subsequently provided by Theodosius: "τὰ γὰρ διπλᾶ οὐδέποτε προτάσσονται [συμφώνων]|because the doubles are never placed before [other consonants]". For ξ and ψ this is pure chance; these are products of declension, wherein a velar or a labial happen to occur before an ending starting with σ, no endings comprising another consonant after their σ. In the case of ζ, the cause is etymology, as ζ is almost always the product of /j/ acting on a preceding voiced mute (or of /j/ alone). /j/ in Greek is a non-syllabic ι (/i/), which occurs only when the following vowel is the core of the syllable. If a consonant succeeds, then the ι has to be syllabic. This may also explain the lack of confusion between voiced σ (if indeed σ was voiced before μ, β, and/or γ) and ζ (both [z]) in early inscriptions by means of a simple orthographic rule: [z] before consonants is spelt with σ, before vowels with ζ (the certainly curious transcription "Ἔσδρας" for Ezra in the Septuagint is most likely due to such an orthographic rule).
Contrary to this curious suggestion of Sextus (late II AD) and Theodosius (III AD), the somewhat earlier Latin grammarian Velius Longus (early II AD) vehemently rejects the theory that z=sd/ds and incontrovertibly teaches (the text is quoted in STUR20, pp. 116-117 and partly in CHAT02, p. 447) that ζ (and z) was pronounced with a uniform sound from start to end when prolonged; he explicitly states that ἀζηχής does not sound like ἀδσηχής (ἀσδηχής in the version of Chatzidakis), but ἀζζηχής does sound like ἀσσηχής. Here, there can be little doubt that he is describing a pronunciation ζ=z=[z]. It is also interesting to mention the theory which he discards, namely the quoted statement of Verrius Flaccus "z litteram per sd scribi ab iis qui putant illam ex s et d constare, ut sine dubio muta finiatur|z is written with sd by those who think that it consists of s and d, so that it undoubtedly ends in a mute [Stu's translation]" (emphasis mine; note the use of "putant|they think/suppose/ponder/consider and "sine dubio|undoubtedly" in the same sentence), Flaccus concluding that z ends in a mute because of an orthographic convention used by those who think that ζ is composed of σ and δ! The (lack of) merit of this clumsy "proof" is even more evident, if it is juxtaposed to Longus' more rigorous method of observing the... actual pronunciation. The question, of course, is "who are 'those' who think that ζ consists of σ and δ and write it with σδ?" From what we know (and from Longus' subsequent statement "Dores enim scimus dicere μελίσδειν, et ipsum παίζειν apud alios dicitur παίσδειν|For we know that the Dorians say μελίσδειν, and even παίζειν is spoken as παίσδειν in certain places [Stu's translation]"), Flaccus can refer to no other but some Aeolians and Dorians, who reportedly used σδ in lieu of ζ. Now, if this were true (see below), it would only mean that ζ=σδ dialectally, almost certainly not in Attic or Koine. But let us investigate this further.
Indeed, there is a tradition according to which Aeolians and Dorians wrote σδ for common-greek ζ (e.g., as alleged by Longus above, but also by others). However, the only occurrences of such spellings is "in the texts of Lesbian poetry", but "they are not found in early Lesbian inscriptions", which led Allen to conclude that "These spellings almost certainly represent a later editing" (ALLE87, p. 59, emphasis mine). It appears that this was also a feature of "some other poets as Alcman and Theocritus" (BLAS90, p. 120), i.e., authors writing in Doric, which seems to confirm the dorian practice reported by Longus. However, as Blass further remarks, "this is only in books, not on inscriptions". What inscriptions actually do show, in some dorian and aeolian varieties, such as Cretan, Laconian, Elean and Boeotian, is the use of δ or δδ for ζ and vice versa: "Οἱ Κρῆτες, οἱ Λάκωνες, οἱ Ἠλεῖοι, οἱ Βοιωτοὶ κλπ. μετεχειρίζοντο δ ἐν ἀρχῇ καὶ δδ ἐν μέσῳ λέξεως πράδδω, δουλίδδω, Δεύς [...] Καὶ πάλιν οἱ Ἠλεῖοι γράφουσι ζ ἀντὶ δ, ζίκαια, ζίφιον, καζαλήμενον|The Cretans, the Laconians, the Eleans, the Boeotians etc. made use of δ in the beginning and δδ in the middle of a word πράδδω, δουλίδδω, Δεύς [...] And yet again the Eleans write ζ instead of δ, ζίκαια, ζίφιον, καζαλήμενον" (CHAT02, p. 444). Here we may have a possible solution of the riddle.
The aforementioned interchange of δ and ζ in some Aeolian and Doric dialects is most likely an indication of (the beginning of) a phonemic merger between δ and ζ.Some scholars have interpreted this as indication of a mere convergence, yet not identity of sound (e.g., CHAT02, p. 441: "Τοῦτο φαίνεται δηλοῦν ὅτι το δ εἶχε καταστῆ ἐν Ἠλείᾳ διαρκής πνευματώδης φθόγγος, ἀλλ' ἄδηλον ἄν τοιοῦτος οἷος ὁ ἡμέτερος δ|This appears to denote that δ had become a spirantic continuant sound in Elea, but it is not known if it was similar to our δ"); in other words, they assume that δ was confused with ζ because they had become [ð] and [z], respectively (I cannot imagine what other "spirantic sound" Chatzidakis might have in mind for δ). However, as correctly pointed out by Méndez Dosuna: "les échanges graphiques ne résultent pas de la seule ressemblance de deux sons. [...] En effet, de telles confusions présupposent la neutralisation de l'opposition|the graphical exchanges do not result from the mere resemblance of two sounds. [...] In fact, such confusions presuppose the neutralisation of the opposition [first emphasis in the original, second mine]"; in other words, had δ and ζ the distinct values of [ð] and [z], there would be no reason for native speakers to confuse the two (the corresponding graphemes are not confused in languages that possess those sounds, such as English and modern Greek), but the confusion can only be the result of uncertainty about which of two graphemes representing the same sound should be used. This general tendency of Aeolian and Doric, which would involve a shift of ζ(=[z]) towards δ=([ð]) or of δ towards ζ or of both towards an intermediate sound, might have not been completed in some dialects, like Lesbian and/or Laconian, and would have left either or both of the sounds in limbo, but still distinct.The Catholics have attempted to accommodate the merger with their assumed model of δ=[d] and ζ=[zd] or ζ=[dz]; however, for an explanation abiding to that model one would necessarily have to assume that the sibilant suddenly disappeared or turned into a stop or that [zd] (re-)metathesised to [dz], which was again an affricate (violating the general rule of loss of dental mutes before sibilants) and then transitioned to [d] (for the use of δ for ζ) or that δ(=[d]) suddenly developed a sibilant, but only partly (for the use of ζ for δ), or some similar voodoo. Instead, there is no need to unnecessarily complicate matters, since the orthodox model provides a simpler and more cogent explanation. Now, assuming that lesbian-et-al ζ was somewhere "between" [z] (the general-Greek value of ζ, from at least the hellenistic age on) and [ð] (the assumed, under the orthodox model, general-Greek value of δ), the Lesbians et al would have no reason to replace the grapheme ζ by δ, because those two would still represent distinct phonemes, which would explain the absence of σδ from inscriptions; on the other hand, an editor who was a Koine speaker and wanted to graphically represent the peculiarity of lesbian-et-al ζ (with respect to common-Greek ζ=[z]) would have to devise an appropriate symbol for this "intermediate" sound and the digraph σδ would provide a satisfactory approximation.A modern parallel would be the rendering of Cretan κ in κι, κε ([t͡ɕi], [t͡ɕe]), which in general Greek are pronounced with a voiceless palatal stop ([ci], [ce]). A native Cretan writing for his compatriots would not need to use a different grapheme, because they would pronounce the κι, κε the Cretan way. However, a Greek editor wanting to emphasise the particular cretan "flavour" as opposed to the general-Greek pronunciation would resort to the common rendering τσι, τσε, even though this would be a mere approximation (pronounced [t͡si], [t͡se]). Even worse, the average Greek reader, upon encountering τσ for κ in the script, would most likely conclude that κ in Cretan is a double letter before front vowels. Thus, this orthographic convention, which was (in Allen's words) "almost certainly [...] a later editing", might account for the conviction of Flaccus, Theodosius, etc that ζ ended in a mute and for the awkward theory that ζ was a shorthand for σδ, yet not always (i.e., not in προσδοκία, ὅσδε, etc).Siamakis has an alternative explanation that might at first sound reasonable: "πρόκειται γιὰ τὴν ἀρχαιότερη ἀποδεδειγμένη χρῆσι προσδιοριστικοῦ γράμματος στὴν παγκόσμια ἱστορία|it is the oldest proven [!] use of a determinative letter in world history" (SIAM88, p. 459, §2,471). In other words, δ in σδ was used only as an indication that σ is to be voiced and was not meant to be pronounced, similarly to h in italian chi, che, which is not pronounced but merely indicates that these syllables should be pronounced [ki], [ke] or rather [ci], [ce] and not [t͡ʃi], [t͡ʃe], which is the pronunciation of ci, ce. As reason for this convention he considers that common-greek ζ was a voiced affricate (he uses the modern spelling τζ, i.e. [d͡z]), whereas in aeolian it became a plain sibilant [z], whence the modern value. Nevertheless, he disregards a critical point: the spelling σδ was not used by the Aeolians themselves (i.e., σδ for ζ is not found in aeolian classical inscriptions), but by the other Greeks in editions of aeolian (and doric) texts. If post-classical non-Aeolians used σδ as a convention for [z(ː)], this would mean that post-classical common-Greek ζ was not [z(ː)], which is against the evidence (e.g., the spellings ζμ for σμ). Siamakis' conjecture is only plausible if we assume an affricate pronunciation of ζ by the majority of the Greeks until a very late date. He also cites i.a. the spellings Ἀσδρούβας, Ἔσδρας (wherein σδ represents semitic ז, presumably [z]) as proof of this convention; however, these may simply be attempts to render the strange (from the perspective of greek phonology/phonotactics) cluster זר(=/zr/), or a phonetic development of [zr] (similar to mn→mbr in Spanish), or due to an epenthetic δ marking the transition from [z] to [r] (similar to π in πέμπτος and p in empty).
To sum it up, after a closer look at the relevant passages, no description can be found that ζ can be used for σδ (or δσ) and vice versa; had that been the case, some Grammarians would have discussed the cases where σδ is the only orthographic variant (exclusively in compounds, such as προσδοκία). The earliest testimony, that of Archinus (if reported correctly by Theophrastus via Syrianus/Alexander), is clearly not compatible with the catholic model, as he argues that ζ is produced near the teeth like δ, not comprising/including δ. Shortly thereafter come two excerpts of Aristotle's, of which the first is a conjectured "restoration" of the manuscripts that do not talk about ζα, but about σμα (in that respect, they could merely relate to the already mentioned voicing of σ before μ); at any rate, even the "emended" version is considered by Allen and Sturtevant to be a probable indication of ζ=[z]; the second one is too ambiguous even for Aristotelian standards, but sense can only be made if we interpret it as referring to σ being pronounced at three different places of articulation, which is consistent with the orthodox value but not with the catholic ones. Then, we have the late-Hellenistic/early-Roman Grammarians (Thrax, Halicarnasseus), who at most teach that there is a σ and a δ in ζ; this is most likely a mere theory developed under the need to explain Ζ1: if one needs to analyse ζ=[z] into two components (for we know that two consecutive consonants make position), then σ=[s] and δ=[ð] are the most reasonable choices, since [z] combines the sibilant nature of [s] with the voice of [ð] (which is the voiced consonant with the closest place of articulation; the theory also works with δ=[d]). Finally, some late-Roman comments (Flaccus via Longus, Sextus? via Bekker, Theodosius) that may initially appear to confirm the identity ζ=σδ, are too late into the post-III BC era (which the Catholics virtually consider as the terminus ante quem for ζ=[zd]) to be taken seriously and are obviously induced by the orthographic convention of using σδ for ζ in later editions of lesbian-et-al texts, a convention that probably has its origins in the shifted pronunciation of ζ in those dialect, in accordance with the general tendency of Aeolian and Doric.
