In Russian (later)
Contents Türkic languages
Contents Türkic in English
Classification of Türkic languages
G. Ekholm Germananic Ethnology
C. Stevens Gmn.-Türkic traits
A. Toth German Lexicon
A. Toth Turkic and English
R. Mc Callister Non-IE in Gmc. languages
Türkic borrowings in English
Türkic in Romance
Alans in Pyrenees
Türkic in Greek
|Türkic in English|
The Indo-European Language Family: Questions About Its Status
Institute for the Study of Man, Washington D.C.
Journal for Indo-European Studies Monograph Series, Monograph 55, 2009, ISBN 978-0-941694-03-2, 978-0-941694-02-5
www.jies.org, Copyright © 2009 Institute for the Study of Man
Pagination per reprint in Journal of Eurasian Studies
Republished with permission
January-March 2015, Vol. VII, Issue 1, http://www.federatio.org/joes/EurasianStudies_0115.pdf
Mikes International, The Hague, Holland, 2013, ISSN 1877-4199
© Copyright Mikes International 2001-2015, All Rights Reserved
© Copyright Angela Marcantonio, 2009-2015
http://uniroma1.academia.edu/AngelaMarcantonio Links to publications
The science of modern linguistics is built on the pattern of the Indo-European (IE) linguistics, rather to the detriment of the science. The pattern of the IE linguistics carries its reasoning and problems to every other linguistic group, bearing on the relationship between Türkic in English and quite a few other “IE languages”. The problems of the IE linguistic theory are massaged from its inception 200 years ago by self-policing acolytes who nearly always found partial fixes, but a systematic synopsis of the problems and solutions has never existed; to see an impartial critical synopsis of the IE linguistics is an event of the epochal scale. Instead of periodical general review done by any other discipline, this field had discretely circulated dispersed criticisms, citations of diverse opinions, and discussions of some particular aspects, all with an implied premise that the theory is right, albeit requires tweaking. The offered citations on the theme of the sanity of the IE linguistics are extracted from the work of Angela Marcantonio, an editor of a linguistic compendium, who wrote an editorial introduction that laid out IE problems, inventoried positions, analyzed situation, and drew conclusions. The punch line is jesting, it rings “it is hard to pin down what the foundations of the theory are actually supposed to be”. In science, any theory rests on fundamentals. If the fundamentals are not fundamental, the term “theory” is a misnomer for speculation, propaganda, a set of beliefs, and the like craze promoting some cause. The past 200 years have proved that, every imbalance of the IE theory was mitigated by shifting foundation over under. These shiftings, hatched to fix specific problems particular to the IE model, as scientific approaches are useless for the IE linguistics, and even less applicable for the rest of the world's linguistics.
This posting cites the snippets taken from the Angela Marcantonio's Introduction, and as always, the devil is in the details, they need to be traced in the original publications, follow the links for the JIES and JOES publications. Few accents may be added.
Any editorial comment added to the text of the author and highlighted in (blue italics) or blue boxes is posting comment unrelated to the views of the author. The author explicitly does not personally believe that there may be a Turkic substratum to IE, or to Sanskrit, not only because there was no IE, but also because, even if IE had exited, the time and area of the two language groups do not add up, each appearing at the scene of history far apart from one another. The author does not subscribe to any links between the Turkic languages and IE, or Sanskrit.
The origin of the “IE theory” was apt to its time, a biblical model where Noah produced Semitic Shem, African Ham, and Eurasian Japheth, otherwise depicted as a Family Tree. The modern linguistic terminology still uses biblical tales, Semites are from Shem, and Hamites are from Ham. The IE linguistics formed as an obverse side of a racial coin, its dignified theoretical side, while the now eroded reverse side was a practical reality. The biblical model implies that at one remote point in time existed a formed language that split off into various dialects that grew into languages and language families. In most cases, that is a fallacy. The tiny bands of hunter-gatherers needed large tracts to forage, reaching into other's territories and initiating social exchanges. Creolization held the day, or rather many millennia. Then, the “Family Tree” had a shape of a grass lawn with stolons extending from every base and crisscrossing with stolons of their neighbors. After a drought, the surviving islands were restarting the process uncountable number of times. A most aggressive weed propagated and conquered. The emergence of the roving agriculture and animal husbandry tilted creolization toward the more fecund side, the Family Tree took a shape of brush spots in a grassy lawn. Leveling became a greater force in communications, all linguistic aspects were affected by leveling, typology, syntax, morphology, and phonetics. That trend grew with introduction of a 3-field system starting in 800s AD, which reduced agricultural roving, with companion linguistic fallout. The resultant leveling gave Family Tree a shape of groves in brushy tracts surrounded by grassy lawns. The printing presses of the Late Middle Ages made leveling ubiquitous at the expense of minority languages and dialects. At the same time, increased productivity of agriculture allowed rulers to engage in permanent warfare that stirred and dispersed subject population along the path travelled by cattlemen ranchers five millennia earlier. It took another 1000 years for the emergence of linguistics and its biblical-type etiology. Some linguists might have thought of the example of biological evolution. It took another 200 years for some linguists to realize that their model is hopelessly faulty, and come up with an alternate Wave model.
