Go to previous page Go to main page Go to next page

And So the Bible Ends

After presenting the main bibliography, it is time to set forth the main result of my research:

a complete revision of the Greek (and partly the Latin) sibilant system.

The Vision

A notable feature of ancient-Greek phonology (as scientifically reconstructed by the great linguists of the 19th and 20th centuries, e.g., ALLE87, pp. 177-179) is its paucity in sibilants, in that there is a single sibilant Σ=[s].Here it should be noted that the shorthand Ζ=Σ+Δ, like Ξ=Κ+Σ and Ψ=Π+Σ, can clearly not count as a (separate) sibilant, while the allophonic value Σ=[z] before voiced consonants is only proven (by means of misspellings such as ΖΜΥΡΝΑ) for the post-classical language, i.e., when Ζ=[z]. A similar remark can be made for the other classical language, Latin, which presumably also had a single independent sibilant, S.Here, again note that it is assumed that X=C+S.

Theory

The necessity for assuming a different value for Ξ and Ψ in ancient Greek can be logically arrived at by the following considerations:

  1. Status
  2. The aformentioned dearth in a fundamental phonological group (the sibilants) for both ancient Greek and Latin is very atypical of a classical language, particularly compared to the Sanskrit phonological model (e.g., as presented in CHAT02, p. 176), which comprised the sounds ç, ṣ and s.

  3. Analogy
  4. If one seeks a similar example of a language with a single sibilant, one does not need to look further than Spanish. But (modern) Spanish is far from being called a "classical language" and, indeed its single sibilant is a fairly recent development having its origin in the collapse of the sibilant system of Medieval Castilian: ç=[t͡s] (later [s]), s=[ś](=[s̺]), x=[ʃ] (see LLOY87, p. 328 ff.). It is, therefore, reasonable to expect that the sibilant system of classical Greek would be similarly rich.

  5. History
  6. The obvious candidates for the "extinct" sounds are the so-called "double consonants" Ξ and Ψ, which in modern Greek are truly double and stand for [ks] and [ps] respectively, the values also assumed for ancient Greek. However, phonological change is an inherent and inevitable feature of every language and, since there is no reason for Greek to be an exception to this universal rule, one cannot reasonably expect that their (modern Greek) pronunciation has been kept constant for nearly three millennia.

Based on the above, the most reasonable guess is that Ξ (presently regarded as the combination of a velar with Σ) was a sibilant with a rear point of articulation, such as [ʃ] or [ɕ], and that Ψ (combination of a labial with Σ) had a forward point of articulation, such as [ṣ] or [ś]. This theoretically derived conclusion is, indeed, supported by excessive factual evidence.

Evidence

There is overwhelming evidence that Ξ (as well as its Latin counterpart X) was a single sound, most likely [ʃ]. There is less evidence for Ψ, but its value can be analogically inferred.

  1. Origin
  2. It is well known that Ξ, in the places and dialects it was used, has its origin in Semitic samekh (see CHAT02, pp. 152-159): it has its place in the alphabet (between Ν=nun and Ο='ayin) and its (Phoenician) shape (samekh). Samekh has never been described as or thought to be a composite letter; we have no evidence that it ever was anything but a simple/single phoneme. Thus, there is no reason whatever for ancient Greeks to adopt it to represent the combination [ks]One should be reminded that the similar adoption of C in Germanic or Western-Slavic alphabets to represent the combination [ts] was due to C having already assumed this (allophonic) value at the time of adoption., since this combination could be easily rendered as ΚΣ (as, indeed, it did in some marginal dialects). It is often argued that samekh's value is [s] and, since this value had already been taken by Σ, samekh was "recycled" by being assigned a new value. However, it should be pointed out that [s] is the value of samekh (ס) in modern Hebrew, while the most usual value of Σ's counterpart (shin or ש) is [ʃ]; hence, we cannot be sure how shin and samekh were pronounced by the Phoenicians or how there were perceived by the Greeks. For example, if samekh stood for an apico-alveolar s ([ś]), it could have been perceived by the Greeks as [ʃ] but evolved in Semitic to [s], just like Medieval-Spanish apico-alveolar s (which evolved into today's [s]) was most often transliterated using Arabic šīn=[ʃ] (LLOY87, p.339).

  3. Raison d'être
  4. As argued above, if Ξ and Ψ had their modern values, they would be completely superfluous letters, since they could be readily rendered as ΚΣ and ΠΣ and their introduction would go against the spirit of alphabetic writing which is to use the fewest possible number of symbols. Allen's explanation (ALLE87, p. 59) that [ks] and [ps] were the only word-final consonant clusters in Greek does not seem to fare well with Latin practice: the Romans adopted Ξ (in the form of X) as, indeed, several Latin words end in it (e.g., REX); however, despite the fact that many Latin words also end in PS (e.g., PRINCEPS), a letter similar to (already available) Ψ was not adopted; such preferential treatment to one of the two "double consonants" does not make sense, unless the sound represented by Ψ did not exist in Latin, while that of Ξ did (and those sounds could not be [ps] and [ks], as explained); even if it is argued that Latin lacked (a letter similar to) Ψ because Ψ did not feature in the western-Greek alphabet (a copy of which was the Latin alphabet; cf. ALLE87, p. 47, "the Latin alphabet, which is based on a West Greek model"), then Allen's argument makes even less sense, considering that western Greeks, whose language did possess words that (allegedly) ended in [ks] and [ps], used a symbol for one consonant cluster and refrained from using a similar symbol for the other.

    Furthermore, had Ξ been a shorthand for ΚΣ, it should be used for every occurrence of the consonant cluster; however, there are many words that are consistently spelled with ΚΣ (e.g., ἔκστασις, ἐκσκαφή, ἐκσῴζω), but never with Ξ (*ἔξτασις, *ἐξκαφή, *ἐξῴζω). One may argue that these are compound words and it might not seem appropriate to use a single letter that would span two words (or two word constituents); however, this consideration did not prevent the ancients from, e.g., writing Αθήναζε (from Αθήνα(ν)ς+δε) instead of Αθήνασδε or Διόζοτος (from Διὸς+δοτός) instead of Διόσδοτος (STUR20, p. 189; CHAT02, p. 443).

  5. Traces
  6. The original value of Latin X (which was a mere graphic variant of Ξ in the western-Greek alphabet, whence the Latin alphabet, and which we have no reason to believe differed from Ξ) survived (as [ʃ]) in Spanish until about the 16th century (LLOY87, p. 265) and does occasionally in Portuguese, while its monophthongal nature is evident from its representation as ss in Italian (e.g., Alessandro, Massimo). Furthermore, its representation as s or ss in late Latin (TAFE60, p. 59) explains its being regarded a "double" consonant (while it seems that it was actually a "doubled" or "geminate" consonant) and the implied phonetic merger with S is more probable for a monophthongal X.

  7. Comparative linguistics
  8. If one compares the Greek preposition(al prefix) ΣΥΝ and its Latin cognate CVM bearing in mind that there is an established equivalence between Greek Υ and Latin V, as well as between Greek word-final Ν and Latin word-final M, the correspondence of Greek Σ and Latin C is not easily explainable, unless one assumes that the original PIE form was *Xum, X being an "intermediate" sound between [k] and [s], which developed either way in Greek and Latin. Indeed, in Attic, ΣΥΝ is written ΞΥΝ (e.g., in the history of the Peloponnesian War, "ξυνέγραψε", "ξύμπαντες", "ξυγγενὲς", "ξὺν τοῖς θεοῖς", etc.), which is obviously the preserved PIE form and the orthographic variation between Attic and general Greek is paralleled by the existence of dual forms with initial [s] and [ʃ] (rendered as X) in medieval Castilian (LLOY87, p. 265). A similar orthographic variation between dialects is attested for the alternative form ΕΞ/Ε(Σ)Σ (HORR10, p. 32) of the preposition ΕΚ (cf. Latin EX, which has apparently preserved the original form of the preposition).

  9. Transcriptions
  10. While talking to an Iranian classmate, I mentioned the famous Achaemenid king Xerxes (Greek Ξέρξης), but he did not understand who I was talking about. I later found out that his Old-Persian name was actually Ḫšayāršā, which is certainly not pronounced the "reconstructed" ancient-Greek way [kserksɛ:s]. Admittedly no conclusion can be drawn from the vowels, because Old Persian (just like Sanskrit) seemed to lack a character for or an exact rendering of the mid vowels ε ([e]) and η ([ɛ:]). But the verdict for the consonants is incontrovertible: in two instances, š corresponds to ξ; while the first correspondence Ḫš↔Ξ is reminiscent of the Attic rendering of Ξ as ΧΣ (assuming that Ḫ stands for the velar fricative [x] or some similar sound and forgetting that Greek Χ has been reconstructed not as a fricative, but as an aspirate [kʰ]), the second š↔ξ speaks clearly for the value ξ=[ʃ] (š). A less clear, but supplemental proof is the transcription Τάξιλα for Indian Takṣaçilā (STUR20, p. 170).

    The Septuagint, in the book of Lamentations, provides a rather cogent proof of the true value of classical Ξ, when it employs its Old-Attic rendering ΧΣ to represent the Hebrew letter ש (ΡΗΧΣ↔ריש, ΧΣΕΝ↔שין), which is known to have the value [ʃ]; regardless of the ("aspirated" or "fricative") value assigned to Greek X, this peculiar transcription cannot be explained, unless the digraph ΧΣ was used with its classical (Attic) value of [ʃ] (the simple letter Ξ having, probably, developed into the modern [ks] and, thus, being unsuitable to render ש).

  11. Inscriptions
  12. Attic inscriptions before IV BC make use of the digraph ΧΣ (e.g., HORR10, p. 40) for the sound which was later represented by Ξ. Bearing in mind that the former's transliteration in Latin would be CHS and remembering the German convention sch=[ʃ] (e.g., Schule|school[GER]), as well as that chs=[ks], (e.g. Ochse|ox[GER]), it is evident that the transition from ΧΣ to Ξ in Attic was due to a metathesis of chs(=[ks]) to sch(=[ʃ]), the former being the old-Attic value (and possibly explaining the tradition Ξ=Κ+Σ of the Grammarians, which led to a "learned pronunciation" of the letter) and the latter being the classical value of the letter.