All in all, the confused accounts of the ancient authors with respect to the real nature of ζ seem to have been caused by metrical considerations (Ζ1) and by literal interpretation of the lesbian orthographic convention.
After having convinced themselves that ζ was [dz] or [zd], the Catholics have started seeing signs. Thus, a number of transliterations, both external (with other languages) and internal (within Greek) are used as evidence for the value ζ=σδ=[zd]; the present pronunciation of z in Italian is used as evidence of "preservation" of the pronunciation ζ=[dz]; a (probably medieval) pun on the name of Zeus is also believed to be explained by the value ζ=σδ. All these bear as much (actually, even less) significance, as Ḫšayāršā↔Ξέρξης, Takṣaçilā↔Τάξιλα do in the "proof" that ξ=[ʃ], and as does the argument of "preservation" of this value in Spanish x.
Here, "some" implies that there are numerous examples; nevertheless, one can only find the following three invariably recurring in the various catholic treatises (STUR20, p. 189; BLAS90, p. 118; ALLE87, p. 58):
The main problems associated with this kind of evidence have already been outlined and the first two have already been discussed. Let us start from the actual facts, not from the numerous assumptions hidden in the catholic arguments. Well, the fact is that what we actually have are the greek forms (Ὠρομάζης, Ἀρτάοζος/Ἀρτάβαζος, Ἄζωτος) in the (post-christian) manuscripts of the works of some classical authors, like Herodotus, Plato and Xenophon. We do not know whether these were the exact spellings used by the ancient authors and we only assume that they refer to these foreign names, since the two (i.e., the greek and the foreign forms) are not found side by side. But even if we overlook this inconvenient detail, there are a number of question that need to be answered before these transcription can be considered to be of any evidential value.
What was the "source" form?
Certainly, it was not A(h)uramazda, Artavazda, Ashdod, as neither was Latin spoken nor was the latin alphabet in use in the Middle East of V BC. The first two are clearly old-persian names and the last one is a semitic form, but probably of non-semitic origin (being a Philistine city). We know them either from later Greek sources, or from "deciphered" cuneiform inscriptions in Assyrian or Persian, or from documents written in the semitic alphabet, such as the Bible, inscriptions, etc. It is not known in which form Herodotus, Plato and Xenophon first encountered the original names. Did they hear them or see them written? If they heard them, did they hear it from native speakers or from speakers of a third language, who may have not pronounced them correctly?We know, e.g., that Herodotus spent a lot of time in Egypt. Furthermore, the Persian empire of V BC was a linguistic peculiarity, since the official language of the Persian Empire was Aramaic. It is also reported that, by the end of V BC Old Persian was almost forgotten. If they saw them written, in which script was it? Was the script capable of representing all the details of old-persian phonology or merely a crude approximation thereof? Was it written in a third language (Aramaic, Egyptian, etc), thus adding another (weak?) link in the transcription chain? It is evident that we know nothing about the origin of the greek forms and their association with the original ones.
What degree of accuracy was intended in the transcription?
Why would the ancient authors strive to represent the sound of a foreign name as closely as possible?The inconsistency of transcriptions is evident i.a. from the deciphered greek names written in Egyptian (Ἀλέξανδρος↔alksantrs/alksnres, Πτολεμαίος↔ptolmês/ptlomês, etc; cf. POPE99, pp. 72-74). Here, there was a good opportunity for near perfect rendering, as the rulers were Greeks and the educated were well versed in both languages. If exact transcriptions were not observed in that case, how do we expect Xenophon and Herodotus to have strived to perfectly represent the zd of Ahuramazda, etc, instead of having provided a mere approximation ([zd]≈[z]) thereof? (cf. also Dārayavauš→Δαρεῖος, Ḫšayāršā→Ξέρξης, Haŋgmatana→Ἀγβάτανα!) We see that the endings were hellenised: -a↔-ης/-ος, -d↔-τος. Why is it not conceivable that the individual sounds were also hellenised? It is possible that the greek authors avoided using consonant clusters that contravened greek phonotactics (σδ was such a consonant cluster, since it does not appear within independent greek morphemes). Why do we assume that the transcription is perfect, as far as ζ↔zd is concerned, but not as far as A(h)u↔Ὠ (in the first example), va↔ο/βα (in the second), d↔τ (in the third)? After all, if we accept the catholic model (αυ=[aw], δ=[d], ζ=[zd]), Αὐραμάζης, Ἀρταυάζης, Ἄζωδος would provide the best possible matches. It follows that phonetic accuracy was not the primary concern of the ancient Greeks that transcribed the foreign names, but that they were more concerned with providing a greek-sounding name that came reasonably close to the original one or to the version that was communicated to them.In that respect note, e.g., the name Σαρδανάπαλος used by Ctesias (V BC), commonly regarded as being the greek version of Ashurbanipal.
It is evident from the above that the Catholics were too eager to accept the mapping ζ↔zd without answering any of these simple questions. But, even if one disregards these general remarks, there are more specific objections for each of the three examples.
So, after some investigation, which none of the Catholics bothered to carry out, it is evident that it cannot be established that the three greek forms were direct transliterations of the three original (persian/semitic) ones and that there are serious doubts about the association of ζ of these three specific examples with zd, rather than with z. Thus, to conclude from a couple of (or single) doubtful, if not forged, examples that greek ζ stood for iranian zd (or semitic shd) is not only naive, it is bad science.
In the face of that, one may wonder why the Catholics were content with these two recurring iranian examples and did not attempt to conform this conclusion with more persian names. Well, it appears that there are no more instances. In fact, there are no more convenient instances, because one may find further examples of persian names having (plain) z in the original and (surprise, surprise) ζ in the greek form. For instance, Xenophon (the very same author that wrote Ἀρτάοζος) reports the names Φαρνάβαζος and Ἀριοβαρζάνης, which correspond to iranian färnəbā´zəs and *Ārya-bṛzāna- (I could not find any iranian form for the name Τιρίβαζος also featuring in the same passage); Herodotus mentions the names Μεγάβυζος and Ζώπυρος, of which the first corresponds to Bag(h)abuxša (undoubtedly a horrible transliteration, but certainly no trace of zd here) and the second most likely does not correspond to Zd-, as this consonant cluster does not appear to introduce any persian words.
Surprisingly, the Catholics neglect to mention one of the most famous transcriptions, the contemporary Zaraϑuštra↔Ζωροάστρης apparently used by Xanthus (mid V BC) and by Plato. Now, if Plato uses ζ to render iranian [zd], why would he use it to also render iranian [z]? One may argue that it was the closest approximation to [z], but then how does it comply with the catholic narrative that until ζ became [z] in late IV BC, σ was used for Iranian [z] (ALLE87: "Later in the 4 c. we begin to find ζ replacing σ used for Iranian z")?I have not yet seen any convincing evidence for that claim (σ↔[z]). The only example I have encountered is the sorry "Zara(n)ka- = Σαράγγαι". Now, this is yet another case of little-thought "proof", as can be seen by some elementary research. It is reported that there are several variants: Σαράγγαι by Herodotus, Ζαραγγαῖοι by Arrian, Zarangae and Drangae by Pliny, all believed to refer to the same people, where "The forms with S or Z reflect the local pronunciation, those with D the Old Persian one". With so much variation, how do we know that Herodotus chose the "local" pronunciation (which, allegedly, could not be accurately expressed in Catholic Greek), overlooking the "Old Persian" one (which would be more easily rendered)? How can we be sure that the name was transmitted to him in the "form with Z" and not in some third form that employed S? In fact, according to Encyclopædia Iranica, there exist "Elamite (Sir-ra-an-qa and variants), Babylonian (Za-ra-an-ga), and Egyptian (srng or srnḳ) versions", which could have been the basis for Herodotus' transliteration. The Egyptian one, particularly, has a high probability of being the source (thus, explaining the use of Σ), as Egypt was the primary travel destination for Herodotus and probably where he collected most of his (hi)stories. It appears that whoever came up with this idea forgot to establish the variant after which "Σαράγγαι" was modelled, before jumping to the conclusion that Herodotus used Σ for lack of a better candidate for [z], while the later authors took advantage of the "mutated" Ζ to express this sound.
In the case of Ζωροάστρης (and to a lesser extend in the case of Φαρνάβαζος and Ἀριοβαρζάνης) we have a strong indication that ζ↔[z], which cannot be ignored or explained away. Why has no Catholic mentioned it or at least commented on it? Could this be a case of honest omission or neglect? Absolutely not! It cannot be argued that the example of Ζωροάστρης did not occur to them, because it is found side by side with Ὡρομάζης in the specific passage of (Pseudo-)Plato cited by Blass and Sturtevant, which means that they deliberately chose to ignore it. This a blatant case of cherry picking and unscientific deceit!
This is a claim, the first part of which ("we find -ζε and -ζ-") is a fact and the second ("for -σδε and -σδ-") a conjecture. The instances cited are
The first observation to be made is that, in addition to being pure guesswork (particularly when the data for "reconstructing" PIE is so scanty), etymology is never proof for pronunciation. For example, we have seen that modern-german sch is considered to derive from west-germanic /sk/, but this does not mean that sch is pronounced [sk] (it is pronounced [ʃ]). Equivalently, modern-greek ζ has in some cases the same etymology as ancient-greek ζ coming from assumed [sd], but it is never pronounced [sd]. Thus, even if we accept the etymology as correct, we do not know whether the phonetic process (fusion, simplification or whatever) that produced the modern pronunciation [z] kicked in yesterday, in classical times or prehistorically. It is, thus, pointless to examine whether the etymological assertions ὄζος←ὄσδος, etc, are justified or not, as they bear no significance for establishing the classical pronunciation.