Every and all IE linguistic broodings see the ancient languages as some frozen codified entities, while our recent experience attests that languages are dynamic creatures with definite lifetimes and a spatial differentiation that makes versions separated by a mountain ridge or 200 km distance mutually unintelligible. So people speak 2, 3 languages, or whatever is needed. One of them is an areal pan-language that is used as a first and last resort, a lingua franca. Each language influences the others in all linguistic aspects, that is an areal Sprachbund. There comes a revolution, a new language becomes a pan-language, say, Greek, but the old pan-language keeps living within all languages of the Sprachbund. There comes a revolution, a, say, Latin language becomes a pan-language. Then the new pan-language and local languages evolve, becoming a tool of the literati, who start to codify them, setting rules of right and wrong, and deliberately pushing the wrongs way down. Then comes a linguist that discovers too many synonyms and homonyms, and a general linguistic chaos that needs to be systematized and classified. A linguistic model is born, full of contraventions, but suitable as an unifying force blind to its inner conflicts in all linguistic aspects. Throwing the effects of the layers of non-codified and mutually unintelligible causes into the IE studies makes the confused theory unmanageable. The model bursts from the inside.
One predominating theme of the Introduction is that linguistics is a science with no laws. The laws are created as needed, adjusted as needed, manipulated at will, and have premises as their conclusions. The only reliable element in the study of the linguistic laws is not taught, because it augurs that there are no laws. No law or a defined set of laws can take a student from Anglo-Sax. to English, in reverse, or in any direction. The only utility of the linguistic laws is the self-perpetuating education industry.
Unfortunately, the typology, a most basic property of language is totally unaddressed, as if it is not linguistics but a whimsical idea. In the ratings of importance within linguistics, typology trails endings in some God-forgotten IE language with a dozen of active speakers. That is not to say that the scholarship of 460 IE languages has 460 dictionaries, in fact it is far from that. And typology trails the last of them
One of the leading linguistic parameters is “genetic connection”, like “Noah begot Shem”, used as a premise (i.e. cognates) and as an objective. Genetic connection is a statistical parameter with 2 levels, that of a word, and that of a paradigm (class of all items). On the level of word, a critical threshold must be present, constituting a lexical paradigm; the other paradigms may consist of other shared innate linguistic properties. Aside from its inability to project into the future, what sets the linguistic genetic connection apart from the science is the absence of a demographic aspect. Unlike languages that are carried by warm bodies, the linguistic languages are carried by spirits entitled “demic diffusion”, “elite dominance”, and the like abstractions. The roles of maternal language imbibed with milk, of generations raised by mamas systematically imported from the same group, and other physical processes of systemic recurrent movements (a la colonization of Americas) and incidental impacts (a la nearly instantaneous spread of literacy) with heavy demographic impact are systematically neglected. The distorted vision turns a lively bouquet of a blend of languages brought into the family of any modern language into an unrealistic scheme fit for inclusion into unrealistic “IE theory”. Languages as diverse as analytical English, agglutinative Armenian and Ossetian, and morphologically rich Romance languages are facetiously combined into a single rubbery class.
A reader should keep in mind that the oldest writing in the world belongs to Sumerians, ca. 4,500 BC, and no claims to deeper knowledge are evidentiary supported. We can do many things theoretically, like weighing stars, because we have hard and validatable scientific toolbox, but the intuitive linguistics falls way short from being a science yet.
Is this analysis auguring the wane of the easy come - easy go biblical model and a coming of a real science? Not so fast, first we need an uncomfortable sober-up period, and a shift of the entire IE education industry. As to the nearly sanctified LIV (IE etymological dictionary), a single critical review by non-IE linguists would go a long way toward its credibility.
Page numbers are shown at the bottom of the page. The posting's notes and explanations, added to the text of the author, are shown in parentheses in (blue italics) or blue boxes.
The Indo-European Language Family: Questions About Its Status
1. 1. The purpose of the present volume is to survey the current state of the Indo-European (‘IE’) theory in the light of modern linguistic knowledge. Included in the survey is also extra-linguistic evidence, such as recent archaeological, genetic and palaeo-anthropological findings. Its ultimate purpose is to revisit the validity of the various tenets of the theory. In fact, when scholars refer to the “IE theory”, they may be referring to one of a number of competing, and sometimes contradictory, models. For example, some regard IE as purely a linguistic classification, whilst others regard it as an attempt to reconstruct a ‘real’ pre-historical language. To take another example, some scholars hold that the original IE protolanguage was a morphologically complex language, similar to Sanskrit, whilst others argue it was morphologically simple.
...in a single volume, these various approaches and views about IE that it clearly and explicitly
emerges how surprisingly different and, often, even deeply contradictory, these views and approaches
may be. Thus, the IE theory is widely accepted despite the fact that opinions differ enormously on
what the theory actually comprises. Opinions may clash even as to the very nature and validity of
many of the underlying tenets of the theory – whether explicitly stated or quietly assumed. For
example, as mentioned, some scholars regard the subject a ‘pure theory’, which helps to describe
correlations between languages, whilst others regard it as a valid means to reconstruct
pre-historical facts. Although there seems to be a widely shared assumption that at least a ‘hard
core’ of the correlations among the IE languages are ‘compelling’ – that is, too striking to be the
result of chance – there may be deep divergences on how to interpret these correlations, as well as
on other, less central but equally important aspects of the theory. In other words, contrary to what
one would expect, the wide acceptance of the IE linguistic classification does not appear to be
accompanied by a parallel acceptance of a coherent and equally agreed set of tenets, a coherent
common denominator, consisting of what one could call the ‘hard linguistic evidence’ and the
‘fundamental principles’ upon which the theory, supposedly, is based.
...This, ultimately, amounts to the task of re-assessing the founding principles of the theory.
