User:Jouna Pyysalo/LT

Criticism
Both supporters and opponents have constantly produced criticism of the laryngeal theory with the aim to modify, improve or reject the theory, in part or as a whole. Thus the laryngeal theory is in a revision process in which large parts of the community of Indo-Europeanists have been participating through the results of etymology, onomastics and comparative mythology. The central fields of criticism defined by the history of the laryngeal theory are presented in the following subsections.

Ever since its beginning the laryngeal theory has been split in two, the orthodox and the revisionist theory (Pyysalo & Janhunen 2018a). The orthodox models, beginning with Hermann Møller, assumed only a single vowel PIE *e and were abandoned after leading to an excessive set of six laryngeals in Jaan Puhvel’s model that still was unable to explain even the basic ablaut PIE *e/o in the absence of PIE *o. The revisionist models assume at least two vowels, starting with Jerzy Kuryłowicz assuming *e and *o. After this was shown insufficient, Heiner Eichner and others assume *e o a. The simultaneous presence of both three laryngeals *h₁ h₂ h₃ and *e *o *(a), however, leads to a circle of ambiguity, because the Indo-European vocalisms can be explained with both *h₁ h₂ h₃ and *e a o. Since by the emergence of the revisionist models all possible permutations of vowels *e a o and *h₁ h₂ h₃ have been tried without success. Unable to offer an universally accepted solution to the IE vowel and laryngeal problem the laryngeal theory is under continuous revision to reach that goal.

Critique of de Saussure’s early ideas
The history of the laryngeal theory begins with Ferdinand de Saussure’s (1877: 386) suggestion that the Neogrammarian ablaut pattern Neogr. *ǝ : *ā contains a common denominator, the coefficient DS (de Saussure's) *A (Neogr. *ǝ) alternating with *A₂ (Neogr. *ā). De Saussure soon explicated this idea in his Mémoire sur le système primitif des voyelles dans les langues indo-européennes (1879:127) asserting that Neogr. *ǝ : *ā stood for earlier *A : *eA, in which *A first colours *e into *aA, which then yields Neogr. *ā via compensatory lengthening.

This suggestion was initially left almost unnoticed as de Saussure’s (1879) other conjectures became criticized as erroneous by his contemporaries and led to general dismissal of his work by except for Hermann Møller and Albert Cuny.

In addition de Saussure (1879: 127) proposed an analysis of the pattern Neogr. *ǝ : *ō for which he claimed a second schwa, Ô, patterning as Ô : eÔ. > oÔ > Neogr. *ō. This, as pointed out in his critical review by Hermann Møller (1880:494n2), was a mistake: All occurrences of Ô can be explained by means of Neogr. *o (in *oA), i.e. the postulate Ô, unlike *A (Neogr. *ǝ), was not implied by a correspondence set (in order to indicate the lack of founding in material, Ô is marked henceforth with ⁺ to indicate its internal origin in opposition to asterisk *, the symbol of a comparatively well-defined item).

Furthermore, the critics noted that de Saussure’s ⁺Ô was inconsistent with the material itself. As pointed out by first by Møller (1880:518), as against ⁺Ô (in Gr. δοτήρ) and ⁺eÔ (Gr. δωτήρ), *A is implied by Gr. δάνος (Neogr. *dǝ-). Another inconsistency was pointed out by Holger Pedersen (1938:180-1) noting that ⁺deÔ- was contradicted by the set Lat. dās: Lith. dovanà.

In addition to ⁺Ô, making de Saussure’s system unsound, Karl Brugmann (1879: 773-4) noted the absence of a counterpart of Neogr. *a, in turn leading to incompleteness of the system, and concluded that de Saussure’s proposal was ‘a purely aprioristic scheme (rein aprioristische Construction) that did not hold water.’

Critique of Hermann Møller’s Indo-Semitic hypothesis
Hermann Møller, a Semitic and Indo-European linguist, dedicated his academic life to establish a genetic relationship between these language families, which due to the processes introduced led Oswald Szemerényi (1996:124) to name him the "the true founder of the laryngeal theory”. Despite the criticism presented against de Saussure, in which Møller had participated, he accepted all of de Saussure’s conjectures as the foundation of his subsequent Indo-Semitic hypothesis. To achieve his goal he abandoned the comparative method for the benefit of a methodology transforming IE vowels into Proto-Indo-Semitic ‘laryngeals’ ⁺E *A ⁺O (⁺h₁ *h₂ ⁺h₃) by means of generalising the (Proto-)Semitic root structure hypothesis CC(C) into PIE (1879:492 and 1906:xiv) to justify the postulation of laryngeals. This led the Neogrammarians to observe that Møller’s theory was ‘not sound in scholarship’ (Koerner 1985:336) , and even after the emergence of Hittite Hans Krahe (1958:97) stated that the laryngeal theory, the heir of Møller’s theory, ‘cannot be held secured neither with regard to its substance nor methodology’.