The other two cases of -ζε(←-σδε) and -ζ-(←-σδ-) are also potential cases of proof by etymology. That is, they may simply derive from the same phonological process that produced ζ from σδ acting upon the cluster -σδ- (to eliminate this "un-greek" combination of consonants) in cases where the two morphemes (-ς and δ-) were no longer considered separate. Otherwise, specific observations about the meaning of ζ in these examples have already been made. In brief:
Threatte triumphantly declares (THRE80, p. 546) that "Two Attic texts conclusively show the pronunciation of zeta as the obstruent cluster [zd] in the sixth and fifth centuries B.C." (emphasis mine). The texts he refers to are a) the already mentioned τοῖσζ' for τοῖσδ(ε) and b) Ζδεύς in a "Dipinto on a vase of the Tyrrhenian group". However, the first example is not a case of ζ for σδ, but of ζ for δ! Threatte's explanation that σζ=[zzd] in analogy to -σστ- for -στ- is imaginative, but involves too many assumptions (i.e., that σδ was replaced by ζ, resulting in an intermediate unattested form τοῖζ' and then σ was added, only in this case, but never in the, allegedly equivalent, forms Ἀθήναζε, etc; his cited Argive equivalent δικάσζοιτο is not a case of ζ standing for normative σδ, i.e., †δικάσδοιτο, so it is only relevant if one a priori accepts that ζ=σδ). But even the correctness of the replacement is doubtful; according to Teodorsson (TEOD74, p. 140), who claims to have personally examined all the inscriptions he cites, the ζ (actually ɪ, the old form of ζ) in this particular inscription is due to an effort of the stonecutter to amend an originally carved Τ; it will never be known for what reason he wrote τ, what his initial intention was and whether the correction was meant to turn the τ into a ζ, an incomplete δ or a... smudge. As for the second example, if ζ=σδ, then it stands for *Σδδεύς, where the double δ cannot be explained by a similar analogy to unattested †σττ or any other of the Catholics' favourite voodoo; if Ζδεύς is derived from the assumed "original" *Σδεύς, then the reason can only be that ζ=[z] replaced the (voiced by regressive assimilation) Σ! What's more, as Threatte himself submits, there is a non-Attic form (Νέσος) on the same vase, and the spelling ΔΒΕΥΣ is found on "another vase of the Tyrrhenian group", which he considers "still awaits satisfactory explanation" (read: "needs to be explained away"); it is curious that Threatte accepts the former instance as unequivocally correct and rejects the latter as suspect. If these are "conclusive" evidence for the Catholics, then one can only imagine how unimportant their "not-so-conclusive" evidence can be.
OK, so? French c is pronounced [s] (before e, i), but this does not mean that it is a relic of some "original" pronunciation of Latin. This simple truth is overcome by the Catholics (in order to back their theory that ζ's original pronunciation was [dz]) by yet more unproven speculation, claiming that out of all places where Greeks lived, the allegedly original pronunciation [dz] survived in the mouths of the Italians: "The pronunciation dz must have been current among the early Italian Greeks; for in Etruscan, Oscan and Umbrian the character ɪ[=ζ] represented the sound ts, with t for δ as in Latin citrus for κέδρος [...] The use of z in Italian in the value of dz (orzo) and ts (grazia, etc.) proves that this primitive value of the character has always survived on Italian soil" (STUR20, p. 190; emphasis mine). Now, have the Catholics ever paused to hear what they are saying or are they just too busy inventing "proofs" and bragging about how "scientific" work they have done? Here is their narrative (quotes from STUR20):
There are several unproven assertions in these speculative ramblings:
It is, thus, evident that the whole theory rests on the imaginary influence of "Italian Greek" on Latin and the other Italian languages, which is never proven or is rather proven "by Blatant Assertion" (ZEMA94). The brothers Grimm would have resigned from story telling, if they were confronted with such competition from the "scientists".
Yet, the simplest solution escaped(?) the venerable scientists: the use of z in Italian as an affricate is an internal phonetic or orthographic development that has nothing to do with any fictional pronunciation of the "Italian Greeks". Instead of making the Italians the arbiters of the genuine greek pronunciation, and imposing onto Greek a sound that could never have occurred (as implied by the general disappearance of dentals before σ), one should try to explain why ζ developed into an affricate in Italian. How it came about can only be a matter of speculation, but here are a few theories for the hungry minds:
Before blaming the present sound of italian z on the Umbrians and Oscans, whose languages disappeared centuries before the first instances of z for di/de, and on the asserted, but not proven conservatism of Italian Greeks, the Catholics might want to consider one or more of the above theories, which are based on the most reasonable premise, namely that z developed into the affricate it is today on Italian soil by Italian(=late-Latin) speakers. But none of the above theories would serve the catholic plans, would it?
True, but we know neither the date of the pun nor its relevance to phonology. The catholic hallucination is that the pun is only justified under the assumption ζ=[zd]: "The [zd] value also incidentally adds point to the comic ὦ Βδεῦ Δέσποτα cited by Tzetzes, possibly referring to Aristophanes, Lysistrata, 940, where the MSS have Ζεῦ" (ALLE87, p. 56). Allen's statement was initially incomprehensible to me. How exactly does ζ=[zd] add any point? If Ζεῦ=[zdeu] what does Βδεῦ (presumably [bdeu]) have to do with it? What is the origin of Β? Where does it come from? Certainly not from any sibilant component in Ζ, e.g., as a result of a lisp, since lisped /z/ is [ð], not [b] (Catholic Β) or [v] (Orthodox Β). The pun is perfectly clear, as a case of alliteration, when I read it with my "barbaric" (as implied, e.g., by Blass; BLAS70, p. 7) pronunciation: [ˈo-ˈvðe-ˈvðe-spo-ta]. Why would anyone look for a different explanation? What is the reason for thinking that Bδεῦ justifies a specific pronunciation of ζ? An investigation of the origins of this argument may provide an answer.
Allen was not the first to cite this statement as "evidence" of the pronunciation ζ=[zd]. It appears that it was first brought forward by Blass (BLAS82, p. 97; BLAS90, p. 117): "verdrehte ein attischer Komiker ὦ Ζεῦ Δέσποτα in ὦ Βδεῦ Δέσποτα, mit großer Härte, wenn es Δσεῦ hieß, dagegen ohne Mühe bei der Aussprache Σδεῦ [German original of the 2nd edition]|the distortion of ὦ Ζεῦ Δέσποτα into ὦ Βδεῦ Δέσποτα by an Attic comic poet would be very harsh if the pronunciation were Δσεῦ, but quite easy if it were Σδεῦ [English translation of the 3rd edition]". This is part of the feud between Blass and Curtius (cf. PAPA89, p. 590) over the order of σ and δ in ζ (i.e., whether ζ=σδ or ζ=δσ), with Blass throwing any (even remotely-relevant, if at all) "evidence" in his argumentation, in his ardor to prove the master wrong. As such, the argument is potentially pure junk, but since it has made it into Allen's "hit list", closer examination is probably required.
At first, the context of the quote must be established. It may seem that it was part of an attic comedy play, but this is only a conjecture. The passage appears in the fragment cited by Blass: "ἕκτον κατὰ ἐξαλλαγὴν ὡς τὸ Ὦ Βδεῦ δέσποτα, ἀντὶ τοῦ Ζεῦ|sixth, by alteration like the [expression] Ὦ Βδεῦ δέσποτα, instead of Ζεῦ"; it is not clear where the fragment is taken from. Allen submits that it is cited by Tzetzes (of XII AD!!!), but without any specific reference that would enable one to locate the context in which it is cited. However, the relevant expression can be found in several scholia, e.g., in the prolegomena de comoedia: "Ὁ δὲ γέλως τῆς κωμῳδίας ἔκ τε λέξεων καὶ πραγμάτων ἔχει τὴν σύστασιν· ἐκ μὲν τῶν λέξεων κατὰ τρόπους ἑπτά. [...] ἕκτον κατ᾽ ἐναλλαγὴν ὡς τὸ Ὦ Βδεῦ δέσποτα, ἀντὶ τοῦ Ζεῦ|the laughter of comedy consists in words and acts [?]; words in seven ways. [...] sixth, by interchange, like Ὦ Βδεῦ δέσποτα instead of Ζεῦ". Here we have virtually the same excerpt with "ἐναλλαγὴν" instead of "ἐξαλλαγὴν". Neither of the two words justifies the assumption that exactly one letter is substituted by exactly one other (allegedly the Σ of *Σδεῦ by Β); ἐναλλαγὴ is merely "interchange" or "variation", while ἐξαλλαγὴ is "complete change, alteration". Thus, the conclusion that "Βδεῦ for Ζεῦ, is letter-substitution" is not justified.
Furthermore, we have no idea which comedy play (if any) features the expression "Ὦ Βδεῦ δέσποτα". Allen speculates that it refers to Lysistrata 940, but he also admits that no surviving copy has the word Βδεῦ. Even assuming that this was its original location (and was, e.g., censored by medieval copyists), the pronunciation [zd] for ζ does not make more sense than the present value [z]. This part of Lysistrata relates to the exasperation of Cinesias in anticipation of his promised intercourse with his wife Myrrhine. She drives him so crazy by inventing several excuses to put off the act that, after she starts covering him with perfume, he exclaims: "εἴθ᾽ ἐκχυθείη τὸ μύρον ὦ Ζεῦ δέσποτα|Lord Zeus, may she soon use up all the myrrh [Lindsay's translation]". Why would he insult Zeus (or invoke some "Lord Fart") for his wife's procrastination? How would the use of Βδεῦ in this context cause the laughter of the spectators, as if they were school kids laughing at the mere allusion to the word βδέω|I fart? If, on the other hand, one considers that Cinesias is by this time so much in agony, it is reasonable that he would not be in full control of his articulation; then, in his attempt to summon the name of Zeus, his stuttering tongue messes up the intended (common) expression "ὦ Ζεῦ δέσποτα" and pronounces "Βδε" instead of "Ζε", in anticipation of the next syllable "ῦ Δέ" ([vðe]), producing a lisping, stuttering and profane/comical effect, all at the same time.
Now, the wit of such a word play would never be grasped by a Catholic educated in an environment of western pronunciation (β=[b], δ=[d], ευ=[ew], etc), who pronounces the pun [o:-bdew-des-po-ta]. Suddenly, Caragounis' thesis that one should have "a thorough knowledge of Greek (preferably in all its periods)" (CARA95, p. 152) in order to understand the texts (particularly authors as late as Tzetzes, who almost surely spoke like a "modern Greek") seems more relevant than ever. If I were a Catholic, I would not even mention this example, because it rather confirms many Orthodox views (ζ=[z], δ=[ð], β=[v], ευ=[ev]), but I would maintain that the fragment is not original, Β having been inserted by a "Byzantine" scribe who thought that it makes more sense because of his "modern-Greek pronunciation".
In their attempt to reconcile the two incompatible theses ζ=σδ and ζ=δσ, after being convinced that both had their merits, the Catholics have abused and undermined the very science they were supposed to be serving. This is understandable for scholars of the 19th century, a time when linguistics was still at its infancy, but unforgivable for the later linguists, who cannot plead inexperience. Let us see how one of the biggest hoaxes in linguistics came about, starting from the actual facts and scrutinising the reasoning.
The primary reason for claiming δσ as the original value of ζ is that they trace it back to PIE *dj, *gj and *gʷj and that it corresponds to δι̯ and γι̯ resulting from inflection, where δ=[d], γ=[g] according to the Catholic doctrine. The Catholics can imagine no other descendants from these clusters than voiced affricates, which they generally represent by dz or dž. This is the first divergence from proper linguistics, as they should have instead used [d͡z] and [d͡ʒ], the affricates, not dz and dž, the consonant sequences. The notation dz dates back to the 19th century, practically the dark age of linguistics, a time during which the linguists (mostly Germans and Anglo-Americans) had a very limited experience with other phonologies and interpreted everything through the prism of the sounds of their own languages, but this practice has survived up to the present time (cf. ALLE87, p. 56; HORR10, pp. 10, 11, 31). One reason is probably their ignorance or, to put it mildly, their unfamiliarity with the conventions of the IPA, even in cases where they have "heard" of the IPA (cf. ALLE87, pp. 11 and 32; we have seen that Allen has misused the term "voiced palatal fricative" for "ž"=[ʒ], instead of the correct [ʝ], which indicates that he does not understand the IPA). But even when they are well versed in the IPA, they prefer to write dz, /dz/, or [dz] because it helps them forget that they speak of an affricate and allows them to juggle the two letters around, as we will shortly see. For the record, let me point out one more time that there is a difference between [d͡z] and [dz], which is not merely semantic ("catch it" is one thing and "cat shit" another). One of the most important points that I have already explained is that the affricates [t͡s], [d͡z], etc are single sounds, and they are so perceived by the speakers of many languages that use single symbols to represent them. This is something that most of the modern (western) scholars fail to understand, undoubtedly owing to the digraphs employed in their languages (and, partly, also in the IPA) to represent them. In that respect, they are no more sophisticated than the ancient Grammarians, who follow the script rather than the pronunciation.