In particular, one might reasonably ask the following questions:
The points of views and perspectives... have hardly been dealt with and confronted with one
another..., despite being inextricably interdependent. As a consequence, the questions raised in the
points (1)-(3) above have hardly been addressed in a targeted and systematic way. ...there are
plenty of publications dealing with the issue of the strengths and weaknesses of the methods of
historical linguistics... However, these publications hardly ever ponder on whether the acknowledged
weaknesses may have a negative impact on the IE theory and, if yes, to what extent. Similarly, there
are many publications revolving around “how real(ist) are reconstructions” (Lass 1993), but their
scope hardly ever extends to encompass the consequent issue of how to best interpret the IE
reconstructed, comparative corpus. On the other hand, there are plenty of publications which...
revolve only around the question of the whereabouts of the (assumed) IE proto-community.
...textbooks of IE linguistics (and, often, specialist publications too), hardly ever mention any of
these ongoing debates. ...the fact remains that textbooks typically present a highly idealized,
monolithic picture of IE: a paradigmatic, problem-free language family, where everything works
(especially sound laws, lexical and morphological correspondences), where there are hardly any
contradictions or ambiguities in the linguistic or extra-linguistic evidence, or even any
significant divergence of views among scholars. ...this idealized picture is in stark contrast with
the messy reality – in terms of variation, high level of exception, contradictory evidence, etc. –
found in practically all the other language families of the world, as well, admittedly, within
branches of IE (such as the Romance and Germanic languages, or the Balto-Slavic continuum, etc.).
This in turn has lead several scholars (such as Grace 1990 & 1996) to come up with the rather
‘aberrant’ idea that there must exist in the world two basic types of language families: the ‘the
exemplary ones’ (IE and just few others) and the ‘aberrant’ ones (all the rest).
There are, of course, textbooks that present... the poor quality and /or quantity of the evidence
in support of otherwise widely accepted theories. One may compare, for example, Szemerényi’s (1973 &
1996:122 ff.) and Gusmani’s (1979) account of the slippery evidence on which the laryngeal theory is
typically based on the Hittite side, or Sihler’s (1995:144 ff.) account of the factual and
methodological obscurities found in Verner’s Law, one of the most revered IE sound laws. However,...
it is difficult to evaluate their potential impact on the validity of the theory, or simply just to
acknowledge them. ...if, at some stage in the history of a theory the amount of evidence counter to
the model reaches what is usually called a ‘point of critical mass’, then a revision of the theory
might be in order, whether to modify or update its tenets, or, if necessary, to reject it
altogether. In other words, minimizing, re-interpreting or adjusting any evidence found to be
inconsistent with the model might be misleading, and therefore not desirable. In fact, scholars
identifying problems in their area of research may wrongly assume that the matter has been settled
beyond doubt in other areas of study, ...in this way, unwillingly, and maybe wrongly, contributing
to reinforce the validity of the theory in question.
2. The comparative method and the sound laws
2.1. The strengths and weaknesses of the comparative method
The debate revolving around the issue of the strengths and weaknesses of the comparative method is an intense, long standing debate. In fact, establishing regular sound correspondences is considered by several (many / most?) scholars to be a crucial part of the process of establishing language families (see for example Campbell (1998:315)). ...this debate has hardly ever been associated with a targeted, extensive investigation of the possible impact the weaknesses of the comparative method may have on the validity of IE as a linguistic classification. In particular, the long standing “Lautgesetz controversy” (for which see Wilbur 1977) subsided without resolution, and despite its recent resurfacing in publications dealing with several linguistic areas / families in the world (see for example Ross & Durie (eds, 1996); Blust (1996)2, Aikhenvald & Dixon (eds, 2001), etc.), it is rarely referred to in textbooks of IE. In fact, these usually assume, whether explicitly or implicitly, that the ‘regularity principle’ and the related family tree model have been established in IE beyond doubt, through the support of an extensive amount of data derived from the various IE languages.
However, this is not necessarily the case... As a consequence, the reader, including general
linguists or even experts in historical linguistics who are not acquainted with the details of IE,
may be excused if they are confused as to the actual ‘quality’ and ‘quantity’ of the phonological /
lexical correspondences conventionally established for IE. Indeed, within IE too irregularity and
exceptions do occur – it would be odd otherwise.
However, the vital questions are:
The reader who would expect an answer to these questions which is coherent, unanimous, and, most importantly, decisive – one way or the other – might be disappointed. To show that this is the case, here is a list of the most common justifications put forward to answer question (A):
1. Only the ordinary nouns in IE, particularly those referring to objects and concepts of everyday life, display a high degree of irregularity (much higher in any case than verbal roots), but this is only because they belong to the lower level of speech, the lower level stratum of the IE population (Meillet 1934: 396 ff.; Benveniste 1935:175 ff.).
2. The irregularities are only, in most cases apparent, in the sense that linguists have not yet found the appropriate explanation to account for them.
3. There are some / many / plenty of irregularities (according to interpretations), but there is nothing to be surprised about. We know that sound changes do not proceed so regularly after all, but this does not have a negative impact on the validity of IE, whose establishment, in fact, has not been based (only) on the phonological / lexical correlations, but on the morphological ones. Such a view is embraced, among others, by Harrison (2003:214 ff), Greenberg (1987:18; see also Croft 2005) and, in this volume, by Kazanas. For example, Harrison (2003: 217) states that: “If one can prove that even one single cognate pair holds over two languages, one has proven those languages genetically related”. The (implicit) claim here appears to be that proper lexical ‘correspondences’ are generally hard, if not impossible to attain. Therefore, linguists must come to terms with the fact that also within IE (like within other language families or across macro-families) the correlations among the assumed cognates are not, in the main, as ‘regular and systematic’ as generally claimed.