    Further inscriptional evidence is provided by a Naxian inscription "where ἔ⸠σοχος Να⸠σίο (Ναξίου) are written with the symbol for the spiritus asper" (BLAS90, p.115), which is to say Ξ is rendered as "hs". Blass' explanation that "actual assimilation [of Attic Χ by Σ into h?!] also took place dialectically and absolutely destroyed the explosive [the Κ of common-Greek Ξ=ΚΣ]" is incomprehensible. Instead, if one assumes that Ξ=[ʃ] and takes into account the English convention of writing sh for the same sound, the representation of Ξ by hs makes much more sense; admittedly, the order of the letters in English is reversed, but it (i.e., hs↔sh) is paralleled by the established (by Allen) correspondence of Greek initial ῥ (rendered in Latin as rh, e.g., rhetor|orator[LAT]) and Armenian or Icelandic hr (e.g., ALLE87, p. 41, hṙetor|orator[ARM] and p. 42, hringur|ring[ICE], respectively); here, as for ῥ (STUR20, p. 165, also endorsed by Allen), the aspiration (h) is a simultaneous feature of the main sound (Σ) and can variably be written before or after it.

  13. Internal evidence
  14. The last-mentioned inscriptional evidence is confirmed by internal evidence in Latin, where X is produced by the coming together of H (not C) and S (in that order): TRAHO→(TRAH-SI→)TRAXI, VEHO→(VEH-SI→)VEXI (TAFE60, p. 23).

    Internal evidence from Greek is to be found in HORR10, p. 23: "in Boeotia, we begin to find the substitution of the typically West Greek velar suffix -ξα- [-ksa-] for 'true' Boeotian -ττα- [ -ttja] in the aorist (past perfective) stem of verbs with an original stem-final dental, e.g. ἐκομιξά-μεθα [ekomiksá-metha] for ἐκομιττάμεθα [ekomittjá-metha] (<*[ekomitj-sá-metha])"; on p. 19, Boeotian [ttj] has been identified as [tʃ]. In view of this, it is evident that Ξ=[ks] makes much less sense as an approximation of the palatal affricate [tʃ] than does Ξ=[ʃ] (by means of the well-known disappearance of dentals before sibilants in Greek).

  15. Grammarians
  16. A more direct testimony about the nature of Ξ and Ψ can be found in the works of the ancient Grammarians, who (contrary to what is widely believed) seem to favour a monophthongal rather than a composite value for Ξ and Ψ:

Conclusion

As seen, one can find plentiful evidence in eight different categories, which is much more than the amount of evidence provided in support of diverging features of ancient-Greek phonology.For instance, Allen provides (ALLE87, pp. 35-37) one hint (use of γ for indicating the velar nasal) and two and a half kinds of evidence (one obscure testimony of a Roman Grammarian, one "anomaly" in conjugation and a single, albeit doubtful, case of preservation in the modern language) in support of his contribution that "the pronunciation [ŋm] is recommended for γμ in all cases". In the face of such abundant evidence, the "consensus view" that Ξ=[ks] and Ψ=[ps], i.e., that Ξ and Ψ had their modern Greek values, is untenable. Other than the present-day pronunciation,which is certainly no proof that Ξ=[ks] and Ψ=[ps], if not a proof to the contrary, since Greek cannot have been a lone exception to the law of sound change and kept the same pronunciation for 3000 years. there is little evidence that may be cited in support of the modern-Greek values. For example, the Grammarians' description of these two letters as "διπλᾶ|double" does not prove that they were composite.Remember that the Grammarians apply the same term for Ζ, which is known (ALLE87, p. 58) to have already become [z] (or [z:], whence its consideration as "double") long before any of the Grammarians was born.

How could this obvious factNote that no evidence provided above is new; I used only what I found already written in various books, the sources always acknowledged. have escaped all the renown scholars that established the reconstructed ancient-Greek pronunciation?At first sight, it appears that Pernot was the only other person who, based on the use of a single symbol for the sounds rendered elsewhere as ΚΣ, ΠΣ or ΧΣ, ΦΣ, suggested that "chez eux [=les Ioniens] ks et ps se prononçaient chacun comme un son simple|each one of ks and ps was pronounced by them [=the Ionians] as a simple sound" (PERN21, pp. 68-70); however, in the note accompanying this statement, he appears to assign to this "simple value" the quality of an affricate ("Il en est encore ainsi en grec moderne. On trouve de même, en Francais, le mot médecin prononcé dialectalement métsin, avec un ts qui est en réalité une consonne simple; et bien des gens disent accent, absent, avec un ks (x) et un ps simples.|It is still like this in modern Greek. One similarly finds, in French, the word médecin pronounced dialectally métsin, with a ts which is actually a simple consonant; and also some people say accent, absent, with a simple ks (x) and a simple ps.")!! It is evident that his (erroneous) proposal for a "single sound" of Ξ and Ψ is substantially different from the (truly) monophthongal pronunciation that I propose above. I believe that they were too busy fighting the conservative and chauvinist iotacists (and sometimes each other), that they simply did NOT look into it; perhaps they were even fooled by the Reuchlinian arguments that naively interpreted the word of the ancient Grammarians in light of the modern-Greek pronunciation.

I believe I have opened a new road in the study of ancient-Greek phonology. Had I been a university professor with a good reputation or some influenceA number of followers having a tendency for fancy things, particularly for anything that goes against tradition, would also help., my tenure or renown would be guaranteed.

Eat your heart out, Herostratus!

The Hallucination

If you were convinced by the "reasoning" provided above about my "reconstructed" pronunciation of Ξ (and Ψ), then you are liable to fall for any kind of elementary argumentation put forward by real "scientists" in support of the Catholic or Orthodox model and have probably a lot to learn from this Website; keep reading.

If you were not convinced by my "reconstruction" because you could identify the weak points in the argumentation, then you should either also start doubting about the "scientific reconstruction" (if you have not already done so) or contribute to this Website (if you have already doubted the great masters of "reconstruction").

If the reason for not believing in what I wrote is merely that I am not a professional linguist or university professor or that most scholars accept the traditional values of Ξ and Ψ, then you are a man of faith who likes to be told by some expert what to believe in and faith is of little value in this Website; may I suggest a different topic for you?

If you are a true "scientist" (i.e., professional linguist) who is concerned about the veracity of my arguments, worry no more, for I want to make a ...

Confession

Before anyone takes this "proof" seriously, I admit it myself that most of the above argumentation is bogus. The funny thing is that there was no falsification of evidence, but the argumentation was based on real data, while the spurious part was, in most cases, the reasoning. Because Allen may proclaim that "The results of any historical study are only as valid as the evidence upon which they are based" (ALLE87, p. xiii), but this is only half the truth; the other half is the completeness of the evidence, as well as the validity of their interpretation.For instance, it is not enough ☺ to look at "plough" and "night" and conclude that gh is silent in English.

The sad thing is that I have used as many of the tricks I learned from the great masters as possible:

  1. Appeal to Prestige
  2. The fact that a language is "classical" does not mean that it is superior in any respect, much less in phonology. Phonological "dearth" may as well be a feature of a classical language (for instance, Sanskrit lacks symbols for [e] and [o], all "original" [e] and [o] having been lowered to [a]; cf., e.g., CHAT02, p. 33 and 177).

    The Model
    WIANPR: "Η ποικιλία των φθόγγων ι, η, υ, ει, οι, υι, ῃ που αντιστοιχούν σε νεοελληνική προφορά [i] και των ο, ω, ῳ με προφορά [o] θα αποτελούσε αδικαιολόγητη και περιττή πολυτέλεια για την κλασική γλώσσα|The variety of sounds [symbols?] ι, η, υ, ει, οι, υι, ῃ that correspond to modern-Greek pronunciation [i] and of ο, ω, ῳ with pronunciation [o] would constitute an unnecessary luxury for the classical language". As a matter of fact, this is a luxury for ANY language, including present-day Greek (not to mention English, French, etc), but this has not prevented (modern) Greeks to pronounce all of them the same.
  3. Analogy as Proof
  4. A certain development in one language (here, the development of Latin into Medieval Castilian) does not necessitate that all languages should also work the same. For example, although the Germanic sound [v] appears to have, in most cases, originated from [w] (cf., the English and German pronunciations of W in, e.g., wine/Wein, word/Wort, wax/Wachs), it would not be correct to identify the original pronunciation of Greek Β (the source of most present-day [v] sounds) as [w] (without, however, disregarding the fact that, in some cases, it does originate from, e.g., the digamma Ϝ and the second element of the diphthongs ΑΥ, ΕΥ, ΗΥ, which might have formerly been pronounced as [w]). Citing analogous phenomena from other languages can only serve to demonstrate that a particular claim is not an impossibility (as, e.g., in ALLE87, p. 27, regarding the clusters of "aspirate consonants").

    The Model
    The use of analogy as proof is a very widespread tool. One example is the extrapolation of the evolution [u]→[y]→[i] of the pronunciation of Greek Υ, based on the evolution [u]→[y] of Latin V into French u (which has not - yet? - developed into [i]). Another example is the conclusion that the result of assibilation of di/δι is [d͡ʒ] or [d͡z] based on giorno|day[ITA] from DIVRNVS[LAT] (and disregarding, e.g., jour[FRE], jornada[POR]). A further example is discussed below.
  5. Promoting Frequent Phenomena to "Laws"
  6. The ubiquity of sound change in the world's languages does not render it an inevitability, particularly for individual sounds; in other words, a certain sound in a certain language does not have to undergo sound shift every few hundred years or so (even though it would not be surprising if it did). For example, there does not seem to be any doubt that A has retained the same pronunciation from ancient Latin all the way to modern Italian (similarly for Α from Attic to modern Greek).

    The Model
    ANNI08 (objecting to Caragounis' claim that Attic phonology was essentially identical to that of modern Greek): "[Greek] is, we are expected to believe, completely unlike any other language in human history known to us, in that it alone has been immune to sound change for nearly two and a half millennia". According to the most conservative views, all Greek sounds have defied the laws of sound change for at least a millennium (the last change, Υ/ΟΙ→[i], presumably took place in X AD) and almost all of them (i.e., with the possible exception of Υ/ΟΙ) for at least a further 5-6 centuries before that; in other words, it is universally admitted that Greek phonology is "stagnant" for at least the last 10-15 centuries. One (and a half) millennium is already an eternity in terms of phonological stability and immunity to sound change; why two eternities (2-3 millennia) would be more of a paradox is something that only those of... little faith understand. ☺
  7. Thinking Modern in the Past
  8. A modern linguist would not have felt the need to use a special symbol for [ks], because he would have immediately recognised it as a consonant cluster, mostly thanks to knowledge gained from century-long research and even longer practice, but the ancient users of the alphabet neither had access to that knowledge nor were linguists. As I have written elsewhere, "it is quite improbable that the ancient Greeks (the first users of a full alphabet) suddenly grasped all complexities of phonemic representation, something that eluded the users of previous writing systems". Thus, one must allow for imperfect and inconsistent representation of the same sound, e.g., that some Greeks considered [ks] a new phoneme (and used Ξ to represent it), while others correctly identified it as a cluster of independent phonemes (and rendered it as ΚΣ).