Having convinced themselves that the later pronunciation was [zd] and only that, they tried to reconcile the "original" and the "later" pronunciations by means of linguinistics: "the presumed pronunciation of these latter forms with [zd] represents a metathesis of the fricative and stop elements" (ALLE87, pp. 56-57)! Now one can see why the use of the special IPA symbol [d͡z] for the affricate is avoided and the convenient digraph dz is employed: it allows them to pretend that it comprises two consonants and then claim that these two consonants can be readily transposed and that consistently in all instances of the affricate.
But there's more. In an attempt to solidify his argument about the plausibility of this transformation, Allen states: "such metatheses are of a particularly common type; R.P. wasp, for example, derives from an earlier and still dialectal waps (cf. Old Prussian wobse); and the particular change in question is closely paralleled e.g. in Old Church Slavonic mežda from Indo-European medhyā; an intermediate stage must here have been medža, which has given Russian meža 'boundary' (Russian méždu 'between' is a borrowing from O.C.S., being the locative dual of mežda)." (emphasis mine). Where can one start with the criticism of this statement?
All in all, the attempt to provide the missing link between the two catholic theories ζ=σδ and ζ=δσ is only valid if the latter is the affricate [d͡z] (so that it can result from *dj, but most importantly from *gj), but behaves like the consonant sequence [dz] (so that its constituent consonants can be transposed). While such theories have their merit in other disciplines, this case of Quantum Linguistics is not only bad science, it is bad science fiction!
So far, I have dissected the arguments put forward by the Catholics in favour of their proposed values [zd] or [dz] for ζ and against the present value [z] and (hope to) have shown that the traditional value ζ=[z] is still compatible with the provided evidence. Subsequently, I will attempt to show that further facts or a re-examination of the already presented ones almost preclude a composite ([zd] or [dz]) or affricate ([d͡z]) sound for ζ and are rather suggestive of a simple sound [z] for ζ. As it turns out, some of the best arguments against the catholic dogma on ζ are provided by the Catholics themselves, during their 19th-century vendetta between the supporters of [dz] (led by Curtius) and those of [zd] (led by Blass). This is, hence, a case of two fighting and the third winning. Before continuing from where I left off (i.e., the etymology of ζ), I would like to illustrate the relations between the various consonants.
I have elsewhere shown that the basic consonants of Greek (with the term "consonants" I am referring to sounds, not letters) can be divided into four groups or so-called series according to place of articulation (labials, dentals, palatals and velars) with the consonants of each series being characterised by two properties, voice and duration; the latter refers to the consonant's ability to be sustained in time (i.e., prolonged), something that is only possible if the airstream is not completely blocked;It appears that the scientific (and nicer) term for the property I name "distance/duration" is "stricture" (e.g., KEHR02, p. 2). thus, if a pair of organs (particularly the tongue or lower lip and the roof of the mouth or upper lip) completely blocks the airstream, the consonant is a stop or plosive, but if they are set at a distance, the consonant is a continuant or fricative. A similar binary distinction is applied on the other property: either the vocal chords participate in the articulation and the consonant is voiced or they do not and the consonant is voiceless. Each series can, therefore, be represented in the form of a 2x2 matrix, one property per dimension. The four greek sibilants, which are all alveolar, can also be arranged in a similar 2x2 matrix and the same can be done for the corresponding post-alveolar (or palato-alveolar) and alveolo-palatal sibilants (which are not featuring in Greek). As the place of articulation of these three series of sibilants is progressively between that of the dentals and the palatals, the 2x2 matrices representing the aforementioned seven series can be arranged along a third axis, thus resulting into a virtual 3D structure, which demonstrates the relationships between the various consonants in terms of the three aforementioned properties. This structure, bar the labial series, which is not relevant for the present discussion, is illustrated in the image to the right. I have also added the so-called palatal semivowel [j] below the palatal fricative [ʝ], as it is located further down(=at a greater distance from the palate) the palatal axis (the column, if expanded downwards, would also comprise [i], [e] and eventually [a], as already illustrated). To keep things simpler, I will use a version comprising only five series, the alveolo-palatals having been removed, as it is often difficult to distinguish them from the palato-alveolars.
This 3D structure, to which I will refer as "The Matrix", is a very powerful visual aid for determining whether a certain phonetic development is plausible. To demonstrate its potential, let us consider the already mentioned development of γ assumed by Allen. He believes that the sound of γ, which he identifies as [g], shifted before front vowels to its present value of [j] (always according to him), "via a voiced palatal fricative [...] Cf. Armenian Diožēn = Διογένης", i.e., [ʒ]. If one is to trace this development in The Matrix, one would obtain the red trajectory in the image to the left, where the (allegedly) original [g] jumped over two intermediate sounds ([ɟ] and [d͡ʒ]) into a different row (i.e., from stop to continuant) and then sunk under the neighbouring fricative ([ʝ]) to semi-vowel status ([j]). As can be seen, the phonetic development proposed by Allen is an accomplishment that even John Carter of Mars would envy. No self-respecting phonetician would subscribe to such kangaroo-phonetics. Instead, if there are no external factors, such as adoption of the language by non-native speakers (in which case, they might approximate the sound with one of their own language's, which may not be adjacent to the original sound), the behaviour of sounds during normal sound change is not so erratic, but is expected to be smooth, progressively shifting to neighbouring sounds. This means that the only possible movements (under normal circumstances) are along the three axes (distance, voice, place) and at single steps (i.e., movements at Manhattan distance 1). Therefore, the development of γ before front vowels towards its present value [ʝ] (and not [j] as claimed by Allen) is a normal case of palatalisation almost certainly from the closest velar value [ɣ] (solid green arrow). If γ ever had the value [g], then it must have first undergone "fricativization" to [ɣ]; alternatively, it might have had a palatal allophone [ɟ] before front vowels (as is normally the case in most languages, including modern Greek), which shifted directly to the "fricative" [ʝ] (dotted green arrows). The Armenian value [ʒ], if not a mere approximation, is a further development from [ʝ] (brown arrow) and not an intermediate greek value as asserted by Allen.
A further demonstration is to use The Matrix for depicting the development of Latin c (undoubtedly original [k], even though probably not before front vowels, at least not in late Latin) into Spanish c=[θ]. The first development is the expected "fronting" when the velar [k] is followed by a front vowel. The instant result must have been the corresponding palatal [c] (LLOY87, pp. 135-137), which was later further "fronted" to [t͡ʃ], its present value in Italian (p. 138), which in Iberia merged with [t͡s], the result of [tj] (p. 259). Then, it moved to the closest fricative [s] (pp. 331-332) and from XVI AD it progressively acquired the pure fricative non-sibilant pronunciation [θ] (p. 334), probably to emphasise its difference from s=[s̺] (p. 336). This stepwise non-jumpy development is depicted on The Matrix in the illustration to the right and is an excellent example of how a sound is expected to develop during regular sound change.
Now, let us return to the origins of greek ζ as provided by etymology, namely the sounds that (we believe) it derived from. Excluding a handful of doubtful cases where it is assumed to have developed from σδ or rather *sd (which still does not preclude these two sounds merging into a [z(ː)]), in the majority of cases ζ can be traced back to the palatal semivowel [j] either alone (*j-) or combined with a voiced dental (*dj) or a voiced velar (*gj), as already seen. Although etymology is often speculative (as in the case of the derivation of ζ from alleged *zd in ὄζος, ἵζω, etc), there is no reason for raising objections against its derivation from *j-, *dj, *gj, since that is mostly based on inflectional patterns. However, for the same reason (i.e., since ζ originates as ἁρπάγ-ι̯ω→ἁρπάζω, κομίδ-ι̯ω→κομίζω, etc) and for not prejudging the phonetic value of these "original" clusters as [j], [dj], [gj], I will use the notations ι̯, δι̯ and γι̯ respectively (which have the advantage of employing the mediae δ, γ of the original forms without assuming any particular pronunciation thereof, as judgement on their values is still pending). Thus, the fact can be formulated as:
In the above notation, ι̯ is a non-syllabic ι (essentially the palatal semivowel [j]) and δ, γ are voiced mutes. Based on the "reconstructed" values δ=[d] and γ=[g], the Catholics argue that only [dz] can result from δι̯ and they also consider it the most likely value for γι̯. As a matter of fact, a voiced affricate value is a reasonable conclusion, under the premise that δ and γ are voiced stops, as these would be responsible for the affricate's occlusion and the semivowel for the (sibilant) release. After all, it is evident (by glancing at The Matrix) that [d͡z](=ʣ) is the immediate neighbour of [d] and that [d͡ʒ](=ʤ) or rather [d͡ʑ](=ʥ) is the sibilant neighbour of [ɟ], which is the expected "fronted"(=palatalised) version of [g] before front vowels. Thus, it is not unreasonable that the assibilation induced by the semivowel [j] on the voiced stops [d] and [g] result in voiced affricates.
This observation is the main point of the proponents of the [dz] pronunciation against those of the [zd] one. The Tafels, for example, devote an appendix of 16 pages (TAFE60, pp. 157-172) to the subject of "Zetacism", where they provide examples from several languages, including Chinese, Tibetan, Mongolian and Modern Arabic, which have developed affricates from the combination of a dental or "guttural" stop and /j/. It is, therefore, considered that δι̯ can develop in no other way but into an affricate, specifically [d͡z]. As the Catholics see no mechanism that would enable the development of a [zd] out of δι̯ (ignoring the slavic evidence put forward first by Blass and the alternative explanation proposed above), they accept the affricate value [d͡z] as the "original" and have attempted to reconcile it with their favourite value [zd] by means of the unscientific "metathesis of an affricate", as we have seen, which later "lost" its [d] to attain its present value [z].
However, before resorting to science fiction, let us take a step back and, instead of looking at Tibetan or Mongolian, have a look at the recent history of the language we are actually investigating. In medieval (or later) Greek, there are at least three instances of conversion of δι̯ to ζ (these are the ones that I have found in a brief search; I do not know whether there are others): ζαβολιά←διαβολιά, ζουλώ←ζουλίζω←διυλίζω, ζουπώ←ζουπίζω←*διοπίζω. ζ here stands for its normal value [z], as the etymologies are medieval, several centuries after the time ζ is assumed (by the Catholics) to have stood for [zd] or [d͡z]. A secondary development (i.e., by means of another, intermediate sound, i.e., [d͡z]) can most certainly be excluded. I am not aware of any instances of conversion of δι̯ to τζ (the medieval spelling of [d͡z], as in Τζέτζης=Tzetzes=[ˈd͡zed͡zis] and other words mostly of foreign origin). So, we have clear evidence that δι̯ in Greek can (directly) develop to [z] rather than to [d͡z] as would be expected by the Catholics.As a matter of fact, among the "zetacistic" phenomena discussed by the Tafels (TAFE60) one finds developments into pure sibilants instead of the expected affricates: e.g., Lithuanian breedi-s↔bree-zha←breedya (p. 161) or Finnish, where "the dental t, when followed by i turns regularly into s" (p. 159). (Medieval) Greek is, therefore, not alone in the development of pure sibilants instead of affricates.