4. Sound changes do, in the main, proceed regularly, but the encountered irregularities are a natural effect of the great antiquity of the IE family, although its precise degree of antiquity is impossible to assess...
...the neo-grammarians worked on sound changes only ex post facto; therefore they were unable to
observe particular changes in progress.
...As to the answer to question (B), we are not aware of any research carried out with the specific purpose of ascertaining whether or not the circularity issue has had a negative impact on the soundness of IE...
...On the issue of the reliability... of sound laws, ...the following statement made by Clackson (2007: 60-61) with specific reference to the laryngeal theory: “The comparative method does not rely on absolute regularity, and the PIE laryngeals may provide an example of where reconstruction is possible without the assumption of rigid sound-laws”. This statement begs the question of why, when and where, and on the basis of which criteria, scholars may – or may not – assume the existence of “rigid sound-laws”.
2. 2. The circularity issue
If – as it seems – the circularity issue has not been solved (yet?), scholars could attempt to
set up some sort of qualitative and / or quantitative constraint to the number of the defining
parameters a given law may consist of. In practice, however, as far as we are aware, there has never
been any such attempt. On the contrary, in the every-day, painstaking practice of establishing sound
laws and correspondences, any mismatch in the evidence (ambiguity or absence of the expected
outcome, exceptions, etc.) can always “be explained away” through a range of procedures, a range of
‘adjustable parameters’, to be added to the original definition of the law. In other words, instead
of casting doubts on the validity of an assumed law (and, if necessary, dropping it) when faced with
exceptions and difficulties, typically the practitioner tries to ‘rescue’ the stated law, even at a
cost of making recourse to a (virtually unlimited) number of (often ad-hoc) adjustments, such as:
1. re-defining the law
The obvious consequence of this circularity deeply embedded in the comparative method is that the explanatory system runs into the risk of becoming so powerful, so flexible, that it can be stretched to match almost any data, in this way making it impossible to compare the results it yields against the predictions of the model. In other words, although each single ‘adjustable parameter’ as listed above may in itself reflect a plausible, genuine linguistic process, the overall cumulative effect of many adjustable parameters added to the definition of a given law may endanger the ‘cumulative effect’, the ‘statistical significance’ any established ‘law’, or even ‘tendency’, should display to deserve these names. This is an issue that has hardly ever been properly and systematically addressed...
Although the supposedly rigorous, ‘scientific’ nature of the comparative method has often been called into question, and more objective quantitative methods of analysis have been at times adopted within historical linguistics, the statistical significance of the IE comparative corpus itself (both the phonological (/lexical) corpus and the morphological one) has never been tested....
...Authors have used the IE family, whose validity is taken for granted, basically as a ‘control
case’ for various kinds of statistical tests within historical linguistics, but have not tested the
statistical significance of the IE comparative corpus itself.
...Marcantonio... argues that the great majority of the conventionally stated IE sound laws lack statistical significance and that, therefore, most of the conventionally established correspondences (within the LIV corpus) are simply similarities, most probably in the given sense of ‘chance resemblances’.
2. 3. Borrowing vs inheritance
...the possibility that the established ‘cognates’ – be they ‘similarities’ or proper ‘correspondences’ – may be due to the common processes of borrowing, diffusion, convergence, or even chance resemblances. As is known, borrowed words tend to integrate into the sound pattern of the receiving language, as well as undergoing the same (more or less regular) changes that inherited words would undergo. Thus, the identification of borrowed elements on the basis of internal, linguistic clues only might not always be easy. Therefore, sound correspondences, whilst fundamental to most approaches in assessing language families, “can be misused”...
...several semantic fields within the IE basic lexicon..., in addition to being mainly irregular,
typically lack a wide distribution across the IE area, being often confined to just two or three
contiguous languages... In contrast, the cognate terms for ‘mother’, ‘father’, ‘brother’,
‘daughter’, etc., display a much wider distribution, and a higher degree of regularity. This factor
has correctly raised the suspicion that processes of loss (and consequent replacement) of original
words, or even processes of chance resemblance, may have been involved in this area. The issue of
the wide vs restricted distribution and the (degree of) irregularity of many basic cognates within
IE is interpreted differently by different scholars... ...patterns of original words are typically
retained, and are therefore easily identifiable even within a context of extensive borrowing...
2. 4. Is the phonological evidence malleable?
...one could certainly argue that the conventional, phonological / lexical evidence on which the IE theory is based, to a closer scrutiny, appears to be rather ‘malleable’ – it is certainly not as decisive (one way or the other) as generally claimed. As a matter of fact, it is always possible to find a plausible, although not always testable, justification to any intervening piece of evidence counter to a stated rule or tendency. Similarly, it is also always possible to provide at least two equally plausible, equally well founded explanations for any given linguistic phenomenon.
This is also the case within non-controversial areas of IE linguistics, such as the postulation of the so-called ‘Indo-Iranian branch / unity’. One can in fact compare the different interpretations given to this unity – (also) on the basis of phonological /lexical evidence... Schmitt... argues that the data from both Vedic and Sanskrit are not with absolute necessity genuinely antique and that, therefore, Old Indo-Aryan is not as close to PIE as still believed by some scholars...
At this point one could object that all these methodological and factual difficulties do not, after all, matter, even if they did impact negatively on the validity of the conventionally established sound laws (and related correspondences and reconstructions). This is because... the lexicon is often considered to be the level of language less (or not at all?) relevant for the purpose of assessing genetic relationships. This would be sound and acceptable if there were indeed a consensus among Indo-Europeanists as well as comparativists in general on the principle that it is morphology the level of language (mostly?) relevant in this context...