    The Model
    BLAS90, p. 10: "in Attica towards the close of the fifth century the entire system was absolutely changed. Here was the opportunity in those cases, where the living sound had here and there deviated from the writing, to bring them again into harmony". Here, Blass envisages a spelling revision of the kind recently undertaken by Spanish and German authorities, but forgets that (some features of the) traditional spelling survived countless revisions before the last ones and that new spelling does not always indicate new pronunciation (cf. his own spellings BLAS82, p. 2, "mußte", p. 6, "andre" vs the modern spellings "musste"[GER], "andere"[GER]). In the particular case (Euclidean reform of 403 BC), the issue was not the change of "the entire system", but merely the adoption of the Ionic alphabet (cf. Suida's description that "τοὺς δὲ  ̓Αθηναίους ἔπεισε χρη̂σθαι τοι̂ς τω̂ν  ̓Ιώνων γράμμασιν|He persuaded the Athenians to use the letters of the Ionians [online translation]") and its spelling conventions (basically, abolishing the use of Η as πνεῦμα and writing ΕΙ/Η, ΟΥ/Ω, Ξ, Ψ for previous Ē, Ō, ΧΣ, ΦΣ), namely a pre-established (by others) system of writing conventions, which probably reflected the current pronunciation better, yet not necessarily perfectly.No reason for the reform is given in the much later ancient account (cf. ARVA37, pp. 42-45).
  9. PIE in the sky, PIG in a poke
  10. The most popular sport in linguistics is the "reconstruction" of words of a hypothetical "Proto-Indo-European" (PIE) or "Proto-Indo-Germanic" (PIG) language, which (allegedly) is to the various Indo-European languages as Latin is to the Romance languages. How this "reconstruction" is achieved is rather vague, although some general guidelines are provided. In "reconstructing" the ancestor of ΣΥΝ and CVM, I tried to follow the method described by Aitchison (AITC04, p. 176), namely identify the common or corresponding sounds and try to reconcile the divergences with some abracadabra. My approach is faulty in both steps: it may be that the correspondence between Latin and Greek is often V↔Υ and -M↔-Ν, but Latin short V more often corresponds to Greek Ο (which is always short), particularly in the endings of the nominative singular where -VS↔-ΟΣ and -VM↔-ΟΝ (the latter also coinciding with the part of CVM that "agrees" with ΣΥΝ); furthermore, regarding the discrepancy of the first letter (C and Σ), a sound "between" [k] and [s] might seem a reasonable guess that would produce C=[k] in one language and Σ=[s] in another, but [ks] (i.e., *ksum) would also do the trick.Incidentally, the reconstructed PIE word for Greek ΣΥΝ appears to be *sṃ, the voodoo that would explain why the syllabic sonorant ṃ developed in that case to ΥΝ and in the case of septṃ to Α (e.g., AITC04, p. 176) being beyond my competence. On the other hand, the origin of Latin CVM is allegedly *kom, since it is curiously identified as "cognate" not with Greek ΣΥΝ, but "with German ge- ('with', collective prefix) and gegen ('toward, against'), English gain-, Russian ко (ko, 'to')" (one has to admit that the resemblance in terms of form, as well as meaning is... striking)!!! I also realised that the Greek version of Wiktionary does identify CVM as a possible cognate of σύν. It is evident how certain the identification of cognates among the various IE languages is and how well defined the procedure is. I rest my case.

    The Model
    In (almost) any work on ancient phonology one can find many references to PIE prototypes that (allegedly) "spawned" the later forms (in Greek, Latin, Sanskrit, etc.) and suggest their pronunciation. It is evident that this is a recursive argument, since the reconstruction is usually based on the pronunciation of these later forms, which is assumed known and can obviously not prove itself. For this reason, this kind of argument has been slightly modified to allude to a vaguely "reconstructed" PIE form by referring to "cognates" in as few as one other language.
    A typical example is the curious etymology of "ὄζος|branch" for which "it is admitted by all that ὄζος comes from [*]ὄσδος (Germ. Ast)" (BLAS90, p. 119), an explanation that has also made its way into official lexica, such as the vague reference "Cf. Goth. asts, Germ. ast"The further example "Goth. asts" was apparently introduced to reinforce the claim, but the "cognate" is still drawn from the same language family, even if one considers further Germanic forms (e.g., "ast" is also the Saxon and Dutch word). under the entry ὄζος in Liddell-Scott. This argument was introduced to "verify" the (preconceived) theory that, in ancient Greek, Ζ=[zd], since the identification of ὄζος and Ast as "cognates" suggests the possible correspondence of ζ and st; however, due to the absence of any similar "cognates" in other languages,I am referring to the cognates identified at the time that the argument was put forward by Blass. Today, it is considered that the Old-Armenian word ոստ (ost) is a further cognate. Its resemblance to the German form is certainly striking, but it is still not clear to me how one can rule out accidental coincidence and be certain that these (two or three) words are the only surviving remnants of an assumed PIE word. the assumption that ὄζος and Ast are related is valid only after having accepted that Ζ=[zd], i.e., the very fact that is to be proved! Had ὄζος been identified as "cognate" with, e.g., RAMVS[LAT] or СУЧОК[RUS]|Russian, the suggested value of Ζ could be (something similar to) [m] or [t͡ʃ]. In fact, the latter would be the best "defence" for the German pronunciation of Ζ as [t͡s] (cf. BLAS90, p. 116), particularly if the pair of "cognates" ἕζομαι|I sit-sitzen[GER] is also taken into account. ☺
  11. Seeing signs
  12. The fact that a letter (X) or its evolution (SS) happened to have a value identical ([ʃ]) or similar ([s:]) to that assumed for an earlier stage of the language is no proof for the validity of the assumption. In this particular case, the origin of Medieval-Castilian x and modern-Italian ss may be (and most probably is) regular phonetic development and not preservation of any "original sound".Incidentally, I have not seen any convincing evidence for x=[ʃ] in Medieval Castilian. The usual example of the French, Italian, etc renderings of "Don Quixote"[SPA] as "Don Quichotte"[FRA], "Don Chisciotte"[ITA] etc (also in LLOY87, p. 343) may not be as much of a proof as one may think (based on the modern pronunciation of these languages). Were Medieval-Castilian x the velar voiceless continuant [x] (i.e., its 19th-century value - cf. the spellings "Texas", "Mexico" vs modern Spanish "Tejas", "Mejico"), how would the languages lacking this sound (such as French and Italian) represent it? Wouldn't (the spelling conventions for representing) [ʃ] be more suitable for this purpose (cf. the French pronunciation of "ich"[GER]=[iç] as [iʃ] and CARA95, note 107, where scholars that have a hard time with Greek χ pronounce χάρις as "sharis")? Furthermore, it is still less clear (to me) what the 16th-century value of French ch was. Since it originates from Latin C in its (purely) velar value [k] (e.g., "cheval"[FRA] from "CABALLVS"[LAT]) and its present value is [ʃ], isn't [x] (or [ç]) a reasonable intermediate value that might have been current any time between Latin and modern French (in which case the Italian spelling may be based on the French one; cf. the use of double t in both the Italian and French versions for single t in the original Castilian)? The "retention of an original sound" is a very common claim and, most of the times, it is just that: a claim (without any provided proof).Here, it should be pointed out that I mainly refer to claims relating to isolated exceptions (e.g., "ξερός" and "θερί" in "The Model") or one of many diverging developments of an original sound (e.g., Castilian x=[ʃ] vs French x=[ks]) and not to nearly-unanimous or unique tradition. The situation is reminiscent of the pious men and women who see, e.g., a tree in the shape of the cross and take it to be a sign from God, whereas it might simply be... a cross-shaped tree!Or, in the words of Chatzidakis (CHAT02, p. 301): "έκλαμβάνουσι μὲν πολλάκις ὡς Ἥραν τὴν νεφέλην|many times they take the cloud to be Hera", with an obvious reference to the story of Ixion.

    The Model
    BLAS82, p. 19: "ebenso wie vor diesem Buchstaben [r] altes η e sich in der Schreibung ε gehalten hat: ξερός [ξηρός], θερί θηρίον u. a. m.|just as before this letter [r] old η e has been preserved in the form of ε: ξερός [ξηρός], θερί θηρίον and so on". The fact that some modern-Greek words have ε where they used to have η, was enough evidence for Blass that the original sound of η (was [ε:] and) was preserved in those cases (while everywhere else it became [i]). However, in the next edition he concedes (BLAS90, p. 21) that "the fact of an η appearing as e ε before r in unaccented syllables (ξερός, θερί for θηρίον etc.) is not due to a retention of the ancient sound, but to a modern phonetic law, according to which every unaccented ir (ιρ, ηρ, υρ) becomes er, as keryaki κυριακή" (emphasis mine). Although his (actually Psichari's) "modern phonetic law" is an inaccurate term (it is rather a "predisposition", since one can readily find many modern words that defy this "law", such as όνειρο, άκυρος, κακομοιριά, πειρατής), he (correctly) acknowledges that the particular examples might have a different origin from that which he (hastily) assumed a few years earlier.Another explanation is provided by Rhankabes, who suggests (RHAN81, p. 23) that "sie stammen von Wurzeln mit ε ab; z.B. ξερὸς (aber auch bei Homer) für ξηρὸς von ξέω|they stem from roots with ε; e.g. ξερὸς (but also in Homer) for ξηρὸς from ξέω" (which I can neither confirm nor reject). Further examples are discussed in the corresponding chapter where the validity of evidence from the modern language is examined.
  13. Redefining Linguistics
  14. The lamest of all arguments above is the one based on an alleged... metathesis into a trigraph! Metathesis is a linguistic phenomenon that involves the transposition of sounds (cf. Argelia[SPA] for Algeria),Contrary to Allen's assertion (ALLE87, p. 57, "such metatheses are of a particularly common type"), metathesis of proximate but not adjacent sounds is the most common type. As a matter of fact I cannot readily think of one example that involves the transposition of successive sounds (a possible example may be βγαίνω from ἐκβαίνω, even though the necessary intermediate form *ἐγβαίνω is not attested, as far as I know), much less of a consistent transposition of the same two sounds in all environments (which is a prerequisite for the validity of the proposed "metathesis" chs→sch). not letters (in other words, "ugess", "cuop", "htere" are not the results of metathesis in "guess", "coup", "there", but mere... typos); hence, a shuffling-around of letters cannot be justified as a case of metathesis, unless they individually define the same sounds in their old and new positions, which is not the case in chs (where ch=[k] and s=[s]) and sch (where sch=[ʃ], not [sk] as it should be in a true metathesis). Furthermore, there is absolutely no evidence to support the allegation that Attic ΧΣ metathesised to ΣΧ, which was either pronounced in the same way as the German sch(=[ʃ]) or (later) monophthongised to Ξ=[ʃ].cf. Allen's objection to his own(?) theory (ALLE87, p. 54) that PIE (word-initial) *y developed to [h] through the (intermediate) prehistoric stage of a palatal (voiceless) fricative [ç], by observing (in note 105) that "the comparable Armenian change of cl. [y] to mod. [h] in historical times (e.g. Yoyn 'Greek' = mod. [hun] etc.) shows no evidence of any such intermediate stage"; how he is sure about all Armenian pronunciations (not spellings) from cl. to mod. times is another matter. The argument would be even lamer, had I suggested a metathesis of German sch (=[ʃ]), i.e., a trigraph for a monophthong, into chs(=[ks]), i.e., a true consonant cluster, which would amount to defining a "metathesis" of the guess→ugess type mentioned above and to establishing a new branch of linguistics, let's call it... "linguinistics".