Why, then, is the outcome different from that of Mongolian, etc? One reason could be that the language lacked the expected sound (here [d͡z]) and was reluctant to introduce an unfamiliar sound, but instead used the closest available (cf. the use of y=[j] or [ʝ] in Spanish for English j=[d͡ʒ], instead of the closer palatal stop [ɟ], which is missing from Spanish); this was certainly not the case in medieval Greek, as the sound [d͡z] was available (under the spelling τζ). Well, the plain reason is that (at least at the time of these sound changes) δι̯ was not [dj], but [ðj]! Having only continuant (or "fricative") components, the original δι̯ could not have (in a primary step) developed into an affricate, which by definition involves obstruction of the airstream (in other words, the closest sibilant for [ð] is [z], as can be seen in The Matrix). If we, thus, allow for the possibility that the ancient Greeks spoke like (present-day) Greeks and not like Germans (i.e., they pronounced δ more like [ð] and less like [d], which is not as unlikely as the Catholics want us to believe), the emergence of the plain sibilant continuant [z] would be the reasonable development for this kind of δι̯. This approach (i.e., the emergence of the present value [z] directly from δι̯) has the additional advantage of not having to assume the extremely problematic and improbable intermediate developments δι̯→[d͡z]→[zd]→[z].
But the origin of ζ in δι̯ is only half the truth; for ζ originates also from γι̯ and from (word-initial) ι̯. While the first one is considered by the Catholics a combination of a voiced stop and the palatal semivowel ([gj]) which would naturally lead to an affricate (however, if similarly to δ we accept for γ its present continuant value [ɣ], the plain sibilant [z] or [ʒ] is the most likely development), the emergence of the assumed affricate [d͡z] from the semivowel ι̯ is incomprehensible, at least as a primary development.To be honest and entirely consistent, the derivation of ζ from ι̯ (more correctly *j-) is less a fact than is its derivation from δι̯ and γι̯. The latter is verifiable from internal evidence, namely observing stems and how they combine with endings beginning with /j/ (such as ἁρπάγ-ι̯ω→ἁρπάζω, κομίδ-ι̯ω→κομίζω), but the former is only based on external evidence, particularly the existence of a /j/ at the same position in various IE cognates (such as "Lat. iugum, Goth. juk, Lith jùngas, OCS igo" for greek ζυγόν) or a "reconstructed"(=conjectured) PIE form (such as *yugóm). Thus, it would not be correct to state as a fact that ζ sometimes derives from ι̯, but rather that it appears where other cognates have /j-/. There have also been attempts to explain such ζ as deriving from the combination of a preceding "laryngeal" and the semivowel, particularly from PIE *γj-; in addition to being so speculative that it is backed by a single pair of cognates between a greek and a "hittite" word (thus being doubly speculative, since we have seen how reliable Hrozný's decipherment is), this theory does not explain any better how the (assumed) affricate's stop component derived from the fricative [ɣ] and how (much) [ɣj] would have differed from the plain palatal fricative [ʝ] (all understandable for someone who has probably never heard a γ in his life). While it is true that english j and italian gi (both [d͡ʒ]) have their etymological origin in latin consonantal i=/j/ (now orthographically rendered as j), these are most likely secondary developments.This probably a good place to present my objection on the official narrative that latin consonantal i(=/j/) developed directly to [d͡ʒ] in Italian and early French, from where English took it, and then "lost" its stop component in later French to yield the modern [ʒ]. The primary reason is The Matrix: the reasonable assumption is that, before being assibilated, the semivowel [j] was consonantised to the closest consonant [ʝ]; [ʒ] is a neighbour of that sound's (i.e., is at Manhattan distance 1), while [d͡ʒ] stands at Manhattan distance 2 (or 2 and 3 respectively, if no intermediate consonantisation is assumed) and would require a quantum leap to emerge from [j] or [ʝ]. Furthermore, if at the time of English adoption of the letter j from French j were an affricate, then the same should have been the case for c (before i, e, y), which almost certainly developed from a velar [k] (or palatal [c]) in Latin to an affricate [t͡s] (or [t͡ʃ]); however, English c has exactly the non-affricate value [s] that french c has today in that environment and English ti=[ʃ] (e.g., in -tion) is the result of the merging of French ti=[sj] (i.e., "palatalisation" induced on [s] by the semivowel /j/), not [t͡si] which must have been the primary development of the assibilation of latin ti; since French had lost all voiceless affricates by the time of English adoption (the case of ch resulting from latin ca, which is [ʃ] in French and [t͡ʃ] in English, will be dealt with on another occasion), the reasonable assumption would therefore be that j was also a plain sibilant ([ʒ]) in French, but it developed into an affricate in English (a development paralleled by that of g, which has the same affricate value [d͡ʒ], although it is the voiced counterpart of c); otherwise, we have to make the improbable assumption that j was turned from a semivowel ([j]) or fricative ([ʝ]) to an affricate ([d͡ʒ]) and then to a fricative ([ʒ]); i.e., most likely [j]→[ʝ]→[ʒ]→[d͡ʒ]→[ʒ]. The story of /j/ in Italian (and probably Occitan) must be somewhat different; it is very likely that the spelling (g)gi used for former consonantal i (e.g., Giulio↔Iulius, giustizia↔iustitia, maggiore↔maior, etc) is a way of indicating a consonantisation of the semivowel [j] to the closest consonant, the palatal fricative [ʝ] (for the equivalence of late-latin gi and j, see GRAN07, p. 114, §272, this j being a continuant, according to Sturtevant's verdict based on transliterations like κόζους↔cojux, etc, STUR20, p. 191, as well as on "the Romance languages" and "the misspellings of late inscriptions and of manuscripts", p. 48), the first stage before assibilation (to [ʒ]) and then affrication (to the current [d͡ʒ]); this would be paralleled by the modern greek convention of writing γι in the few cases where a standalone (i.e., not preceded by a consonant) non-syllabic ι is consonantised, e.g., ιατρός|(medical) doctor, physician→γιατρός=[ʝaˈtros], Μαΐου|of May→Μαγιού=[maˈʝu], and it would be consistent with the use of gu in late Latin to denote the consonantal nature of germanic w, e.g., wërra→guerra, wîsa→guisa (GRAN07, p. 143, §344; cf. modern-greek γου for english w, e.g., Γουόλτερ Γούντμπερν←Walter Woodburn); incidentally, such a value for gi would also explain the variant spellings with y in Old Spanish fuyo←fugio[LAT], also oyo←audio[LAT] (LLOY87, p. 296). All in all, there are several indications that the English and Italian (also Occitan?) affricates [d͡ʒ] corresponding to latin consonantal i are secondary rather than direct developments. By accepting a value ζ=[z] (or, in general, some similar fricative voiced sibilant), one does not need to make any further exotic assumptions, like parthenogenesis of *gj- out of *j- to desperately explain the emergence of an affricate [d͡ʒ].
We have, then, our first explanation of the emergence of ζ=[z] out of δι̯, γι̯ and ι̯ (illustrated in the figure to the right):
This course of events is very plausible, under the premise that δ=[ð], γ=[ɣ] at the time of assibilation of δι̯, γι̯ and ι̯. Since this sound change is certainly pre-alphabetic (as ζ was adopted to denote its result), we have virtually no information on the actual values of δ and γ at that time.
However, there is another plausible development of the clusters that does not rely so much on the exact values of δ and γ, but on their correspondence with latin d and g. In that respect, it is considered that (in english notation) late-latin dy(=de̯, di̯) and gy(=ge̯, gi̯) were reduced to y(=[ʝ]), the consonantised version of earlier "consonantal i"(=/j/);There are plenty of references to this phenomenon in the literature, e.g.,
STUR20: "It appears from the Romance languages and from the misspellings of late inscriptions and of manuscripts that consonantal i came to be identical in sound with g before e and i, with gy and dy" (p. 48); "In Vulgar Latin di and de before vowels came to have the same sound as consonantal i, and they were confused with the latter in spelling" (p. 110);
GRAN07: "Dy and gy in the latter part of the Empire, probably were reduced to y in vulgar speech" (p. 114);
LLOY87: "The combination of [dj] and gj] then became identical with intervocalic [-j-] and suffered all changes that affected this consonant" (p.133). this [ʝ] was subsequently assibilated to [ʒ]LLOY87, p. 132: "The palatal semiconsonant [j] likewise strengthened its articulation about the same time so that it too became clearly consonantal and developed a strongly fricative pronunciation in word- and syllable-initial position, e.g., ZANUARIO for IANUARIO, in which the letter Z may represent something like a palatal fricative [ž] or an affricate [dž]". in French, Portuguese and Old Spanish, [ʒ] being probably also an intermediate value in Italian (and Occitan?) before its affrication to [(d)d͡ʒ].Contrary to the catholic narrative, the regular reflex of latin dj in Italian is [(d)d͡ʒ], e.g., poggio←podium, oggi←hodie, giorno←diurnus, rather than [(d)d͡z], which is the exception rather than the rule (that the palato-alveolar value [(d)d͡ʒ] can only come from an original palatal, while the alveolar [(d)d͡z] from an original dental in Italian is evident from the analogous voiceless developments [(t)t͡ʃ]←[c(j)] and [(t)t͡s]←[tj] in the same language). I suspect that the emergence of italian [(d)d͡z] in a few cases corresponding to latin prevocalic di and de is a later development, a second assibilation, which kicked in after d attained the plosive pronunciation ([d]) that it today has in Italian, in a few cases where the clusters di and de survived (i.e., never transformed to [j] or [ʝ]); e.g., di and de have survived in medius and hordeum yielding romanian medi, spanish medio, sardinian mediu (even italian medio) and spanish hordio, catalan ordi, besides italian mezzo and orzo. If we, thus, consider that the clusters δι̯ and γι̯ shared the fate of late-latin dy and gy, we have an alternative explanation for the emergence of ζ=[z]:
So far, the proposed developments do not aim at "proving" that the pronunciation of ζ was [z]. They are submitted (backed by analogous developments in later Greek and the Romance languages) only as proof that the modern value could have arisen from the "original" sounds δι̯, γι̯ and ι̯-, i.e., that the etymology is not against the orthodox value of ζ. What about the catholic values? Could a [dz], [d͡z] or [zd] have arisen from δι̯, γι̯ and ι̯-? Let us consider them separately, assuming that δ=[d] and γ=[g], as per the catholic doctrine.
So, we have two possible alternatives, the orthodox [z] and the catholic [d͡z], compatible with the evolution of δι̯, γι̯ and ι̯- in primitive Greek. This does not mean that they are equally likely though. It suffices to look at the outcome of the corresponding voiceless clusters τι̯, θι̯, κι̯, χι̯, which yield (σ)σ in general Greek (SMYT20, pp. 29-30, §112-115); while there is (still) a lot of speculation that general-greek σσ and attic ττ were affricates [(t)t͡s] (which is not very credible for reasons already laid out), the (plain) σ resulting in such cases as the "characteristic innovatory 'assibilation' of original [t] before [i] (i.e. [ti] > [tsi] > [si])" of"East Greek" (HORR10, pp. 19-20) and in the etymologies παντ-ι̯α→πᾶσα, μεθ-ι̯ος→μέσος, τοτ-ι̯ος→τόσος (SMYT20, p. 30, §113) cannot be considered an affricate. If the voiceless clusters (at least τι̯ and θι̯) yielded the sibilant continuant (σ)σ=[s(ː)], then why would the result of the voiced clusters (at least of δι̯) be anything other than the voiced version of (σ)σ? In other words, if we conclude, based on the voiced developments, that ζ=[d͡z], then we have to also conclude that σ=[t͡s]; conversely, if we accept that σ=[s], the only consistent assumption would be that ζ=[z]. Even if the (unattested) affricates [t͡s] and [d͡z] are assumed as intermediate values, it makes more sense that they both had the same fate, i.e., lost their stop character together and not with a lag of several centuries (any potential voiceless affricate value corresponding to later σ would have to be pre-alphabetic, while the plain-sibilant pronunciation of ζ as [z] is allegedly "Hellenistic"). Moreover, the affricate value [d͡z] suffers from a number of other problems, most of which are pointed out by the Catholics themselves (albeit of the other denomination ζ=[zd]) and which we will see in the following.