3. The role of morphology
3.1. Is morphology the most reliable indicator of genetic inheritance?
Since the very times of the establishment of IE the prevailing opinion appears to have been that grammar can offer the most reliable evidence for assessing genetic relatedness. Grammar has typically been considered to be a rather stable level of language and, often, totally resistant to borrowing – in contrast to the volatility of the lexicon. These properties have made morphological correlations quite popular among many historical linguists, even if it has always been (more or less openly) recognized that it is not clear what the criteria are, if any, on the basis of which to identify and evaluate morphological similarities.
In fact, not even the regularity principle is expected to consistently operate at this level, due to the overwhelming interference of the analogical principle.
In addition, there has never been... any systematic attempt to define a possible measure, a ‘unit
of similarity’ (in form and / or function) – as it were – to be applied in the practice of comparing
morphemes. This measure of similarity would work as a common denominator against which to evaluate
the at times vaguely similar, at times very similar and at times identical morphemes occurring among
(most / some) IE languages. Thus, the problem is that the morphological correlations are typically
observed and established by intuitive, visual inspection (often by single scholars), whereby
considerable latitude may be allowed when it comes to phonological forms as well as to similarity of
functions. For the morphological (and morpho-phonological) correlations to be rigorous, to be proper
‘correspondences’, one should certainly require regular phonological correspondences between morphs
which also indicate similar (but how similar?) meanings and /or functions – condition which is hardly
ever met. Indeed, one often reads in the literature that the grammatical correlations within IE are
(still nowadays) simply and purely ‘obvious’ to the ‘naked eyes’ of the trained philologist (see
Nichols 1996a), exactly as they appeared to the first scholars who dealt with them a couple of
3. 2. The degree of ‘borrowability’ of grammar
In recent years a mounting body of evidence has been accumulating according to which not only
grammar is found to be ‘borrowable’, but, given the appropriate historical and social context, it
may rate quite high on the scale of borrowability. It could therefore be difficult to determine
whether shared grammatical innovations are the result of genetic inheritance or of areal
convergence. ...is grammar borrowable to such an extent that historical reconstruction becomes
...within IE studies there is still an open question regarding the nature of the original morphological structure of PIE: was PIE rich in morphology (as is the case, mainly, of Greek and Indo-Iranian), which has then been ‘reduced’ or ‘lost’ in the other languages, or was it rather poor in morphology, in which case the complex morphology of Indo-Iranian and Greek is the result of parallel, shared innovations, rather than of genetic inheritance?...
Still on the issue of the ‘borrowability’ of grammar, it has been claimed at times that the IE morphological correlations are, on the whole, similar enough to be considered valid correlations but different enough so as not to raise the suspicion that borrowing might have been involved. However, certain IE grammatical forms typically reported in textbooks as ‘obvious’ examples of genetic inheritance – such as the paradigm of the verb ‘to be’, or ‘to bear, carry’ – are so similar, if not in some forms identical across the area, that the suspicion of borrowing may indeed arise. In fact, one would normally expect much more divergence from a long process of inheritance and development. ...In addition to this, one should take into consideration the numerous morphological correlations which supposedly connect the IE family with other contiguous, but different language families, as argued for by the supporters of the so-called macro-families...
3. 3. Is the morphological evidence malleable?
At this point one could object that the risk of reconstructing false matches within IE grammar is rather low, since one can rely on a wealth of shared (inflectional and derivational) morphology in a great variety of areas. Furthermore, although much of the grammatical evidence put forward by traditional IE studies is certainly rather intuitive and subjective, as it happens, this evidence turns out to fall within the range of what Nichols (1996a: 49 & 64) calls “diagnostic evidence”:
Traditionally linguistic kinship was identified on the basis of diagnostic evidence which is grammatical and combines structural paradigmaticity […] and syntagmaticity with concrete morphological forms. The Indo-Europeanists’ intuitive feel for what was diagnostic evidence of relatedness corresponds to a computable threshold of probability of occurrence […]. A grouping can be regarded as established by the comparative method if and only if it rests on individual-identifying evidence
The idea is that one can compute the probability of
occurrence in the languages of the world of certain features, certain patterns, such as consonant
sequences (in the dimension of syntagmaticity), or the co-occurrence of some grammatical categories
and their morphological indicators (in the dimension of paradigmaticity), etc., whereby the Author
assumes that a probability of occurrence of a given phenomenon of 1 in 100.000 is
“individual-identifying”. Within IE one finds several, arguably a statistically significant number
of instances of individual-identifying diagnostic evidence.
The following data, as proposed by Nichols herself (1996a: 47), are among the most quoted ones for the purpose of this discussion (notice that to the Latin and Greek endings reported in this table at least the corresponding Sanskrit endings should be added to complete and strengthen the picture...)
...there is plenty of evidence that wholesale (nominal and verbal) paradigms of the type under
discussion can also be the effect of borrowing, as shown by Matras’ contribution... Second, not all
the IE languages do possess those rich, (more or less) consistent, paradigmatic morphological
systems we would need to establish a wealth of ‘individual-identifying’ grammatical evidence.
Whatever the case, one has to admit and reflect on the fact that the morphological reconstructions are essentially based on a few core languages, whose morphological correlations – whether the result of archaism, or innovation, or borrowing, or convergence – are nevertheless well attested, whilst “the attribution of the other (so called “IE”) languages to the (“IE”) family is necessarily done on partial evidence”...