    The Model
    The most typical example of "linguinistics" is the frequent allusion to metathesis of an affricate, particularly [dz] (actually [d͡z]) to [zd]; however, I will leave its discussion to the respective chapter on the pronunciation of Ζ. Here, I will just point out how misunderstood is the notion of an affricate, even in the works of scholars who write on phonology.
    An affricate is not a mere consonant cluster (as are, e.g., [ks] or [kt]), where the consonants are spoken in succession, but rather a blend of two sounds (a stop and a continuant) sharing the same place of articulation.Elsewhere I have explained the difference between consonant clusters and affricates by pointing out that [ks] (="x") is a true double consonant ("Mac sale" and "Max ale" sound the same), but [t͡s] (also denoted [ts]) is not necessarily equivalent to [ts], as is evident when [t] and [s] are on opposite sides of a word boundary (cf. "it seals" and "its eels"). This is not a mere contrast of heterosyllabic [t.s] vs tautosyllabic [ts] (the syllable syllable boundary [.] having, anyway, no associated sound); the point of (tongue-mouth) contact of [t͡s] is not as forward as that of [t]. A further example regarding the distinction of [t͡ʃ] and [tʃ] in English is provided by Wikipedia. The most common affricates are sibilants that combine the properties of a dental/alveolar plosive (generally transcribed [t] or [d]) with those of a sibilant (usually [s], [ʃ], [ɕ] or [z], [ʒ], [ʑ] respectively). Their pronunciation is comparable to that of the (real) aspirates (cf. also KEHR02, p. 212): in pronouncing the aspirate [tʰ] the plosive ([t]) is released into an exhalation ([h]), whereas in the case of the affricates [t͡s], [t͡ʃ], [t͡ɕ], the plosive ([t]) is released into a hiss ([s], [ʃ], [ɕ]).Allen, in his "Phonetic Introduction" (ALLE87, p. 4) defines an affricate as "the gradual release of a stop" and states that "English examples are palato-alveolar as in chest (voiceless) and jest (voiced)". Admittedly, the (audible) difference between an "affricated" and a separate pronunciation of the two "elements" (e.g., between [t͡s] and [ts]) is very small (and possibly imperceptible for the untrained ear), but it exists and is as small as the difference between, e.g., diphthongal and disyllabic pronunciation of two vowels or nasal vowel and vowel+nasal.It appears that the present stance of linguistics towards affricates with a sibilant component is to consider them strident (=sibilant) stops (e.g., KEHR02, p. 11), a claim that I have made for the Greek affricates. According to one theory (KEHR02, p. 4), affrication is merely a phonetic mechanism for making small differences in place of articulation (as is the difference between the point of contact of [t] and that of [t͡s]) more prominent or noticeable.
    No less a scholar than Sturtevant, while commenting on an ancient theory that Ξ and Ψ are called "double consonants" because they occupy the room of two letters, declares that "it may represent a theory based upon orthography and as false as the opinion of most speakers of English that the first letter of Jane represents a single sound" (STUR20, p. 188), obviously implying that English j(=[d͡ʒ]) is a sequence of two sounds, namely [d] and [ʒ]. He disregards the fact that the affricates are represented by single letters not only in English (where the present pronunciation of j is a phonological development, after the orthography had already been fixed), but in many other languages (e.g., c|Slovene, z|German, ц|Russian, צ|Hebrew, ţ|Romanian for [t͡s]), including some that devised single letters for already established affricates. Less than a decade after his book, the Turkish reformers ignored Sturtevant's observation and chose to "burden" their alphabet with two unnecessary (according to S.) letters, C(=[d͡ʒ]) and Ç(=[t͡ʃ]), and to "mislead" the public, instead of using the more "correct" digraphs DJ and TŞ respectively! Even in languages where [d͡ʒ] and [dʒ] do not contrast phonemically, the affricates (particularly the dental ones) are not equivalent to consonant clusters (like Ξ=[ks] and Ψ=[ps]) and Sturtevant's comparison is unfortunate and his verdict on English J (I firmly believe) erroneous (unless by "single" he means "uniform").
  15. Perpetuating a Hoax
  16. Blass cites the Naxian inscription, where allegedly Ξ=hs, without (evidently) having seen it, but with a mere reference to "Röhl no. 407=Bechtel no. 23". I came across the (original facsimile of the) inscription in question while reading the very instructive book by Arvanitopoulos (ARVA37, p. 78; the inscription is also available online) and I realised that the alleged use of "the spiritus asper" to write Ξ as hΣ was not supported by the facts: where a Ξ would be expected, we find the "digraph" glyph for hs (when the direction is from left to right) or glyph for sh (when the direction is from right to left), the first element of which is in the form of an empty square (square) not used anywhere else in the inscription;Note that the second elements of the left-to-right and right-to-left variants of the digraph have different shapes; in the former, a three-line zig-zag letter is used, which coincides with the sigmas elsewhere in the inscription, but the latter comprises a four-line zig-zag letter, which looks like present-day sigma (Σ) but is not found anywhere else in the inscription. a different symbol, an elongated square with a line in the middle (glyph for Η), similar to a "squared" Θ, is found mostly where an Η in its vocalic function would be expected (however, in two occasions where a long Ε would be expected, a plain Ε is found); in one case, it may stand for Ἑ ("spiritus asper"+Ε) and in its more curious occurrence it is squeezed between a Φ and a Ρ and is apparently considered to be Ρ's "aspiration" induced by (the "aspirate") Φ. From these facts, it does in no way follow that the "empty square" stands for "the spiritus asper".The same spelling is found in another naxian inscription, where in lieu of the ξ of Νάξιος we find the "digraph" glyph for hs (written from right to left), whereas glyph for Η is used for the η of Εὐθυκαρτίδης (Εὐθυκρατίδης?) and for the "rough breathing" of ὁ.
    What the symbol stood for is, in absence of other evidence, a mere matter of conjecture. As a matter of fact, one could consider several alternatives, e.g., that it stands for a squared variant of Ϙ (a grapheme used sometimes in lieu of Κ) or for a sound that is imperfectly rendered by the Greek Κ and Attic Χ in the digraphs ΚΣ and ΧΣ respectively, before reaching the awkward conclusion that it stands for [h], which rarely combines as the first element of a composite sound (not digraph), certainly nowhere else in Greek.
    An interesting interpretation is that of Conrad, who appears to regard the empty square square as a symbol for ξ itself (e.g., ἔξσοχος for ἔξοχος, the σ being written pleonastically, under the assumption that ξ=[ks]).
    Despite this realisation, I went ahead and used this dubious statement of Blass' to support my assertion that Ξ[ʃ].

    The Model
    I was not the first to notice this discrepancy between Blass' description and the actual text (obviously the discrepancy is due to Röhl and/or Bechtel). A mere decade after Blass' book, Chatzidakis observes that "οἱ Νάξιοι δὲν γράφουν ἀληθῶς ἐν τούτοις τὸ σημεῖον glyph for Η ἀλλ’ ἕτερόν τι, το square|the Naxians do not really write therein the sign glyph for Η but another one, the [sign] square" (CHAT02, p. 453) and two decades later Sturtevant gives a more accurate description: "In early Naxian inscriptions (IG xii v, p. xxiv = SGDI 5419, 5423) ξ is represented by the combination glyph for hs" (STUR20, p. 156). Nevertheless, they both perpetuate the allegation that the two symbols are related ("ἤτοι μεταβεβλημένον|namely altered [glyph for Η]", "square is probably a differentiation of glyph for Η, the character which in this inscription denotes h, η and hε", respectively) and use it to substantiate mere speculations (cf. "Τὰ ἐν Ναξίᾳ ἀρχαιοτάτῃ ἐπιγραφῇ τῆς Ϛ´ ἑκατοντ. ἀναγινωσκόμενα ΝαΗσίō, ΦΗράΗσō, ἔΗσοχος ἀντὶ τοῦ ἐν νεωτέρᾳ ἐπιγραφῇ Νάχσιος, Ἀλχσήνōρ φαίνονται ὅτι μαρτυροῦσι πνευματώδη τοῦ χ προφοράν, ἀφοῦ τοῦτο διὰ τοῦ Η παριστᾶται|The readings ΝαΗσίō, ΦΗράΗσō, ἔΗσοχος in a very ancient Naxian inscription of the 6th century instead of the Νάχσιος, Ἀλχσήνōρ in a more recent inscription appear to attest to a spirant pronunciation of χ, since that is represented by Η"Here, Chatzidakis appears to be (reciting and) discussing someone else's assertion, with which he only partly agrees. and "If so the orthography must have originated at a time when glyph for Η denoted a spirant (compare German Ochs, etc.)"!!!). Of course, all this does not preclude the possibility that square may indeed stand for [h] (or any other sound for that matter), but basing a theory on such a highly speculative assumption would be tantamount to drawing conclusions about ancient Latin from the Praenestine fibula.cf. LLOY87, p. 80, note 26, "Of course, any of the evidence that Pulgram takes from the fake 'Praenestine Fibula' must be discarded".
  17. Mapping at Will
  18. Transcriptions from one language to another is a tricky matter: if one takes them literally, i.e., letter for letter, one may run into insurmountable difficulties, e.g. (STUR20, p. 170 and p. 98): catapulta↔καταπέλτης (u↔έ?), Bēþlahaím[GOT|Gothic]↔Bethlehem (a↔e?), Catamitus↔Γανυμήδης (!! c↔γ, t↔ν/δ, a↔υ, i↔η??). Thus, one has to selectively apply the principle of "mapped letters" and the question is how? There do not seem to be any guidelines and everyone maps at will. In general, interpreting transcriptions entails the same problems as those of the introductory C↔Κ example, namely uncertainties about the pronunciation of the "source" language (how do we know that, e.g., Latin X was pronounced [ks]?) and about the desired degree of accuracy (couldn't the Spaniards come up with a better transcription than maikelyason for Michael Jackson? Why s for cks, when x=[ks] is available in Spanish?).An additional factor, when one of the languages is not Latin or Greek, is the accuracy of the "latinised" spelling, the origin of which is most of the times vague and doubtful (e.g., where do the vowels come from in Semitic words?). Having all that in mind, one is at a loss when having to decide the correspondence between ξ in Τάξιλα and any (combination) of the letters kṣaç in Takṣaçilā (note also that, due to the rules of Greek accentuation, the final α in the Greek word cannot be long, as is the final ā in the Indian one); any conclusion that ξ=[ʃ] or anything similar would be unjustified. Similarly, the correspondence Ḫšayāršā↔Ξέρξης cannot be easily translated into a correspondence of individual sounds and any ruling about the pronunciation of ξ based on this isolated transcription would be too precarious.Note also that the assertion about lack of "e-sounds" in Iranian cannot be true, since there is at least the example of par(i)dēza-|garden (whence, paradise) (ALLE87, p. 70); hence, the έ of Ξέρξης must also be explained.