ζ (or rather Ζ), as a grapheme, is (unlike, e.g., φ, χ, ψ and the vowels) not among the greek innovations. It rather occupies the place of semitic "zayin" (ז) and its shape (particularly the original ɪ) can be linked to that of the semitic letter. Thus, a further fact can be pretty safely established:
Now, here comes another paradox: the semitic letter has (today) the universal value [z]! This is its value in Hebrew, in Aramaic, in Arabic and even in Syriac. There is no reason to assume that zayin had any other value at the time the alphabet was adopted by the Greeks.Lately, it appears that some semitologists contend that most Phoenician sibilants were affricates, zayin in particular having the value "[dz]" (evidently [d͡z]). Just how this recent discovery came about is not clear. The associated reference to Hackett (on p. 86 of Woodard's collection "The Ancient Languages of Syria-Palestine and Arabia") states: "Since Phoenician is no longer spoken, its phonology must be reconstructed on the basis of (i) transcriptions found in Hebrew, Assyrian, Greek, and Latin writings; and of (ii) comparative phonology of the Semitic languages.". With respect to (i), it would be interesting to know how the transcriptions reveal the pronunciations of zayin as "/dz/" (a certainly weird notation); Hebrew must have essentially used the same alphabet, so I would be surprised if the transcriptions indicate anything other than a correspondence of the phoenician and hebrew zayin; if "Assyrian" refers to Akkadian, which was written in cuneiform, it would be doubtful what conclusions one might draw from the script, since reportedly "Cuneiform was in many ways unsuited to Akkadian: among its flaws was its inability to represent important phonemes in Semitic" (if it refers to Aramaic, then as per Hebrew); the greek and latin transcriptions could not have been prior to or contemporary with the adoption of the alphabet (before VIII BC), as these languages did not have a script to use at that time; after the adoption of the alphabet, I have not heard of zayin being represented by anything other than ζ/z (as in Zama ?), which only proves an affricate value, if one a priori accepts ζ/z as affricate, or σδ/sd (as in Hasdrubal), which is most likely an attempt to write [z] before [r] (cf. Esdras) and almost certainly no indication of an affricate value (note that dental-sibilant clusters, like ts or tz, that might indicate an affricate appear to be used exclusively for ṣade). With respect to (ii), it would be interesting to know what kind of "comparisons" and with which languages led to the "reconstruction" of zayin as "/dz/", as the reflections of the (conjectured) proto-semitic phoneme *z (which appears to have been represented by zayin) in the various semitic languages are unanimously [z] (cf. p. 231: "The early dental/alveolar affricates *ts, *ts’, *dz were deaffricated in most of the attested languages, becoming *s, *s’ and *z, respectively"; we are not informed in which attested languages it was preserved); if zayin was ever an affricate in Proto-Semitic, it must have ceased to be one before its split into the daughter languages, otherwise its parallel independent development into [z] in all the languages would be the most successful conspiracy theory. If, at the time of transmission of the alphabet, the grapheme was adopted for the same value, i.e., ζ=[z] as it is today, all falls into place and no further assumptions need to be made. On the other hand, if the grapheme was adopted to denote a different value, e.g., ζ=[d͡z], then we have the paradox that, after a 5-century adventure, ζ ended up denoting the very value it should have had, had the phoenician phonology been respected in the first place; what a devilish ploy that would have been!The paradox would be even greater, if it is assumed that the phoenician value was [d͡z], as recently posited by some scholars. If the (semitic) alphabet employed zayin for [d͡z] and was transmitted with this value from one language to the other, how can it be explained that it has ended up having the same universal value [z] in all the semitic languages and Greek? What would account for such a co-ordinated development? No common history can explain this, as the greek "transition" to [z] is admitted by the Catholic to have occurred before the hellenistic expansion into semitic territory (mid IV BC vs. late IV BC) and some of the semitic languages were outside the hellenistic sphere of influence.
On the other side of the schism, the Catholics have a weird approach, if they even condescend to provide an explanation for the discrepancy. First of all, the idea of borrowing that letter for the purpose of representing a composite value, particularly [zd], is rejected by the Catholics themselves: "it is difficult to see why a sequence [zd] should not have been represented by σδ instead of by a special sign" (ALLE87, p. 57).Nevertheless, some Catholics do not find the use of a special symbol for a consonant cluster curious, but they counter that ξ and ψ also represented clusters of consonants. We have already seen how likely it is (not) that ξ and ψ were adopted/devised for representing composite sounds (it is another matter whether they were later used by non-ionic dialects to represent their native, non-evolved [ks], [ps] or semi-evolved [xs], [fs]). Furthermore, the usual catholic argumentation is that ξ=[k(h)s], ψ=[p(h)s] were adopted on the model of "ζ for [dz]" (ALLE87, p. 59), not the other way around. In any event, no-one has bothered explain why the same people credited with the discovery of (truly) alphabetic writing (i.e., one sound, one symbol) decided to burden their masterpiece with at least one symbol (be it ξ/ψ or ζ) that stood for a double sound. No satisfactory explanation has been given to-date and the most reasonable conclusion would be that each letter represented a single sound at the time of its introduction (either as borrowing or as innovation) into the Greek alphabet. Strangely, though, they do not extend the same reasoning to the sequence [dz] (d representing the catholic value of δ), which could have easily been rendered as δσ (probably because only few scholars are linguistically-literate enough to understand the difference between the sequence [dz] and the affricate [d͡z]). Thus, the only reasonable catholic value for the zayin-derived ζ is the affricate [d͡z]. The fans of the [zd] value are lost in their identification of the sequence [dz] with the affricate [d͡z], which justifies their "metathetical" fairytale, e.g., ALLE87, p. 57: "it has also been suggested that the affricated combination was at this early period a single phoneme and so preferably represented by a single symbol" ("affricated combination" instead of "affricate"? was it a "double phoneme" at a later period? what phonological reason dissolved the single sound into something else?). Their conviction about the originally affricate value led them down mysterious paths fantasising about mutual mixup of the values, names, positions and shapes in pairs of semitic sibilants (initially in TAYL83, pp. 97-102; see extensive note in the relevant passage). And all that without consideration of the aforementioned paradox that ζ, if initially an affricate in Greek, went a long way to end up to the value of the semitic grapheme, after which it was modelled. None of these supernatural theories is necessary, if one draws the obvious conclusion: ζ was from day one what its predecessor was on day one's eve and what it is today, namely [z].
If it did not occur to the inventors of the Greek alphabet to render the allegedly composite value of ζ as a similarly composite writing (e.g., δσ or σδ), one would expect that it would have at least occurred to its users in the five or more centuries during which it allegedly had a composite value. Alas, a single instance remains elusive: "Τὸ δὲ ζ, ὅπερ παλαιοί τε καὶ νέοι γραμματικοὶ ὀνομάζουσι «διπλοῦν» οὐδέποτε ἐν ταῖς ἐπιγραφαῖς ἐμφανίζεται διγράμματον (« =διπλοῦν»), ἀλλὰ πάντοτε καὶ ἐν πάσαις μονογράμματον. Σπανίως εὑρίσκεται ἐν ἀρχαϊκαῖς ἐπιγραφαῖς Ξ (=ξ) ἀντὶ ɪ (=ζ), και ɪ ἀντὶ Ξ γεγραμμένον.|ζ, on the other hand, which both old and new grammarians call 'double', never appears as a digraph(='double') in the inscriptions, but always and in all of them as a single letter. In ancient inscriptions we occasionally find Ξ (=ξ) written for ɪ (=ζ) and ɪ for Ξ." (ARVA37, p. 70). The following further fact can, therefore, be safely established:
It may, at first, look like this contradicts Ζ4 above; however, they relate to rather complementary facts: Ζ4 refers to the use of ζ for expected (but not attested) σδ (for instance ΑΘΗΝΑΖΕ for conjectured *Ἀθήνασδε, with no attested †ΑΘΗΝΑΣΔΕ), whereas Ζ9 refers to the want of renderings of established ζ with σδ or δσ (for instance, no instances of †ΣΔΥΓΟΝ or †ΔΣΥΓΟΝ for the normative form ΖΥΓΟΝ). The lack of digraph variants for ζ is perfectly analogous to the similar lack of variants for σσ (Attic ττ). We have seen Allen's (reasonable) verdict on the latter: "it is scarcely credible that the existence of an affricate sound would not have been revealed in any inscriptional spelling outside those mentioned above (e.g. as τσ)". For reasons of consistency, one would expect the Catholics to draw the same conclusion in the case of ζ, but they do not even acknowledge the fact, much less make the inevitable association. They seem so content with the few cases of their imaginary ζ for σδ (discussed under Ζ4) that they do not feel the need for seeking the reverse substitution.
To wrap up the lessons from the use of the script, let us note that composition has provided many instances of σ+δ: προσδίδωμι, προσδοκῶ, εἰσδύω, etc. Had ζ been equivalent to σδ, it could hardly be avoided that one of the myriads of users of the alphabet substitute this cluster σδ by ζ (i.e., Ζ9's reverse). However, not one instance of ζ for normative σδ is reported (in the case of -ζε for expected -σδε, the former rather than the unattested latter is normative; Threatte's two imaginative examples are not ζ for σδ, but σζ for unverified σδ and ζδ for ζ). The usual counterargument that κ+σ (actually, ἐκ+σ) is also not rendered by ξ does not hold water; Threatte reports (THRE80, p. 586) many instances of ΕΧΣ (Old Attic) and ΕΞ (Ionian) for ἐκ+σ in Attic inscriptions.
Before leaving the script-related evidence behind, it might also be worth noting that the orthodox values δ=[ð], ζ=[z] more easily explain some inscriptional instances of (lisping-like) use of δ for ζ and vice versa. For example, Threatte reports "rare cases of Δ for Ζ on graffiti from the Agora: ἐπιτραπέδι[α Agora 21, p. 10, no. B 13 (350-300)" (THRE80, p. 550). A similar reverse lisping is to be found in Elean of VI and V BC with ζ being used for expected δ, such as "ζίκαια, ϝειζώς, ζίφυον, Ὀλυνπιάζων, Ζί, ζέ, οὐζέ, ζᾶμον" (CHAT02, p. 441).Chatzidakis concludes from this interchange that δ was [ð] in Elian but [d] in Ionian, whence the Eleans found the Ionic letter unsuitable for their sound and used ζ(=[z]) instead. However, such a conclusion is narrow-minded: from the use of σ in Tsakonian for general (modern-)greek θ, e.g., σέρι, σάτη for θέρος, θυγάτηρ (CHAT02, p. 458), would he conclude that θ is a fricative in Tsakonian, but not in modern Greek and that the Tsakonians found it unsuitable to represent their fricative value? This is certainly not the case and the simple reason why the Tsakonians do not use θ is because in their dialect the fricative [θ] has turned into a sibilant [s](=σ). It is, therefore, also conceivable (and, in fact, more likely) that in Elean (a dialect related more to Tsakonian than to standard Greek) δ(=[ð]) had turned into a sibilant [z] or at least come very close to it and was therefore represented accordingly by ζ=[z]. This is also suggested by the reverse spellings δ(δ) for ζ (CHAT02, p. 444). Subsequently, he seems to suggest that the Eleans reverted to using δ again and concludes that this was because the Ionian δ had become a fricative (and, hence, suitable to represent their sound). In other words, Chatzidakis suggests that the Eleans used the letters of the alphabet according to the pronunciation of these letters in other parts of Greece, as if it were a writing system imposed on them, in parallel to their own writing system. But since Elea had and used its own version of the alphabet, this claim is as absurd as expecting the Spaniards to spell their d(=[ð]) as z or s, because it does not agree with the value of the same letter in Italy, France and Germany. For the record, if one needs an explanation for the reversion to the "correct" spelling δ, it most likely indicates a sense of orthography which gained ground in Elis after V BC. Similarly, a semi-educated islander might occasionally spell their assibilated palatal κι ([ci] in standard Greek, [t͡ʃi] or [t͡si] in some islands that have been under italian or venetian influence) as τσι(=[t͡si] in standard-Greek orthography), but an educated one would still write κι, even if it does not correctly(=according to standard-Greek pronunciation) reflect the actual pronunciation.