3. 4. Are morphological reconstructions predictive?
It is reasonable to ask at this point whether the circularity issue (as discussed above with
regard to phonology and lexicon) could have a negative impact also on the morphological
reconstructions, including the area of diagnostic evidence. In fact, here too scholars can increase
the power of the comparative method by making recourse to a number of ‘adjustable parameters’.
Actually, the risk of circularity is higher at the morphological level, because of the (supposed)
pervasive operation of analogy and the lack of a rigorous definition of ‘morphological correlation /
similarity’... some case endings cannot be reconstructed in an economic, straightforward way, or
cannot be reconstructed at all, so that a number of ‘adjustable parameters’ have to be introduced in
order to overcome the observed mismatches. The adjustable parameters in question include:
1. a chain of (at times unverifiable) assumptions, including syncretism and reshaping through
3. 5. Counter evidence within morphology
It has been argued above that the morphological evidence on which the IE theory is founded
appears to be rather ‘malleable’, this problem being caused by the general lack of criteria and
guidelines on how to evaluate morphological correlations. This in turn means that it is equally
difficult, if not impossible, to identify evidence potentially counter to the model (with perhaps
the only exception of the evidence from languages in contact that wholesale morphological paradigms
can be borrowed). For example, the absence of the fundamental IE category of feminine gender in
Anatolian23 could be considered a paradigmatic example of the inability on behalf of the IE theory
(and the methods of historical linguistics in general) to make very clear-cut, testable predictions.
There have been no claims (as far as we know) that the absence of this ‘diagnostic’ category should
be classified and set apart as potential evidence counter to the IE origin of Anatolian (Hittite,
for example, has no nominal declension corresponding to the feminine stems). Quite the contrary,
efforts have been devoted only to finding traces of the feminine gender here and there or to justify
its absence in the most plausible, natural way... The same could be argued with regard to one of the
most puzzling aspects of Hittite verbal morphology, the ‘qi-conjugation’, which, admittedly, does
not easily slot into any reconstructed category of IE (although there are claimed to be some good
lexical correspondences between verbs adopting this conjugation and verbal roots in other IE
languages...). The situation does not appear to be much different at the level of morpho-phonology,
as is evident from the state of the research regarding the IE Ablaut (vowel alternation / apophony)
(A vowel whose quality or length is changed to indicate linguistic
distinctions (such as sing sang sung song). Here, the scanty and irregular
distribution of vowel alternations across the IE area has generated an array of complex,
sophisticated explanations, none of which however appears to be either testable or satisfactory...
Carruba suggests that the original function of Ablaut is to be sought in the ‘deixis’ – a rather
archaic, elementary function, but a very productive one within IE.
He ascribes the poor attestation of coherent patterns of vowel alternations across the IE area to the very ancient origin of phenomenon itself. This is certainly a plausible explanation; however, again, there may be alternative explanations. For example, it could be argued that the Ablaut present in Sanskrit (and partially in Greek), does not represent a phenomenon to be traced back to the IE proto-language, but on the contrary, a language-specific phenomenon. ...Sanskrit would be the only IE language that has roots and a “proper vowel gradation”. ...‘statistical significance’ (or lack of it) of the present vs perfect vowel alternation...
3. 6. Is there hard core evidence at the foundation of Indo-European?
Bearing in mind the shortcomings of the morphological correlations as discussed above, it is
understandable how several scholars, such as Campbell (2003a & b) and Morpurgo Davies (Cambridge
Seminar 2005; see note ) have reiterated and emphasized the role of the phonological (/lexical)
correspondences for the purpose of assessing genetic relations. One could compare also the case of
Proto-Dravidian, where it is the lexicon that has seen the greatest amount of relevant, comparative
work, as illustrated by Steever (1996). As a matter of fact, phonology has at times offered us the
possibility of testing the validity of the comparative method, in the sense that some predictions
made by the comparative analysis have subsequently been proven correct in connection with the
discovery of new linguistic material. For example, new evidence from Hittite and Linear B Mycenaean
has confirmed the validity of earlier reconstructions, in this way also allowing us to redefine the
processes which lead to the attested languages...
These are undoubtedly remarkable discoveries, and fine analyses. However, the data and analyses under discussion are not void of difficulties... Furthermore, one could observe that the more linguistic material, the more comparanda one brings into the equation whilst doing comparison, the easier it is to find a phonological /lexical (as well as morphological) match with a given form or reconstruction. As Campbell (1998: 277) puts it, referring to the issue of mass-comparison and macro-families: “The potential for accidental matching increases dramatically … when one increases the pool of words from which potential cognates are sought or when one permits the semantics of compared forms to vary even slightly”... Hock (1993:221ff.) observes the following: “Preliminary results [of an experiment in which English, Finnish and Hindi are compared] suggest that enlarging the data base does not improve the reliability of the method. In fact, if there is any change at all it may consist in a slight increase of false friends”. On the other hand, Hock (ibidem) also recognizes that working through a comparison of only a small number of languages (three in the specific case), may be equally misleading.
Finally, one could always argue that a few instances of fulfilled predictions – the occurrence of the right, predicted reflex at the right place – are not enough to prove the supposed predictive nature of all the conventionally established IE laws.