    The Model
    From the transliterations A(h)uramazda↔Ωρομάζης, Artavazdâ↔Ἀρτάβαζος, the Catholics conclude (STUR20, p. 189; BLAS90, p. 118; ALLE87, p. 58) that Ζ↔[zd], but not that Ω↔[a(h)u] (first example) or that Β↔[v] (second example). Assuming that the "latinised" names are faithful representations not only of the original Old-Persian names (which I have not been able to confirm), but most importantly of the pronounced or written names that the ancient Greek writers used as basis for the transliterated "hellenised" forms (which can in no way be confirmed), the selective mapping of Ζ, but not of Ω and Β, should have raised a few eyebrows, but it went through uncontested.
  19. Lost in Translation
  20. In interpreting the phonological descriptions provided by ancient grammarians and writers, no statement is more pertinent that Sturtevant's: "A very few of the ancient descriptions of sounds are quite clear and satisfactory" (STUR20, p. 4). The description is often so involved and the terminology so unusual that the ambiguous text can be interpreted any way one sees convenient.

    In the passages cited above, it is not clear what the ancients refer to with the term "semivocales" or "ἡμίφωνα"; however, I chose to follow the (arbitrary) interpretation of the "experts" (WIANPR) that "άφωνα ονόμαζε τα στιγμιαία ή κλειστά σύμφωνα, ενώ ημίφωνα καλούσε τα διαρκή ή εξακολουθητικά|he [i.e., Thrax] named the stops άφωνα [i.e., 'mutes'] and the continuants ημίφωνα [i.e., 'semivowels']". This distinction is used by the Catholics to establish that Φ≠[f], Θ≠[θ], Χ≠[x], since these letters are not classified as "continuants" (i.e., ἡμίφωνα); but, if the (alleged) stop component in Φ=[pʰ], Θ=[tʰ], Χ=[kʰ] disqualifies them from being considered ἡμίφωνα, there can be no stop component in Ξ and Ψ, which are explicitly described as ἡμίφωνα. Hence, if one wants to be consistent, one has to either accept the above-suggested value for Ξ or discard the arbitrary interpretation of the terms ἄφωνα and ἡμίφωνα and concede ignorance.Unfortunately, it seems that the zealots of the Catholic or "scientific" pronunciation would most likely opt for the adoption of a new value for Ξ (the pronunciation of which occupies, in the treatises on ancient-Greek phonology, an insignificant fraction of the total "real estate") rather than distrust an established but erroneous interpretation.

    In the interpretation of the text of Syrianus, there is a possible mistranslation ("Ψ being generated when Σ is collocated with Π"), which however is due to the absence of καὶ between π and σ (in the text as quoted by Sturtevant), thus allowing for a different interpretation. This example is typical of the ambivalence of many of the ancient texts, which is not necessarily due to the original author, but sometimes due to later copying and editing (in this particular case, it may simply be that Sturtevant's quote is flawed). As for the expression "σύγκειται ἐκ" found in many of the ancient descriptions, it does literally mean "is composed of", but the texts do not specify the "composition" with any more detail: a train may be composed of a locomotive and a car (implying sequence), while the body is composed of intermingled bones and nerves (implying mixture).

    One point in the quoted texts of the Grammarians that is rather unambiguous, is Dionysius' strange choice of words for describing how the two constituent sounds of the so-called "double letters" are combined with each other, namely "συνεφθαρμένων ἀλλήλοις ἰδίαν φωνὴν λαμβάνοντα". Liddell-Scott translates the corresponding verb συμφθείρω (as used by Plutarch in connection with colours) as "melt or die away into each other" and in general as "destroy/perish together"; it is, thus, clear that there is no question of juxtaposition of sounds in this description, but rather of coalescence (particularly taking into account that a new sound, or "ἰδία φωνὴ", is produced). In the treatises on ancient-Greek phonology, the meaning of this statement is blatantly ignored and conclusions are drawn only from the fragment "διὰ τοῦ κ/π και σ" (which is interpreted as indicating addition κ/π+σ!). The interpretation of the quote by Dionysius of Halicarnassus is probably the only sound statement in my argumentation on Ξ and Ψ and, if one is to take the descriptions of the Grammarians seriously, then one must necessarily discard the conventional assignments Ξ=[ks], Ψ=[ps], Ζ=[zd].

    The Model
    An example of mistranslation (albeit of minor importance) is STUR20, p. 163, where Lucian is thus quoted and translated "ὅτι δὲ ἀνεξίκακόν εἰμι γράμμα, μαρτυρεῖτέ μοι καὶ αὐτοὶ μηδέποτε ἐγκαλέσαντι τῷ ζῆτα σμάραγδον αποσπάσαντι καὶ πᾶσαν ἀφελομένῳ Σμύρναν|That I am a longsuffering letter, you yourselves are my witnesses, since I have never brought suit against Zeta for taking an emerald from me and robbing me of all Smyrna [Stu's translation]". Here, I am quite sure that Sturtevant's interpretation of the last word is not correct. He obviously understands it as referring to the city of Smyrna (modern Izmir), as suggested by the use of the capitalised form (Σμύρναν) in his quoted Greek text and its translation "as is" (Smyrna) in the English text; apparently, he is led to this conclusion by the city's English name (Smyrna) and/or by a possible capitalisation (Σμύρναν) in the "original" (i.e., its modern edition by some Westerner, since lowercase letters were not known to Lucian or any of the ancients). However, the city's name, particularly in Lucian's time, was Σμύρνη and Lucian (who spoke and wrote in the Koine) would have no reason to use the old Aeolian form (Σμύρνα); furthermore, "robbing" someone of (actually taking away/removing from someone) "all" of a city makes little sense. On the other hand, if one recalls that, according to the Gospel (which is contemporary with, if not predating, Lucian), the Magi brought to newborn Jesus "δῶρα χρυσὸν καὶ λίβανον καὶ σμύρναν|gifts gold and frankincense and myrrh" (Matt. 2:11), where "σμύρνα" is (certainly) not the city but an oitment, Lucian's passage makes perfect sense (particularly, considering that "σμύρναν|myrrh [acc.]" is juxtaposed with "σμάραγδον|emerald [acc.]").
    An example of a statement that completely misinterprets (or rather misuses) the quoted text is WIANPR: "ο Διονύσιος ο Αλικαρνασσεύς αναλύει την προφορά του φθόγγου κ ως εξής: λέγεται μὲν τῆς γλώττης ἀνισταμένης πρὸς τὸν οὐρανὸν ἐγγὺς τοῦ φάρυγγος καὶ τῆς ἀρτηρίας ὑπηχούσης τῷ πνεύματι (Περὶ συνθέσεως ὀνομάτων 14,140). Από την περιγραφή προκύπτει ότι ο φθόγγος κ ήταν άηχος υπερωικός (όπως στις λ. καρπός, κόρη, κωφός) και δεν είχε ακόμη εμφανιστεί το ουρανικό αλλόφωνο [c] (όπως στα νεοελληνικά καιρός, κερί, κύκνος: [cerós], [cerí], [cíknos]).|Dionysius of Halicarnassus analyses the pronunciation of the sound κ as follows: it is spoken with the tongue raising to the palate near the throat and the windpipe echoing to the breath [Sturtevant's translation] (Comp. Verb. 14,140). From the description, it follows that the sound κ was voiceless [or mute?] velar (as in the words καρπός, κόρη, κωφός) and the palatal allophone [c] (as in modern-Greek καιρός, κερί, κύκνος: [cerós], [cerí], [cíknos]) had not yet appeared." Ex-squeeze me? Baking powder? ΑΗΧΟΣ ΥΠΕΡΩΙΚΟΣ|VOICELESS [OR MUTE?] VELAR??? How can one conclude all that from Dionysius' description? It is not clear whether "άηχος" (which literally translates as "soundless") is meant as "voiceless" or "mute", but neither follows from the description.The feature of consonantal voice is absent not only from the quoted passage, but from the entire corpus of ancient (Greek and Latin) grammarians (cf. STUR20, p. 91, "The Greeks and Romans have left us no account of the distinction between voiced and voiceless mutes"; ALLE87, p. 30, "European phonetics was slow to discover the nature of 'voice', i.e. glottal vibration, as a distinctive feature of consonants [...] it remained completely... unnoticed th[r]ough the middle ages, and only began to be recognized in the nineteenth century"). The term "mute" is a bad (Latin) translation of the term "ἄφωνα" (cf. the translation on STUR20, p. 172) used, e.g., by Dionysius Thrax in BEKK16, p. 631: "ἄφωνα δέ ἐστιν ἐννέα, β, γ, δ, θ, κ, π, τ, φ, χ· ἄφωνα δὲ λέγεται, ὅτι μᾶλλον τῶν ἄλλων ἐστὶ κακόφωνα, ὥςπερ ἄφωνον λέγομεν τραγῳδὸν τὸν κακόφωνον|there are nine voiceless [consonants], β, γ, δ, θ, κ, π, τ, φ, χ; they are called voiceless, because they are more cacophonous than the others, in the same way that we call voiceless the cacophonous bard"; it is evident that there is no relation between the text of Thrax describing the (so-called) "mutes" and that of the Halicarnassean cited above, much less between the terms ἄφωνα and ἄηχα. Furthermore, the term "υπερωικός|velar" does not feature in the cited passage; on the contrary, the quotation includes the terms "οὐρανὸν|[here:] palate [acc.]" and "ἐγγὺς τοῦ φάρυγγος|near the pharynx [throat in the translation of Sturtevant]" and it would be at least equally (if not more) correct to conclude a palatal or pharyngeal articulation from this text. As a matter of fact, Sturtevant's comment (STUR20, p. 186) to the entire paragraph (which describes the articulation "Τῶν δὲ καλουμένων ἀφώνων ἐννέα ὄντων|Of the so-called 'mutes,' which are nine in number [Stu's translation]", the quoted excerpt relating to the pronunciation of not only κ, but also γ and χ) is that "this passage does not help us to decide whether δ, θ, and τ were alveolars or dentals, or whether γ, κ, and χ were palatals or velars"! The only valid point is that Dionysius does not speak about any allophones of κ, γ, χ (which exist in modern Greek, but also in most languages); however, no ancient (Greek) grammarian describes any kind of allophony, even though we know that it existed at least in the case of the so-called "velar γ"(=[ŋ]) and "voiced σ"(=[z]) and, if one concludes from the cited passage that κ, γ, χ had a uniform value in all environments (i.e., if one denies the existence of any allophones), one would have a very hard time explaining, e.g., how this value of γ (i.e., [g] as believed by the Catholics) emerged in some cases from ν before velars (e.g., σύγγραμμα, ἔγχυσις, πάγκαλος).