While the lack of ζ for normative σδ resulting from composition speaks against the established value ζ=σδ, the aversion of Greek to dental/sibilant sequences speaks against the other Catholic value ζ=δσ. Had ζ the latter value, it should have arisen in cases where inflection brings δ and σ together, e.g., ελπιδ-ς→ἐλπίς (not †ἐλπίζ), ελπιδ-σω→ἐλπίσω (not ἐλπίζω, which comes from ἐλπίδ-ι̯ω), ποδ-σι→ποσί (not †ποζί). However, δ (like the other dental mutes) disappears before σ leaving nothing behind (except compensatory lengthening in most cases, evident in ποδ-ς→ποῦς).
This aspect of greek phonotactics is enough for some scholars to declare the cluster δσ(=dz, as they believe) incompatible with greek phonology (thus CHAT02, p. 446); however, this fact is not very conclusive, as it can be explained away by assuming that either
As there is a marginal probability of validity for at least the last assumption, Ζ10 does not rule out the affricate value ζ=[d͡z], as per the Presumed-Innocent Principle; however, it does serve (together with Ζ8, Ζ9 above and Ζ11 below) to reduce the probability of its validity, according to the Incompatibility Principle.
Another internal fact that is used against the "dz sect" is the disappearance of nasals before ζ. This concerns primarily the preposition σύν in composition, e.g., σύν+ζῶ→συζῶ, for they believe that "there was no reason for the rejection of the ν", if "d was the prior element in the compound letter ζ" (BLAS90, p. 117). This feature is reportedly peculiar to συν-, as no other type of composition appears to lose ν before ζ (e.g., ἔνζῳος and μελάνζοφος, although these forms are almost certainly postclassical); but cf. ἑκατόζυγος (not †ἑκατόνζυγος) in Homer. The loss of nasals before ζ is also suggested by etymology, e.g., σαλπιγγ-ι̯ω→σαλπίζω (cf. σάλπιγγ-ς→σάλπιγξ); in this case the nasal is [ŋ] (I cannot think of similar examples with [n], i.e., stems in -νδ-). The relevant fact can, therefore, be stated as
From this, the reasoning is that, if ζ=δσ, the nasals should behave before ζ as before δ; since it does not, ζ≠δσ. The faithful of the "dz sect" have bought into this argumentation and propose that ζ had a normal affricate value [d͡z] and an allophone [z] after nasals (that is to say, if the theory does not agree with the evidence, the theory is only partly given up for the incompatible evidence). As much as I would like to confirm the catholic conclusion (since it disqualifies the only marginally credible catholic alternative), I have to point out that this is yet another example of confusion between [dz] (the sequence) and [d͡z] (the affricate). If a certain sound has a certain behaviour before [d], it does not need to behave the same before [d͡z]; for one, the point of contact of the former is somewhat more forward than that of the latter.Despite the misleading IPA representation of the voiced affricates as [d͡z], [d͡ʒ], [d͡ʑ], their "stop component" (if such can be isolated from the sibilant release) is not exactly [d], but a coronal voiced stop whose point of contact is progressively receding towards the palate (whence the ordering [d]→[d͡z]→[d͡ʒ]→[d͡ʑ]→[ɟ] in The Matrix). This is also the reason why the dental/sibilant sequences are acoustically different from the respective affricates: in the former the dental stop retains its original place of articulation, while in the latter it is assimilated to that of the respective sibilant, i.e., [z], [ʒ], [ʑ], which are clearly ordered from front to back in the IPA. Thus, Ζ11 speaks against the value [dz], but not necessarily against [d͡z]. Nevertheless, a different fate of the stop [d] and its neighbouring affricate [d͡z] (which is practically also a coronal stopIt appears that academic linguistics has finally recognised that the affricates are stops and has started treating them as such (in particular, they advocate "a description of affricates as strident stops"). Kehrein (KEHR02), in particular, has presented a very cogent and illuminating theory under the term "Generalized Stop Approach" (GSA). In a few words, GSA considers that what we perceive as affricates are stops from a phonological point of view, but surface phonetically as stops with a distinct release (strident, lateral or nasal). The strident affricates that are discussed in this chapter are to be considered "strident stops" and the reason these two features appear as "phased", i.e., first the oral closure and then the sibilance, is that "stridency is simply imperceptible during oral stop closure" (p. 9). The most important observation (for the purposes of the present chapter) is probably that the particular realisation of the affricates as stops with a distinct release is due to the fact that "the release phase of stops is the most prominent acoustic cue to perceive phonological specifications in stops", which "explains why stops are realized as 'affricates' [tθ, ts, tɬ, tn] rather than as their mirror images [θt, st, ɬt, nt];" (p. 8). Despite all this progress, the unscientific (and unprecedented) "metathesis of affricates" (which would rather correspond to the replacement of the affricates by their utterly inferior "mirror images") is still the official doctrine of Catholicism.) is yet to be reported; for example, modern Greek has practically levelled the distinction between the sequence nasal+voiced affricate ντζ(=[nd͡z]) and the plain voiced affricate τζ=[d͡z], both being indifferently pronounced [(n)d͡z] by many speakers (the nasalisation of the latter being a case of hypercorrection), but this is probably a result of the state of the (word-internal) voiced dental stop, which is rendered as ντ and often pronounced indifferently as [(n)d]. A possible solution would be to assume a different value for δ, e.g., δ=[ð], but then ζ=[d͡z] could not have derived from δι̯=[ðj]. All in all, Ζ11 does not disqualify the value ζ=[d͡z], but it renders it less probable.
Another type of internal evidence used against ζ=δσ, as well as against the orthodox value ζ=[z], is the reduplication in verbs starting with ζ-, which appear to follow the norm of verbs starting with σC-, wherein the augment is ἐ- (e.g., σκοπέω→ἐσκόπημαι) and not that of verbs starting with σV-, which take the prefix σε- (e.g., σημαίνω→σεσήμασμαι). I am not sure I understand the objection against ζ=[d͡z] in this case. Does it imply that [d͡z] is a single sound (which it actually is) and should be repeated in the augment? In that case, does it mean that stem-initial consonant clusters (like the assumed ζ=[zd]) do not feature in the augment? Then why κτείνω→ἔκτακα, but κτάομαι→κέκτημαι? why ξέω(presumably κσέω)→ἔξεκα, but κλείω→κέκλεικα? why does the imperfect augment of verbs starting with σC- appear as ἑ-, e.g., ἕζομαι (presumably from *σέσδομαι with σε-→ἑ-), but the perfect augment as ἐ-, e.g., στέλλω→ἔσταλκα, instead of ἑ- (i.e., †ἕσταλκα)? and why is the initial σ retained in σείω→σέσηκα instead of following its normal development to a "δασεῖα" (i.e., †ἕσηκα)? The argumentation used from either side is beyond my competence, as I have trouble distinguishing myth(=PIE fairytales buried under an army of asterisks) from reality (=attested forms, morphological necessity). I am afraid I have to defer judgment on this argument until I have a better understanding of the rules governing reduplication (if such rules can be formulated at all).
In the various catholic treatises, only two candidate pronunciations are considered: [zd] and [dz]. The latter is variably considered as a sequence of consonants or an affricate ([d͡z]), in order to justify the contradictory conclusions suggested by the evidence (e.g., [dz] to explain its consideration as "double consonant", but [d͡z] to explain its derivation from δι̯, γι̯ or ι̯-). This charlatanism must stop! At any given time (and place), only one value must have been current, either [dz] or [d͡z]. The present value [z] is only mentioned as a post-classical development and rejected right from the start (if they even condescend to mention it) for the classical and pre-classical language. This prejudice must stop! The traditional value must be rejected on account of evidence and not a priori. Thus, we have to examine four different candidate values for ζ: [zd], [dz], [d͡z] and [z].
The first one, the Catholics' favourite, is (to say the least) absurd! To invent a symbol for a random consonant cluster, at a time where simplification (from ideo- or syllabographic to alphabetic) was the main objective, would be the utter folly. The notion that [zd] would be the value of ζ is only based on a strict interpretation (or probably misinterpretation) of the theories of the roman-time Grammarians (Z2), who have lived centuries after the alleged transition to [z] and, in any event, at most suggest that there is a σ and a δ in ζ. None of them describes ζ as shorthand for (i.e., equivalent to) σδ, much less bother to explain how it related to normative σδ in compounds (such as προσδοκία, etc) and why these retained the spelling σδ.Another point that seems to have no satisfactory explanation under the mainstream-catholic narrative is why σδ in these compounds did not evolve to [z], as is supposed to have happened with ζ(=σδ according to this theory) from IV BC. The aphorism that these were "transparent" compounds has nothing to do with phonology; if ζ were equivalent to σδ and each consonant belonged to different syllables (as the narrative goes, particularly for explaining metre), how would the σδ in προσ-δο-κία differ phonetically from that of ἐλ-πίσ-δω(=ἐλπίζω)? What would prevent the same sound law that (allegedly) converted the ζ=σδ of the latter to [z] to operate on the σδ of the former and also convert it to a [z] (i.e., to a "modern" †προζοκία)? As we have seen, this weird theory of the "Grammarians", the origins of which are obscure, is an obvious attempt to explain ζ's effect on metre (Z1) and is probably instigated by the late-antiquity practise of rendering as σδ the ζ of some dialects where its value was converging towards that of δ or by the fact that [z] combines the sibilant nature of σ(=[s]) and the voice of δ (irrespective of whether that is [d] or [ð]). The older testimonies, such as those of Archinus and Aristotle, who were synchronic with the alleged time of ζ=[zd], actually exclude a composite value for ζ, as they speak of σ being pronounced along the teeth, like δ.