As to the laryngeal theory, it is undeniable that it represents one of the most outstanding
examples of a successful theory within historical linguistics. As Andersen (2007)30 puts it: “From a
purely algebraic theory in Saussure; given putative phonetic justification by Hermann Moeller; then
actually justified by the discovery of some of the posited segments in Anatolian; found to correlate
with the 'prothetic vowels' and 'Attic reduplication' in Greek, and of 'final lengthening' in Homer;
and then with the Baltic and Slavic distinction between acute and circumflex long vowels and
Nevertheless, the laryngeal theory is not uncontroversial, at least with regard to the number of laryngeal segments to be postulated. ...However, the main problem associated with the laryngeal is that the advantages attained by making recourse to the laryngeal theory are counterbalanced by a number of disadvantages. For example, Winter (1990: 20-1) referring to Saussure’s original idea of the ‘coefficient sonantique’ and subsequent developments, describes one of the known ‘disadvantages’ of the laryngeal theory as follows: "We attempt to simplify our statements by subsuming overtly differing phenomena under one common formula which may or may not require positing directly unattested elements conditioning the differences [….]. This analysis had the tremendous advantage that now all subtypes of ablaut in Proto-Indo-European could be treated alike and that canonical forms of roots could be set up; along with this, however, went the disadvantage that the Proto-Indo-European system of vowels apparently had to be reduced in an unreasonable way."
Sihler (1995:111 ff.), who reconstructs an Ablaut system “considerably leaner” than others thanks
to the support of the laryngeal theory, appears to (implicitly at least) identify another, major
‘disadvantage’. In fact, he states that the economy of his model is achieved by “a complication
elsewhere in the system”, because his reconstruction also requires “a number of sound laws” applying
to the postulated laryngeals. Here, the fundamental questions (which have not been asked yet, to our
Thus, if we accept the analyses, objections and counter objections expounded thus far, it is still not clear what exactly the uncontroversial, hard core evidence that lies at the foundation of the IE theory is supposed to be.
...Last, but not least, one should take into consideration also those further limitations of the
comparative method as pointed out in recent years by scholars such as Dixon (1997), Aikhenvald &
Dixon (2001) and Nichols (1992, 1993/1995, 1996b, 1997 & 1998); see also Andersen (2006). These are:
4. Is there a pre-historical reality behind reconstructed Indo-European?
4. 1. Conventionalism vs realism
The ‘conventionalist’ vs ‘realist’ approach to linguistic reconstruction, and related concept of proto-language (in general as well as within IE), is another long standing debate within historical linguistics, and one for which, yet again, there does not appear to be much of a consensus... It is true that between the extreme conventionalist approach on the one side and the full realistic approach on the other there are plenty of more moderate, intermediate positions. Nevertheless, it appears that the fully realistic approach, which in turn is based on the controversial method of palaeo-linguistics, is the one that has so far attracted many supporters, and not only among archaeologists (such as Mallory (1989 & 2001) and Renfrew [1987, 1990]), but also among linguists.
For example, in his criticism to the thesis by Gimbutas (1970) regarding the location of the IE home land, Schmitt (1974: 283) observes that “[with] the methods of linguistic paleontology anything can be proved as Proto-Indo-European, but it can not be proved as typically Proto-Indo-European”. Thus, Schmitt appears to have confidence in the validity of the realist approach, although he has doubts regarding the factual interpretation of single reconstructions and the hasty conclusions reached about the reconstruction of single ‘pre-historical facts’. It could be said that in this case too (like in the case of the ‘regularity principle’) it is not always made explicit what the motivations are, what the evidence is, that provides the basis for the choice of one approach, or the other.
...Di Giovine, on the basis of the analysis of the IE verb inflectional system..., argues that PIE was not a compact language or not even a community language at all.
4. 2. The Aryan debate
Within the camp of the realist approach the debate has revolved mainly around the issue of the whereabouts of the (assumed) IE proto-community. A strand of this debate has recently re-surfaced under the definition of: 'the Aryan Debate'... The Aryan debate centers on the following, major interconnected issues (all questions are linguistic, although framed demographically, but independent from the popular linguistic schemes of elite domination, code change, lingua franca, etc.):
1. Was the assumed homeland located in the west, somewhere in Europe, or in the east, somewhere
in North India /Pakistan? In other words, were the bearers of the Old Indo-Aryan (Vedic and
Sanskrit) culture indigenous or intrusive in North India?
The issues raised in points (3) and (4) are of particular relevance, not only because they are
strictly connected to the home land issue, but also because the assumption of the great antiquity of
(Vedic) Sanskrit is (or, at least, it has been) one of the pillar tenets at the foundation of the IE
theory, and related traditional reconstructions. As we have seen above (par. 3. 3.), this tenet is
now under attack, since several scholars believe that the Greco-Aryan morphological paradigm
represents in fact innovation, and not archaism. As to the more specific issue of the archaicity of
Sanskrit, Schmitt in this volume calls into question this traditional view, arguing that:
“the archaism particularly of the Old Avestan language makes it only too clear that, despite the old age of the earliest Vedic texts, the Old Indo-Aryan language is not the only fundament of the IE proto-language […] and that its data are not with absolute necessity genuinely antique; therefore Old Indo-Aryan (both Vedic and Sanskrit) is not so close to PIE as many people think”.
On the other hand, Kazanas... supports the traditional way of thinking as well as the indigenist model claiming exactly the opposite: Old-Indo-Aryan is indeed the oldest language within the family (actually, much older than conventionally assumed), as well as the closest language to PIE, even though a PIE language cannot be reconstructed. In turn Drinka, in her contribution to this volume, rejects the indigenist / ‘Out-of -India’ model, arguing against it (mainly) on the basis of those very “archaic” morphological isoglosses Indo-Aryan shares with some IE languages and those very “innovative features” it shares instead only with Iranian and Greek.