It must, by now, be clear that, in this informal deception pageant, the qualifications of the professionals are second to none. All I had to do was follow in their footsteps and voilà: the traditional (or "received") value of yet two more Greek letters demolished!It should be pointed out that I do not consider that, with the refutation of my own argumentation on Ξ=[ʃ], the traditional value of Ξ=[ks] has been proven beyond any doubt. I favour the traditional value, yet I believe that we have more reasons to believe in a deviating pronunciation of Ξ and Ψ than to many a "scientifically reconstructed" sound.

Let us now take a closer look at the remarkable formula that permits one to distort the evidence in support of (virtually) any conclusion one seeks.

The Recipe

In fending off yet another attempt of the "Hellenocentrists"I believe their "credo" is summarised in one of the representative magazines, ΕΛΛΗΝΩΝ ΙΣΤΟΡΙΑ|History of (the) Greeks, no. 25, p. 56: "Δὲν χρησιμοποιοῦμε 'φοινικικὸ' ἀλφάβητον. Το ΑΛΦΑΒΗΤΟΝ ὑπήρξε 'Ἑλλήνων ἐξεύρημα'. Δὲν ὁμιλοῦμε 'σανσκριτικά'. Δὲν εἴμεθα 'ἰνδοευρωπαϊκῆς' καταγωγῆς...|We do not use 'phoenician' alphabet. The ALPHABET was 'invention of Greeks'. We do not speak 'sanskrit'. We are not of 'indoeuropean' descent." to prove that Greek was the source of all language, Dr. Moshe sets forth the three main steps towards linguistic fraud:

  1. Preconceived Theory
  2. "Η αναζήτηση τεκμηρίων, όταν στο βάθρο έχει ήδη τοποθετηθεί ένα μύθευμα, καταδικάζει εξ αρχής την προσπάθεια, ακόμη και αν πηγάζει από γνήσια αγάπη για τη γλώσσα.|The search for evidence, when a myth has already been placed on the pedestal, condemns the endeavour right from the start, even if it emanates from genuine love for (the?) language": The first step is to create a myth or "preconceived theory" that will prove itself. In the case investigated by Dr. Moshe, the myth is that Greeks taught everyone how to speak. In my "reconstruction of Ξ" above, my wish was for Greek to possess more sibilants. In the case of many Orthodox partisans, the preconceived theory is that ancient Greeks spoke with the same sounds as modern Greeks, while the Catholics strived to establish a phonology closer to the one(s) they were familiar with.It must have already been made clear from the timeline of the question that the Catholic pronunciation was first (in the 16th century) declared the genuine one based on personal preference and occasionally on circumstantial and superficial evidence and then (mostly in the 19th and 20th centuries) "scientifically" justified or (in a few cases) "refined". However, this is how science works in general: a prima facie interpretation of (some of) the available data is constructed; most of the times, this theory is nothing more than a hunch; being such, it is worthless without verification, but often this last step is forgotten in "reconstructive science" and the unverified speculation is presented as fact ("Shoot first, ask questions later").

    An infamous example of speculation that made its way into textbooks and has been established as a fact in the mind of most of us is Kretschmer's simplistic theory on the coming of Greeks to Greece. Horrocks (HORR10, p. 18) presents it as "The traditional solution to the problem of the distribution of the Greek dialects was provided by means of a theory of three successive 'waves' of invaders (Kretschmer (1896, 1909)), according to which Greek was supposed to have developed as a separate branch of the Indo-European family somewhere outside the Balkan peninsula and to have split into dialects prior to the settlement of the Greek mainland" and observes that "This approach, however, has now been shown to entail quite serious archaeological and linguistic difficulties".I remember struggling to properly deal with a friend's question "When was it that the Dorians came to Greece?"; I did not know whether to give the traditionally "accepted" date (XIII BC) or try to explain that the whole theory was a hoax. One may argue that it was merely an unfortunate but honest attempt to explain the distribution of Greek dialects, but if Jannaris' assessment of Kretschmer's book is correctJANN97, p. v, n. 1:"PKretchmer’s recent volume Einleitung in die Geschichte der Griechischen Sprache (Göttingen, 1896) is not what the title professes to be. It is virtually an attempt to fix the original seat of the Aryan (Indo-Germanic) race in Europe and particularly in Germany (p. 60), and then an ethnological study of the various non-hellenic races (βάρβαροι) which in prehistoric times occupied countries north of Greece and Asia Minor. As a matter of fact, there is not a single paragraph in the book about the Greek language in its historical period". In the part of his book cited by Jannaris, Kretschmer indeed identifies the limits of the "Indo-Germanic homeland" as North and East Germany, the Alps, the Balkans and the Steppe., the theory may be due to a myth (Indo-Germanic origin in/near Germany) that needed to be proved.

  3. Evidence Selection
  4. "Ο αυτόκλητος ερευνητής θέτει συγκεκριμένο στόχο (π.χ. να παρουσιάσει την Πρωτοελληνική σαν μητέρα-γλώσσα) και κατόπιν επιλέγει τα δεδομένα που ταιριάζουν στην εικασία του. Συρράπτει ετερόκλιτα στοιχεία με βάση εκκινήσεως την εξωτερική ομοιότητα και πασχίζει να συμμορφώσει το γλωσσικό υλικό που είναι διαθέσιμο|The self-proclaimed researcher sets a particular target (e.g. to present Proto-Greek as a mother-language) and then selects the data that fit his conjecture. He puts together incompatible elements based on outward similarity and strives to appropriately configure the available linguistic material": Once focused on the target (see first step), select the appropriate evidence or bring together incompatible evidence. This is what the subjects of Dr. Moshe's study did, based on a few similarities between the vocabularies of Greek and other languages; this is what I did with, e.g., the inscriptional evidence when I ignored the numerous examples that have ΚΣ for Ξ. This is also what the Orthodox zealots are doing when, e.g., assuming a [v] value for Ϝ (digamma). This is also what the Catholics are doing when, e.g., choosing to ignore inscriptional evidence.Even Allen's more "impartial" review of the evidence is not free from selective treatment, where the evidence is (not ignored, but) downplayed (e.g., when discussing the "aspirate [h]" in ALLE87, p. 52: "There are admittedly quite frequent omissions; but some were due to the fact that ... [what about the others?]; and the more significant fact is that false writing of H is rare [why should correct spelling weigh the same as incorrect, when orthography may not be keeping up with pronunciation?]", emphasis and comments mine).

    But all the above are misdemeanours compared to the skillfulness of the sages of reconstruction. Enter Bedřich Hrozný, the celebrated "decipherer" of the "Hittite language". After being praised by his peers (who were, obviously, thrilled to see a new research field opening up) for having discovered the oldest written Indo-European language, he was seeing Hittites everywhere, particularly in the Indus valley and in Crete and the Pelopponese". Chadwick (CHAD90, p. 28) exemplifies his fallacy by juxtaposing his (Hrozný's) and their (Chadwick's and Ventris') translation of tablet PY Eb 317"Here is his version of a Pylos text (given in English translation of the French of his publication):
        Place of administration Ḫataḫuâ: the palace has consumed all (?).
        Place of administration Saḫur(i)ṭa (is) a bad (?) field (?): this (delivers in) tribute 22(?) (measures), 6 T-measures of saffron capsules (p. 304).
    We now translate this text as follows:
        Thus the priestess and the key-bearers and the Followers and Westreus (hold) leases: so much wheat 21·6 units.
    "
    and concludes that "The arbitrariness of Hrozný's work is so patent that no one has taken him seriously"This statement is not 100% correct. For example, a review of Hrozný's book on that topic begins by stating: "On ne peut qu'admirer l'ampleur des vues de M. B. Hrozný et la maîtrise avec laquelle il embrasse du regard le monde antique depuis l'Inde jusqu'à la Crète en passant par l'Asie Mineure. Plusieurs conjectures restent incertaines, mais peuvent être confirmées par de nouvelles découvertes.|One can only admire the breadth of views of Mr. B. Hrozny and the skill with which he beholds the ancient world from India to Crete, through Asia Minor. Several conjectures are still uncertain but may be confirmed by new discoveries." It is evident that, not only did some scholars take him seriously, but they were convinced that confirmation of his theories was only a matter of time. It is only after Ventris presented his far more rigorous work that the world looked at Hrozný's highly conjectural work with skepticism and (perhaps) contempt.. Here, Chadwick refers only to Hrozný's later (post-Hittite) works, implying that they do not measure up to his "decipherment" of Hittite; however, it is hard to see how the quality of his later works differed from that of his renown onescf. his translation of the "Hittite code of laws", which allegedly "gave a vivid insight into the life of a society that had not been seen or heard of for more than 3 millennia", e.g., §17: "Si d'une FEMME LIBRE son fruit quelqu'un fait sortir, [s]i (c'est) LE 10. MOIS, 10 SICLES D'ARGENT il donne; si (c'est) LE 5.(? 6.?) MOIS, 5 SICLES D'AR[GENT] il donne; ses obligations il remplit|If from a FREE WOMAN her fruit someone brings out [extracts?], if (it is) THE 10th MONTH, 10 SHEKELS OF MONEY [SILVER?] he gives; if (it is) the 5th(? 6th?) MONTH, 5 SHEKELS OF MONEY [SILVER?] he gives; his obligations he fulfils". and why his newer method was not as rigorous as his earlier one.Even though I have not delved into the specifics of his decipherment, the popular depiction of his main clue is rather telling:
    "he began working with a particular set of rhymed lines that read
        NU NINDA-AN EZZATENI WATAR-MA EKUTENI
    At this point Hrozny was inspired to follow a new train of thought. Recognising the Babylonian sign for bread, 'ninda', he considered the probability of the next word, 'ezza', to mean 'eat' and thus its potential as a cognate of the Greek 'edein', Latin 'edere' and German 'essen'. Then seen like this, the other words leapt out – 'nu': now, 'watar': water – leaving Hrozný with his first successfully deciphered sentence: 'Now you will eat bread and drink water'.
    "
    This method ('ezza'↔essen, 'nu'↔now, 'watar'↔water!!!) is as rigorous, as would be to conclude that the Praenestine Fibula ("MANIOS MED FHE FHAKED NVMASIOI") is a fake based on the identification of "FHAKED" as an English participle!
    One would be rather inclined to agree with Siamakis (SIAM88, §1,606) that "μ’αὐτὸ τὸ σενάριο 'ἐπιστημονικῆς' φαντασίας, ποὺ δὲν ἔχει καμμία σχἐσι μὲ τὴν πραγματικότητα, ἀπέδειξε ὁ Χρόζνυ, ὅτι οὔτε τὰ 'χετταϊκὰ' τὴς Μ. Ἀσίας διἀβασε πράγματι, ἀλλ’ὅτι τέτοιες φαντασιώσεις καὶ φακιρικὲς ἀρλοῦμπες εἶναι καὶ οἱ 'ἀναγνώσεις' του ἐκεῖνες|with this 'science'-fiction scenario [i.e., his 'reading' of Linear B], which has nothing to do with reality, Hrozný proved that he did not really read the 'Hittite' of Asia Minor either, but that those 'readings' were similar delusions and phony nonsense" and that (§1,607) "ἄν ζοῦσε ἀκόμη λίγο, δὲν θἄφηνε στὸν κόσμο γραφὴ ἀδιάβαστη· θὰ διἀβαζε ἀκόμη καὶ τὶς πτυχὲς τῶν κιόνων τοῦ θησείου|had he lived a bit more, he would have left no unread scripture in the world; he would even read the grooves of the columns of Theseion"!