The other evidence provided in defence of the value [zd] are at most circumstantial. Some speculative etymologies that trace ζ to PIE *sd are too few (less than a handful) and too loosely supported by actual evidence (e.g., a single "cognate" Ast) for reaching safe conclusions and, in any case do not preclude the two elements of *sd "fusing" into [z]; the endings -σδε and -ζε, even though they may both have the same etymology, are not interchangeable (i.e., the stonecutters did not have a choice to use one or the other, but in each case only one of them is consistently used); finally, in this type of "evidence" (Z4), a few non-Attic spellings like Θεόζοτος and Διόζοτος are not cases of ζ for σδ, but most likely of ζ for δ. The transcriptions of Iranian names, where ζ is used where we believe that zd is the correct, are isolated (exactly two) and have attested alternative forms with z, which could have very well been the original models; the location and identification of the city Ἄζωτος of Syria mentioned by Herodotus are as certain as those of Atlantis; what's more, there are several concurrent examples of use of ζ for (verified) [z], Zarathustra↔Ζωροάστρης being the most prominent one, which are conveniently silenced by the Catholics; all in all, the transcriptions of foreign names (Z3) are not suggestive of a value ζ=[zd], but rather of ζ=[z]. The pun "ὦ Βδεῦ Δέσποτα" (Z6) is a boomerang, as it makes perfect sense under the (entire) orthodox model, but hardly any under the assumed ζ=[zd]. Most importantly, the absolute want of inscriptional misspellings σδ for ζ (Z9) virtually diminishes the probability of ζ=σδ, the identity upon which the entire catholic edifice is based.
The only reason why the value ζ=[zd] is almost universally accepted today by the majority of scholars is because someone (Blass?) told them so and convinced them with the childish arguments discussed above. They do not even bother check this against the more important evidence and even devise unscientific methods for justifying it. In order to overcome the objections that [zd] could not possibly be the product of phonetic development of δι̯, γι̯ and ι̯- (Ζ7; the OCS development *dj→[ʒd] might explain a derivation of [zd] from δι̯, but it would explain neither the lack of a corresponding [st] from the voiceless original τι̯ nor the transformation of γι̯ and ι̯- to [zd]) and that a special symbol for the consonant cluster [zd] was not necessary (Ζ8), they are willing to accept the affricate value assigned to ζ by their catholic adversaries, but only as "initial" and pre-classical and "explain" their favourite value as the result of a "metathesis" of the affricate, a blatant unscientific lie that is sustained only because (in a true "Goebbelian" practice) it has been repeated so many times with such seriousness by venerable linguists that the "newbies" do not dare doubt it. If [zd] is the result of a metathesis, the original sound could only be [dz] (the sequence, not the affricate), which is equally incompatible with Ζ8 and Z9, as we will shortly see.
To base our judgement on the value of ζ on the statement "ἐκ τοῦ σ καὶ δ|of/from (the) σ and δ" of I BC, which was propagated by the subsequent "Grammarians", is tantamount to using the story of Noah in the Bible, according to which he managed to accommodate a pair of every animal species in the arc, for drawing conclusions about the evolution of species (e.g., to claim that there were far fewer species a few thousand years ago than the hundreds of thousands that there are today) or about the technology of ancient maritime technology (e.g., to claim that the arc was far greater than the aircraft carries of today); only a religious fanatic would subscribe to such a folly. The adherence to the main catholic doctrine ζ=[zd] is the result of a similar religious fanaticism.
For virtually the same reasons, the reverse sequence [dz] is an unlikely value of ζ: why would a symbol be employed for a sequence (Ζ8)? why are there no inscriptional misspelings δσ for ζ (Ζ9)?Allen's constructive imagination has explained the spellings ζζ for ζ in αζζειοι, βυζζαντιοι, κλαζζομενιοι (in attic inscriptions of V BC) as possible attempts "to represent an affricate of the type [dz]; a spelling δσ would, by recessive assimilation, be mispronounced as [ts], and ζ as [zd], whereas a spelling ζζ = [zdzd] would at least include the required sequence [dz]", which he seems to consider as the classical pronunciation in all those "places in Asia Minor" (ALLE87, p. 58, n. 116). Disregarding the details, like the fact that Βύζας, the "eponymous founder of Byzantium", was not an asiatic Greek, but from Greece proper (and there was no reason for the Greeks of Asia to mispronounce his name or that of the city named after him), only a prejudiced mind would think that [zdzd] is an acceptable approximation for [dz], better than δσ (even with "recessive" assimilation of voice) or δζ(=[dzd] according to Allen's phonological model). how would [dz] be derived from γι̯ and ι̯- (Ζ7)? It is not even supported by circumstantial evidence (as does [zd]). As a matter of fact, [dz] is even more unlikely for two further reasons: δσ (the closest graphic equivalent in Greek) is not a valid sequence, as δ and the other dental mutes are dropped before σ (Ζ10), and a nasal before ζ behaves differently than before δ (Ζ11).
[dz] features in the list of candidates only because some manuscripts state that ζ is composed of δ and σ (instead of the usual σ and δ),Blass' claim (BLAS90, pp. 115-116) that "the grammarians make it consist of σ and δ (in that order)" is not entirely correct, as there are manuscripts that comprise the inverse sequence (cf. ROBE10, p. 144, comment on line 11). naively interpreted as indicating the identity ζ=δ+σ, and because it is often (conveniently) used to represent the affricate. Once one is linguistically mature enough to understand the difference between the affricate [d͡z] and the sequence [dz], the value [dz] will have to be dismissed without second thoughts (if not for anything else, because it is incompatible with Ζ10).
The value ζ=[d͡z] is the only catholic value that is halfway credible. It is not equivalent to the unacceptable sequence [dz] and almost no evidence is conclusively against it, as there is always at least a barely plausible explanation. Thus, a value [d͡z] or [d͡ʒ] is a reasonable development for δι̯ and γι̯ and probably (if we draw an analogy with Italian) for ι̯- (Ζ7), particularly under the assumption that δ=[d] and γ=[g]; to use a special symbol for an affricate (Ζ8), which (contrary to what Sturtevant is preaching) is a single sound,cf. KEHR02, p. 11: "Affricates [...] sound like (homorganic) combinations of two segments (stop + fricative), but behave as single segments with respect to phonotactics, sonorancy etc" (emphasis in the original). is neither strange nor unheard of, since many languages have introduced new symbols into existing alphabets specifically for expressing affricates; the lack of misspellings δσ for ζ (Ζ9) is also understandable for an affricate value of ζ, as [ds] (the alleged value of δσ) is not a satisfactory equivalent or even approximation of [d͡z]. The non-equivalence of [dz] and [d͡z] may also explain the (at first sight incompatible) evidence Ζ10 and Ζ11: intolerance to dental-sibilant clusters does not suggest intolerance to the sibilant stops that the affricates are; the slightly different place of articulation (point of contact) of [d] and [d͡z] might explain the different behaviour of nasals before them, as the disappearance of the nasal may be due to the sibilant (not the plosive) nature of the affricate.
If [zd], [dz] and [d͡z] were the only candidate values, then the latter would be the regrettable winner. Regrettable, because it is not problem free. How likely is it that the δσ resulting from inflection evolved to a plain σ (Ζ10) instead of an affricate (and thus spelt ζ=[d͡z] according to the affricate theory)? How likely is it that a language possesses a voiced affricate, but not its voiceless equivalent (i.e., [t͡s])? Why would δι̯ and γι̯ (allegedly [dj] and [gj]) develop to [d͡z] (Ζ7), but κι̯ and τι̯ to σ(σ)? Why would the Greeks use for ζ=[d͡z] the symbol for zayin (Ζ8), which is consistently [z] in Semitic, instead of, e.g., tsade, which is (often) an affricate? And if it is argued that zayin also stood for the same affricate in Semitic, how likely is it that the same symbol has evolved from the same affricate value [d͡z] to its corresponding continuant sibilant [z] both in Greek and in all Semitic languages? If an affricate value is the reason why the Grammarians see a δ and a σ in ζ, why is the assumed sound [d͡z] never misspelled as δσ, but consistently represented by ζ? Taking all these into consideration, the probability of validity of the affricate value is infinitesimal.
The only evidence that may speak for an affricate value is the present pronunciation of z in Italian (Ζ5). However, its significance for the ancient value of greek ζ is not enough to make the likelihood of ζ=[d͡z] considerably higher. Why would Italian preserve the original value of a greek letter better than Greek itself? And if the affricate value were a peculiarity of "italian Greek", as it is often alleged, would the Grammarians (who flourished in the Roman era and did not miss an opportunity to report dialectal divergences) not mention this (noticeable) oddity? How can we rule out that the emergence of an affricate out of the Greek ζ was not an internal development of Latin or Italian? Why let the italian tail wag the greek dog?
It should also be mentioned that an affricate value would not provide a more satisfactory answer to the metric problem (Ζ1). If we accept that position is only produced by two successive consonants (in which case, we would have a hard time explaining the impact of long vowels in metre), then a plain affricate would not do the trick, but a geminate affricate would have to be assumed, conventionally rendered as [dd͡z], which would suffer from most of the problems of [dz], e.g., that it should have at least once been misspelt δζ or τζ.
All having been said, it should also be pointed out that [d͡z] is not that incompatible with the orthodox value [z], as it is essentially a plosive version thereof; in other words, [d͡z] is to [z] as [d] is to [ð], as [g] is to [ɣ], as [b] is to [v] (cf. The Matrix). If we bear in mind that the distinctive property of the voiced consonants δ, γ, β was voice and that duration (i.e., the distinction between a plosive and a continuant) would not serve to further distinguish them from other "competitors", it is conceivable (as is also pointed out in the corresponding section) that the realisation(=pronunciation) of the voiced consonants could vary (for different speakers) between plosive and continuant values without any impact on the phonological model. In that respect, the values [z] and [d͡z] for the voiced sibilant ζ might be considered as practically equivalent (albeit one more "preferred" than the other).
[z] is a nice value: it establishes the symmetry of the phonological model, it is easy to pronounce, it is familiar to the French, the English and many others. But more importantly, it is the value of ζ handed down to us by tradition. So, why doubt it as the original and only value of ancient ζ? The reason cannot be metre (Ζ1); even if we are convinced that only a succession of two consonants makes position, a geminate value (i.e., [zː]) is the straightforward solution. Instead, in the cases where the Catholics condescend to explain why they rule out the traditional value of ζ, the primary reason provided is invariably (BLAS90, p. 115; STUR20, pp. 187-188; only implied in ALLE87, p. 53) the testimony of the Grammarians (Ζ2). But the relevant passages are anything but clear on the nature of ζ, some of them essentially ruling out a consonant cluster. Those that do speak of ζ being composed of σ and δ (but not equal to σδ) are (particularly in light of Ζ9) mere theories, probably instigated by spelling, etymology or phonological dissection, equivalent to claiming that portuguese [ɲ] (spelt nh) comprises a n and an h, that german [ʃ] (spelt sch) comprises a s and a ch and that modern-greek [b] (spelt μπ) comprises a μ and a π. Thus, the rejection of the traditional value and the adoption of an affricate value is due to a lousy interpretation of a lousy theory driven by a strong desire of many (mostly German) scholars to bring ancient-greek phonology closer to their questionable practice (of reading greek texts according to pronunciation rules of their own language). All worked well until Blass decided to make a "contribution", doubted the then Pope, Curtius, interpreted the doubtful theories of Dionysius' even more literally, invented a handful of "confirmatory" evidence and established the most nonsensical value for a greek letter. His successors adopted this value without bothering check all the evidence and even augmented his arguments with the unscientific "metathesis of affricates".
If we understand the fallacy of their arguments and examine all evidence closely, we will realise that not only is [z] compatible with the circumstantial and questionable evidence thrown against it, but it is the only value that explains the most important, solid facts that follow from etymology (Ζ7), origins (Ζ8), synchronic spelling (Ζ9) and phonotactics (Ζ10, Ζ11). There is absolutely no need (and no reason) to assume a different value for ζ for any time period since the introduction of the letter in greek writing (i.e., since the adoption of the alphabet).
If it works (particularly, where everything else fails), don't fix it!
|
|
|
|
|