The real problem encountered by Møller was the need to to circumvent the ambiguity *oA and ⁺eO in de Saussure’s system. In order to do so Møller (1906:XIV) declared that in PIE, like in Semitic, there were only roots in PIE *e corresponding to the Semitic roots in *a. This ‘monovocalism hypothesis’ denied the existence of PIE *o, whence *oA could be removed from the ambiguity *oA or ⁺eÔ. In reality, however, *o had been shown to exist by means of Brugmann’s law (Collinge 1985:13-21), and was to be confirmed soon by the law of the second palatalization (Collinge 1985:133-142). Møller’s elimination took place only on paper.

In addition Møller (1879:151n1) suggested that also the long vowel Neogr. *ē was to replaced with ⁺eE where ⁺E stood for the subsequent h₁. This proposition created a similar ambiguity as de Saussure’s ⁺Ô. As pointed out by Schmitt-Brandt (1967: 2), ‘in reality there was nothing in the ablaut patterning *ei̯ : *i and *ē : *ǝ that would necessitate inferring to *eǝ’, whence also ⁺E has only a quasi-reconstructed status. Quite generally as mentioned by Szemerényi (1996:122), the whole methodology of replacing long vowels with *e and ‘modifying elements’ (i.e. the ‘laryngeals’ ⁺h₁ *h₂ ⁺h₃) remains considerably speculative’.

Next Møller (1879:150) proposed a partial solution to the absence of *a in de Saussure’s reconstruction: Accordingly, Neogr. aC- reflected *AeC > AaC- > Lat. aC-, Skt. aC-. Although this is now almost generally accepted, it should be noted that Møller’s (1880: 492-4n2, 1906: vi) generalisation of this for Neogr. *eC- < ⁺EeC- and Neogr. *oC- < ⁺ÔeC- was not valid due to the erroneous postulation of ⁺Ô and ⁺E.

Since the Indo-European data available in the 19th century did not preserve any segmental laryngeals that would have justified their reconstruction as the output of the comparative method, Møller, using his assumption of genetic relationship, applying Proto-Semitic root structure consisting of two or three root consonants CC(C) to PIE as well. By means of this methodology Møller implemented the postulates *A ⁺O ⁺E to the earlier roots *aC *oC *eC *Cā- *Cē- *Cō- regardless of their actual absence except for in *aC and *Cā- as subsequently shown by the Anatolian data. Perhaps the strongest criticism against this has been presented by Oswald Szemerényi (1967:92-93), who denied the validity of the reconstruction of the laryngeal theory as a whole: “there is no intrinsic reason why we should attempt to reduce all IE ‘roots’ to a single tri-phonemic pattern of the CVC-type [...]. On the contrary, it is clear that such notions were due to a double influence from Semitic linguistics: (a) in Semitic all words begin with a consonant; (b) in Semitic the general root-shape is tri-radical. But, of course neither feature is binding for IE.”

The impact of Hittite on the orthodox theory (Møller and Puhvel)
Bedřich Hrozný’s proof (1917) of the Indo-European character of Hittite turned the attention of scholars to Hitt. ḫ, a segmental laryngeal which had no counterpart in the previously known IE languages and was thus absent from the Neogrammarian reconstruction. A decade later Jerzy Kuryłowicz (1927), Albert Cuny (1927) , and Edgar Sturtevant (1928) and independently confirmed that Hitt. ḫ stood in positions in which de Saussure and Møller had predicted *A.

Despite the correlation between *A and Hitt. ḫ, the success was only partial, because problems emerged in the interpretation of ⁺E and ⁺O. The Anatolian evidence – as pointed out by Szemerényi (1967: 91-2) – does not show any laryngeal in the respective correspondence sets, i.e. de Saussure and Møller were correct only in one third of their predictions, namely *A.

The failure in the initial interpretation of Hittite by Benveniste, Kuryłowicz and others was later explained Johann Tischler (1980: 495, 498), who criticized the authors for ‘not approaching the problem inductively from the linguistic material itself, but deductively based on the theories of de Saussure, [Møller], and Cuny, a point manifested in the subsequent tendency to interpret the laryngeals missing in Anatolian as ‘lost’ rather than never existing in the first place as the data itself, showing no traces of ⁺E and ⁺O, implied.

Of the early interpreters Émile Benveniste (1935) continued Møller’s orthodox (or monovocalist) laryngeal theory, but by early 1960s it had become obvious that three laryngeals ⁺h₁ *h₂ ⁺h₃ and a single vowel *e was not sufficient to explain the attested data: The IE vowel qualities and the presence/absence of Hitt. ḫ implied six distinct correspondence sets instead of three. Perhaps the first to realise this state of affairs was Jaan Puhvel (1960, 1965, 1984ff.), who doubled the number of laryngeals. In his system in two parallel sets of identically coloured laryngeals, *h₁ *h₂ *h₃ for items attested in Hittite’, and ⁺H₁ ⁺H₂ ⁺H₃ ‘lost’ in Hittite appeared. While this enabled him to represent the six required correspondence sets, Puhvel’s suggestion was without a chance from the beginning, since the six-laryngealism only contained *e and could not explain the ablaut *e/o, and was abandoned by scholars except for Fredrik Lindeman (1982).