    And this is not an isolated case. On the question of the "Minoan language" alone, an issue that had not been satisfactorily answered for a mere 50 years, Chadwick lists five further prominent "scientists", who make charlatans look like Einsteins.Gordon "read" the accounting tablets as elegiac poems; Stawell "read" the Phaestos disk and some Linear A tablets by replacing the symbols with their Greek names, then keeping the first syllable of each and expanding (!!) the syllables into full Greek words; Ktistopoulos "translated" the Disk into Semitic, in which he was "inexpert" (!), resulting in a religious paean than involved "of the eggs the white"; Persson "read" the name of Poseidon in a decorative pattern (!) on the rim of a jar; Georgiev "read" the "Minoan writings" as "an ingenious mélange of linguistic elements, which resembled Greek when it suited his purpose and any other language when it did not". (CHAD90, pp. 27-31) It is, therefore, evident that the poor chaps mercilessly pummeled by Dr. Moshe are an amateur drop in an ocean of delusional professionals; while their attempt to establish Greek culture as superior is patently nationalistic, the "work" of the scholarly establishment is not innocent either and is sometimes driven by overweening ambition (e.g., Hrozný) and sometimes by concealed chauvinism (e.g., Kretschmer and Georgiev). In other words, what Dr. Moshe scornfully calls the "self-proclaimed researcher" is merely a leisured plagiarist of venerable illusionists, who make a living out of it.

  5. Scientific Claim
  6. "Αξιόλογοι συνάδελφοι γλωσσολόγοι εκφράζουν ανησυχία για τη διάδοση των παραγλωσσικών και πρωτογλωσσικών μύθων. Αναρωτιούνται δικαιολογημένα πώς είναι εφικτό να ανασκευαστεί κάθε πλάνη, πώς είναι δυνατόν να συμμαζευτεί κάθε ψεύδος που έχει κυκλοφορηθεί με τον μανδύα τής πρωτότυπης έρευνας|Respectable fellow linguists express their worries about the spread of para-lingual and proto-language myths. They rightfully wonder how it is feasible to overturn every deceit, how it is possible to do away with every lie that has been circulated in the guise of original research": In view of the numerous examples of professionals that have made a name for themselves by producing myths, it is evident that the undermining of linguistics is an inside job. One can only wonder why the "respectable fellow linguists" strain out the amateur gnat and swallow the professional camel on account of their having a monopoly on "science". Which one of the venerable "scientists" has strived to make linguistics a rigorous science? It is evident that Reconstructional Linguistics is in dire need of a good deal of housecleaning, but little effort has been taken up so far (if at all).cf., e.g., LLOY87, pp. 8-9: "Since linguistics is (or aspires to be) an empirical science (as far as it is possible), one might think that historical linguistics would have dedicated much time and effort to observing sound change in progress in order to gain some idea of how changes have occurred in the past. In reality, however, very little investigation of current sound change has been undertaken. The majority of historical linguists have been content to rely on elaborate 'thought experiments' in determining how sound change actually happens, rather than on direct investigation of real speech communities (Labov 1970, 202). Much more effort seems to have been expended in explaining away any inconvenient data that fail to fit into preconceived theories than to elaborating theories that accurately fit the data." The author of this philippic against the "self-proclaimed researchers" is not free from sin either; we have already seen how the Greek-Wikipedia article on Classical Greek pronunciation (WIANPR), which is to a large extend his brainchild, misrepresents the testimony of an ancient Grammarian, in order to serve the preconceived theory of lack of palatals in ancient Greek (more examples will be examined later).

    To draw a couple of parallels, if more and more people start comparing their children's Kindergarten drawings with famous modern art and argue that their offsprings are artistic geniuses, it may make more sense to reconsider the appreciation of modern "art" (or at least better explain what is so artistic about it), rather than address every such claim; as long as conventional medicine uses dubious methods for treatment (e.g., "let's try laser treatment to see if your condition improves"), alternative medicine (e.g., "many people find acupuncture helps, why don't you try"), will flourish and be accepted. Similarly, as long as linguistics shelters the Hroznýs and the Kretschmers, as long as it presents its conclusions in a narrative, occasionally citing some circumstantial evidence as support, the same method will be used by "self-proclaimed linguists" in support of obscure ends. If you do not clear your field of the weed, you do not get to decide which myth is acceptable and which is not. Linguistics has a lot to learn from other fields of science, particularly mathematics.

One may wonder how all that (myths from comparative linguistics and ancient-script decipherment) relates to the issue at hand, namely the correctness of the Catholic or Orthodox pronunciation. As we will see below, many myths are also created in the course of trying to establish extravagant features for the classical languages. The above "recipe" is a recurrent pattern: first a desired or suspected value is assumed for a letter or combination of letters; evidence that may support the assumption are selected, fabricated or distorted; once as little as one piece of favourable evidence is found, the assumed value is declared the only true and "scientifically" proved value.

Anatomy of a Fallacy

Since all theories on ancient Greek will be separately addressed in subsequent chapters of this site, let us have a closer look at an extravagant theory on Latin.

In his "Vox Latina" (ALLE78, pp. 12-13), Allen observes that "there are some Greek words containing initial voiceless unaspirated consonants which are borrowed into Latin with voiced consonants; thus κυβερνῶ becomes guberno, πύξος becomes buxus, κόμμι becomes gummi, κράβ(β)ατος becomes grab(b)atus, and so on" and concludes that this "piece of evidence, though indirect, is rather suggestive in [a] connexion" between Latin and English, namely that "the Roman listening to Greek was in much the same situation as the Englishman listening to Hindi, i.e. that the voiceless plosives of his own language, at least in initial position, tended to be aspirated". In other words, based on a few examples where Greek initial Κ and Π correspond to Latin initial G and B respectively and on the assumption that ancient Greek "distinguishes [phonemically] between aspirated and unaspirated consonants" (note that his "Vox Latina" predates his "Vox Graeca", so this assumption is, at least for the reader of the former book, an axiom), he considers it all but certain that word-initial voiceless plosives were aspirated in Latin much like in English.

How can this argumentation, which is based on isolated Latin-Greek word pairs, be more serious than the conclusion (of the "self-proclaimed researcher" mentioned above) that Greek spawned (or is at least related to) Finnish based on isolated Finnish-Greek "cognates" (vesi↔Fύδωρ, aina↔αἰωνίως, tuli↔σέλας, sulo↔θάλλειν, riekale↔ῥάκος + -ύλλον, puu↔φύειν)? Of course, no one can 100% exclude the possibility that Romans aspirated their initial plosives and that (at least some) Finnish/Uralic words were borrowed from Greek/PIE, but that remote possibility is not made more certain by the less-than-circumstantial evidence presented. So, let's take a closer look at Allen's evidence.

The first problem with Allen's "evidence" is that the Latin-Greek word pairs that Allen provides are anything but numerous. To conclude, based on the correspondence Κ-/Π-↔G-/B- in a handful of words, that word-initial voiceless plosives in Latin were most likely aspirated is tantamount to claiming that Greek and Turkish are related, based on the "cognates" beş|five[TUR]πέντε|five, dört|four[TUR]τέτταρες/τέταρτος|four/fourth. In order to overcome this drawback, he plays the card of the "colloquial character" of the borrowed words, namely "non-classical borrowings, in which the actual speech is likely to be reflected rather than a literary consciousness of the Greek spelling". In other words, he alludes to a much larger number of similar "borrowed words", which are not attested because they were not "classical". It does not need to be pointed out that this is plain speculation, but he carries on down the same path citing words from "Vulgar Latin, as e.g. (reconstructed[!]) botteca from ἀποθήκη (cf. Italian bottega), or […] blasta from πλαστ-", as well as "Cicero's statement (Or. 160) that Ennius used always to say 'Burrus' for Pyrrhus".

A systematic study of the phenomenon, which would be the only way to verify or discard Allen's theory, is undermined by his conjecture that the vast majority of counterexamples (i.e., examples where the normal correspondence between the Greek and Latin word-initial voiceless plosives is maintained, e.g., κέντρον→centrum, κιθάρα→cithara, κόραξ→corax, κρύπτη→crypta, κατάλογος→catalogus, κνημίς→cnemis, κρόκος→crocus and so on) are products of a literary tradition and only a few traces of the actual pronunciation remain in the written records. However, in addition to the paucity of evidence, another significant problem is the nature of the isolated examples put forward by Allen, which do not stand up to scrutiny:

Furthermore, the conclusion (aspirated word-initial voiceless plosives) introduces more problems than it actually "solves": how does it fit in the general context of Romance languages? After all, as admitted by Allen himself (again on p. 12), "The Romance languages also generally agree in lacking aspiration (e.g. the pronunciation of Spanish tiempo, from Latin tempus)". Given that there are no traces of the alleged Latin tendency either before (proto-Latin or PIE) or after (late Latin or Romance) the classical times, it must have been a one-off phenomenon.