The impact of Hittite to the revisionist theory (Kuryłowicz, Eichner and Melchert)
An alternative to Møller’s and Puhvel's orthodox models, the revisionist laryngeal theory, was initiated by Jerzy Kuryłowicz (1935). Kuryłowicz’s model, like its subsequent versions, is otherwise identical to Møller’s theory in containing laryngeals ⁺h₁ *h₂ ⁺h₃ and the Semitic root hypothesis (CC·C), but in addition to PIE *e it also postulates PIE *o, that is it abandons the monovocalism hypothesis of Møller and Puhvel. Oswald Szmerényi’s (1967: 91) criticism, noting the absence of counterparts of ⁺h₁ and ⁺h₃ in Anatolian, meant that also Kuryłowicz’s early version had become outdated and revisions were required.

In a series of articles Heiner Eichner (1973, 1978 , 1980 , 1988 ), suggested a new revisionist model. Eichner accepted Kuryłowicz’s *e and *o and

(a) assumed PIE *a, thus replacing the functions of Kuryłowicz’s fourth (‘a-colouring, lost’) ⁺h₄.

(b) explained the lack of colouring in Hitt. eḫ, Hitt. ḫe with PIE *ē (allegedly not affected by the colouring effect of *h₂)

(c) accepted the complete absence (or ‘loss’) of ⁺h₃ in Old Anatolian, Hitt. ḫ consequently always reflecting *h₂ as in Oswald Szemerényi’s monolaryngealism (1970: 131).

Doing so Eichner essentially provided an alternative to three new laryngeals assumed by Puhvel by positing three short (*e a o) and three long vowels (*ē *ā ō) in addition to ⁺h₁ *h₂ ⁺h₃.

A revisionist competitor to Eichner emerged when Sara Kimball (1987) and Craig H. Melchert (1987) independently proposed a model essentially identical to his except for ⁺h₃ being reflected as Hitt. ḫ in root-initial position. However, as already pointed out by Eichner (1978: 162n77) earlier ‘all examples of h₃- can be explained with *h₂-.’ Since Eichner’s treatment contains only comparatively secured *h₂, while Melchert resorts to ⁺h₃ lacking a comparative status, Eichner’s model is clearly superior to the latter. In addition, as all alleged examples ⁺h₃e- can be shown to contain either *h₂o or *o (i.e. no laryngeal at all), Melchert’s model is a detour (Pyysalo 2016: 200-204).

The main problem of the revisionist models is the same as that of de Saussure’s original: The simultaneous assumption of more than one laryngeal and one vowel results in an unsolvable ambiguity, and the more laryngeals and vowels are simultaneously assumed, the more widespread the ambiguities become (Pyysalo & Janhunen 2018a: 1-7, 10-11).

The situation is aggravated by the existence of competing revisionist models, i.e. Eichner’s, Melchert’s and Kortlandt’s (Pyysalo & Janhunen 2018b), which provide mutually conflicting reconstructions. This can be illustrated with Hitt. a- : Lat. o- in the root-initial position, for which the models allow the following alternatives:


 * Hitt. a- : Lat. o- > ⁺h₃e- or ⁺h₂o- or ⁺h₃o- or ⁺h₁o- or ⁺h₃a-.

Since there is no laryngeal in the data (Hitt. a- : Lat. o-), postulated in the laryngeal theory only on the basis of the root hypothesis CC·C, the data do not provide any means to decide which reconstruction or model is the correct one, if any (Pyysalo & Janhunen 2018a: 11). Ultimately this is a result of the simultaneous use of Møller’s theory (explaining the IE vocalisms with laryngeals ⁺h₁ *h₂ ⁺h₃) and the Neogrammarian theory (explaining the IE vocalisms with vowels *e *a *o). Since these explanations are contradictory, the revisionist models based on that are a logical fallacy (Pyysalo and Janhunen 2018a: 6).

As all mathematical permutations of the vowels *e a o and three laryngeals have been tried and this did not yield a universally accepted solution to the problem of IE vocalisms the solution remains an open research question (Pyysalo & Janhunen 2018a).

Critique of the laryngeal theory from the viewpoint of philosophy of science
The critiques of have repeatedly contrasted the deductions the laryngeal theory based on Hermann Møller's initial hypotheses to the comparative method of reconstruction, preceding inductively and empirically. In order to motivate coining the term ‘laryngealism’ Voyles and Barrack (2015: 100) motivate this by explaining that rather than considering the laryngeal theory a theory they consider it an ideology. Instead of an ideology it would be, however, preferable to use the established terminology in philosophy of science: The laryngeal theory represents the hypothetico-deductive model of reasoning while the comparative method relies upon the scientific method: The laryngeal dispute, manifested in the top-down deductions of the laryngeal theory versus the bottom-up induction of the comparative method.

In chapter six of their book Voyles and Barrack place the laryngeal theory with other falsified theories in various sciences such as the phlogiston theory in chemistry. . After no version of the laryngeal theory has been able to win general acceptance even after it has exhausted all possible permutations of vowels and laryngeals, a comparison with other failed theories can be drawn except for the healthy core, the correlation between *A (*h₂) and Hitt. ḫ.