To sum it up, there is a correspondence Κ-/Π-↔G-/B- in a few Greek-Latin word pairs, which Allen explains using the following assumptions:

  1. Greek Φ, Θ, Χ were distinguished from Π, Τ, Κ in that they were "aspirated" (namely [pʰ], [tʰ], [kʰ]).
  2. The Latin words were direct loans from Greek.
  3. Latin initial voiceless plosives (P-, T-, C-) were aspirated (namely [pʰ-], [tʰ-], [kʰ-]), as in English.

Assumption #1 will be investigated more closely in the particular chapter on Θ, Φ, Χ, but we can accept it as a working hypothesis in this chapter for the earlier examples (we have already seen that it is invalid for the late-Latin/early-Romance examples). The elementary research presented above demonstrates that assumption #2 is the most problematic of all and that, in all but one cases, it cannot be safely established. Assumption #3, which is actually the conclusion that Allen draws from this weak correspondence or his explanation of the phenomenon, is incompatible with the history of Latin.

Despite the above objections, many may choose to adopt Allen's theory, because it is "explaining" the aforementioned phenomenon (Κ-/Π-↔G-/B-). This is often the case in linguistics (but also in other sciences), since, in the words of Jannaris (JANN97, p. vii), "many a theory, old as well as modern, enjoyed almost canonical deference not because of its intrinsic merits, but rather because of the absence of a better theory"; in other words, a particular theory, despite its patent shortcomings, may be adhered to and advertised as the only truth due to lack of... (serious) competition (Kretschmer's theory mentioned above is another example)!

Sturtevant's (STUR20, pp. 98-101) explanation of the above word pairs, but also of others where Γ→C, C→Γ, Δ→T, D→T, B→Π, P→Β both word-initially and word-internally (which are, however, overwhelmingly outnumbered by the "normal" transliterations P↔Π, C↔Κ, etc), is a full correspondence of P, B, T, D, C, G with Β, Π, Δ, Τ, Γ, Κ, respectively, in terms of "energy" using the conveniently vague terms "fortis" and "lenis", whatever these may mean. I do not find such a... metaphysical explanation very convincing.

My favourite explanation, when the data are so trifle and the theories so problematic, is:

"WE DO NOT KNOW!"

It could have been mere chance, influence of a third language, or unforeseen circumstances.

But this explanation would still not be very popular, since in the mind of most people it is not acceptable for science to ignore the answer to any question; I, therefore, have a better alternative to offer. Voicing of initial plosives can be found in a number of foreign loans in Greek:

OriginalColloquial"Official"cf.
pyjama[ENG]μπιτζάμα=[bid͡zama]πιζάμα
penalty[ENG]μπέναλτι=[benalti]πέναλτι
tomata|tomato[SPA/ITA]ντομάτα=[domata]τομάτα (?|not sure if it exists)κόμμι→gummi
pisello|pea[ITA]μπιζέλι=[bizeli]πύξος→buxus (?)
travaglio|trouble, suffering[ITA]ντράβαλα=[dravala]τράβαλακράβ(β)ατος→grab(b)atus
tellal|crier, bellman[TUR]ντελάλης=[delalis]τελάλης
teneke|tin[TUR]ντενεκές=[denekes]τενεκές
perde|curtain, screen[TUR]μπερντές=[berdes]
tevatür|rumor[TUR]νταβατούρι=[davaturi]ταβατούρι
plongeon|dive, plunge, header[FRA]μπλονζόν=[blonzon]πλαστ-→blast-
pardon|pardon, forgiveness[FRA]μπαρντόν=[bardon]παρντόν
prise|(electrical) plug[FRA]μπρίζα=[briza]πρίζα
pabesë|unfaithful, disloyal[ALB|Albanian]μπαμπέσης=[babesis]κυβερνῶ→guberno
paroma|headfast, mooring line[VEN|Venetian]μπαρούμα=[baruma]Πύῤῥος→Burrus

All these examples relate to foreign loans that have entered the Greek vocabulary through "unofficial" channels. For some of them there is an "official" parallel form that reflects the original pronunciation and spelling. Thus, the above table relates to exactly the same phenomenon in Greek as the Latin one investigated by Allen (in the last column, the closest example of Allen's is adduced - note that all primary examples put forward by Allen, excluding the lamentable †botteca/bottega, have a parallel in present-day Greek). However, assumption #3 is certainly not true for the receiving language (Greek), which knows no "aspiration" whatsoever. Furthermore, with the exception of English (first two examples), assumption #1 is not true for the lending languages (French, Italian, etc), where there isn't the faintest idea of aspiration.A similar voicing of initial T and C is also reported for Latin loans into Basque: TEMPORA→dembora, CORPUS→gorputz (LLOY87, p. 72). So, there must be another explanation for the phenomenon.

I believe that the key is the second consonant of the original words; in all of the above cases, the second consonant is voiced. It is, therefore, very likely that the cause is a regressive assimilation of voice induced by the subsequent consonant (cf. patata|potato[SPA]→πατάτα=[patata], i.e., no voicing). One may, of course, wonder why such assimilation is not more general in Greek, but (primarily) targets foreign loans. The only reasonable explanation is that foreign loans, being words which do not "bond" well with the rest of the native vocabulary or with which native speakers are not "familiar", are less resistant to assimilatory phenomena.I do not claim to have discovered any "phonetic law" or to have found the only explanation for word-initial voicing; after all, there are some counterexamples, where voicing is not accompanied by a following voiced consonant, e.g., tef|tambourine[TUR]→ντέφι=[defi], or is the result of internal evolution, e.g., κρημνός|cliff→γκρεμ(ν)ός=[grem(n)os]; there are also cases of word-initial devoicing, e.g., (διφθέρα→)defter|notebook[TUR]→τεφτέρι=[tefteri].

Now, if one looks at Allen's examples, they exhibit the same characteristics: the second consonant in almost all cases is voiced; in the case of buxus, there is the uncertainty about the exact value of x (see above) and whether intervocalic x was voiced as in English (cf. Alexander=[ˌælɨɡˈzændɚ] vs Alex= [ˈaːlɛks]), while †botteca/bottega is not to be taken seriously (for the reasons laid out above). If assumption #2 is correct (otherwise Allen's argumentation is moot), the above explanation (for word-initial voicing of foreign loans in Greek) is simpler, more reasonable and does not involve any "exotic" features never evidenced in Latin or Romance.

How could Allen have come up with so flawed a theory? Well, he followed The Recipe:

  1. Preconceived Theory
  2. It is evident that Allen was desperate to make a contribution to the question of the pronunciation of Latin. Since the topic had been extensively studied, all he could do (if he were not to be considered a heretic by putting all this academic work in question) was to collect the arguments and conclusions of others (which, to a large extend, he did) and to confine any contribution to matters of "marginal" importance. Word-initial aspiration is one of these "unimportant" details, which has the further advantage that it is in line with English practice in pronouncing Latin. Perhaps his "secret" aim is to be found in his statement (p. 13) "one should probably not insist too strongly on the complete avoidance of aspiration in Latin".cf. Blass' declaration (BLAS70, p. 39) "Unsere [d.h. die deutsche] Aussprache ist in allen andern Punkten des Vokalismus [außer ΕΙ, ΕΥ und den so-genannten "langen Diphthongen"] fest genug begründet als die wenigstens annähernd wahre und echte|Our [i.e., the Germans'] pronunciation constitutes the, at least approximately, true and genuine in all other matters of vocalism [i.e., articulation of vowels and diphthongs other than ΕΙ, EΥ and the so-called "long diphthongs"]", which is essentially an effort to "scientifically" justify the pronunciation of ancient Greek by the Germans, which had already been establish for anything but "scientific" reasons (basically, because it was the pronunciation applied to Latin transliterations of Greek according to German pronunciation rules).

  3. Evidence Selection
  4. In order to support this preconceived theory, Allen chose the examples that suited his purpose. As already mentioned, there is a larger corpus of Greek-Latin word pairs where voiceless plosives in one language correspond to voiced plosives in the other (e.g., STUR20, pp. 98-101)A correspondence between Greek Γ and Latin K is also suggested by Plutarch (Quaestiones Romanae, 277d), based on the association of macella|butcher stalls and μάγειρος|cook, but this correspondence is suspect and not very reliable., but also (in the case of Ennius) Φ-→B-. This contradictory evidence, together with the prevalent "normal" transliterations of Latin words into Greek and vice versa, could serve as basis for a number of "unconventional" theories, according to one's objectives. Allen chose to concentrate on the ones where Κ-/Π-→G-/B- (which suited his theory) and to ignore (actually, withhold) the others.

  5. Scientific Claim
  6. After having manipulated the data in support of his theory, Allen triumphantly declares (p. 13) that "the discussion will have served to show how light may sometimes be shed on ancient linguistic problems by the observation of modern parallels". The statement, as such, is correct; this is a general practice, particularly in linguistics, wherein the past is looked at through prisms of the present; however, different prisms provide different views of the past and the question is which prism would be the most appropriate. As we have seen, Allen chose a prism that projects a skewed view of the past, essentially, a hallucination. Allen did not scrutinise his data and his theories and did not investigate other alternatives, but jumped to the first conclusion that was visible through his Anglo-German prism.

On the back cover of his book, a quoted review states that "This book is a model of clarity and succinctness". Well, maybe it is, when compared to others, but objectively it comprises many examples (such as the theory discussed above) of how extravagant speculation has hijacked science, albeit with a superficial scientific façade.

The End of Religion

In this chapter, we have seen how the data can be manipulated in support of an outrageous claim following the example of the great masters of reconstruction; we have analysed the reasons why that path leads away from the truth, outlined the steps towards linguistic fraud and seen how this recipe was put to practice by one of the gurus of ancient phonology. After this discussion, I can feel the desperation of those that place their trust in sanctified figures:

"If Allen is not to be trusted, who can we trust?"

I do not aspire to be a new Joseph Smith and create my own dogma. I have strived to demonstrate that the key in scientific matters is not trust, but thorough examination of the data, scrutinisation of the arguments and application of the rules of logic. It is, therefore, time to leave the den of religion and enter the realm of science.

Notes

Go to previous page Go to main page Go to next page