4. 3. Sanskrit and the South Asia linguistic area
This already articulated debate is in turn connected with the other, equally tangled issue of how to interpret the lexical, phonological, structural and even morphological correlations that have been identified as existing between the Old (and Modern) Indo-Aryan languages and the other non IE languages of India, such as (mainly) Dravidian and Munda. In fact, these (and other) various languages / language families of India are widely claimed to form what is usually referred to as the ‘South Asia linguistic area’, for which see Steever (1993:10 & 1996:11 ff.); Masica (1976) and Emeneau (1956, 1971 & 1980). In particular, on the basis of (supposed) lexical borrowing from Dravidian into Indo-Aryan and from the modern geographical distribution of the Dravidian languages, it is often claimed that Dravidian and Indo-Aryan must have been in contact – since prehistoric times– in those extreme northwestern areas of India first inhabited by the Indo-Aryans...
Thus, the relevant question here, the question which has been hotly debated and whose resolution, if ever attained, could help to break the deadlock of the indigenist vs migrationist model – together with the issue of the degree of antiquity of Sanskrit – is the following: Are the non IE features present in Old-Indo-Aryan the result of sub-stratum, super-stratum, ad-stratum, convergence, or even genetic inheritance?...
Several scholars... have compiled long lists of (Rig-) Vedic words believed to be of Dravidian or, more generally of non-Aryan origin. Other scholars, particularly Hock (1996: 36 ff.), have drawn attention to the fact that the situation is not that clear-cut after all. In fact, in many instances it is difficult to trace back the origin of the lexical as well as structural similarities observed in this intricate linguistic area, and to sort out whether they are attributable to borrowing, convergence, or chance resemblances. On the other hand... Indian scholars... challenge the traditional classification of Sanskrit as an IE language, observing that under a different socio-politico-cultural context, the similarities under discussion could have been interpreted differently...
For example, they could have been ascribed to the postulation of a shared, common ancestor for Sanskrit and Dravidian, and the studies of the South Asian linguistic area, as we know them today, might have taken a totally different course, generating a different production of knowledge, and, possibly, a totally different proto-language as the ancestor of Sanskrit. In turn, these scholars are accused to be strongly biased in their research, being driven by nationalistic feelings and political motivations. Certainly – it has to be said – there are elements of truth in this statement, since the main concerns (mostly unexpressed) of these Indian scholars are for the cultural unity of India under a Sanskritic proto-banner, whereby Dravidian would be some sort of Sanskritic Prakrit...
Nevertheless, it is also fair to recognize that had the events, the accidents of History been different, in other words, had Sanskrit not been brought to the attention of the western scholars (as a consequence of the English colonization of India,...), the cultural scenario of historical linguistics might well have been totally different indeed, and one can easily imagine how a different paradigm might have been created to account for the origin of Sanskrit and the other languages (/language families) of India. Actually, the fact that the initial choice of comparing a certain pair / set of languages, rather than others, is by its own nature ‘intuitive’ and may be also influenced, or even dictated by circumstances external to linguistics, can increase the risk of circularity embedded in the traditional methods of analysis, as discussed above... (The classification of Sanskrit is not the only known instance of linguistic classification which has been influenced by historical /political events. ...“The operation of the comparative method does not guarantee a language’s place in the family; only the initial recognition that two or more languages are related can do that […]. When does a linguist decide that there is enough material to relate a language to the IE family? There is no absolute set of criteria beyond the general rule that the evidence must convince both the individual linguist and the majority of the scholar community”).
It would therefore be desirable to be able to ‘demonstrate’ that the correlations shared by Sanskrit and Dravidian, for example, are the genuine effect of a typical process of Sprachbund convergence, rather than simply ‘labeling’ them as ‘contact-induced’ just because it is widely accepted that the languages in question are not genetically related... ...The thesis that the correlations shared by Sanskrit and Dravidians are the genuine effect of Sprachbund.... ...the issue of the origin of the non IE features present in Old (and Modern) Indo-Aryan languages... argue that Dravidian is a clearly distinct linguistic family from Indo-Aryan, although the latter does indeed display Dravidian features. In other words... “some of the four genetically distinct language stocks in South Asia have clear genetic linguistic relations outside of the subcontinent”. ...a comparison of two classical languages of India, Sanskrit and Old Tamil,... would reveal and illustrate a typical Sprachbund situation, where “two languages may be genetically distinct, yet grammatically related”.
At this point it is interesting to observe that there is at least one aspect within this
intermingled debate about which there appears to be consensus among all scholars: there is no
archaeological evidence for the assumed migrations of (part of the) original IE community, either
from a supposed western homeland eastward into North India, or the other way round...
...the explanation of the spread of the Indo-European languages as “the spread of the culture of a concrete group of people from a cradle […] has not been proved by anthropology and archaeology as yet”... ...“all attempts to reconstruct the old culture of the Indo-Europeans as existing in a concrete cradle, by the means of ‘linguistic paleontology’ are wrong”.
...scholars are faced here with yet another area of IE studies where the ‘evidence’ (or lack of it) is malleable and un-decisive. As a matter of fact, Häusler himself believes that “the IE linguistic community must have existed at some point in ancient times, since the linguistic classification has been fairly safely established, even if this (assumed) linguistic community is not at all retrievable by the means of archaeological and anthropological research”... Whatever the case, the fact remains that, thus far, the required, supporting (one way or the other) archaeological and palaeo-anthropological evidence has not been found.
As to the palaeo-linguistic evidence – place and river names, terms for flora & fauna, names of deities, etc. – mainly from Sanskrit, some scholars claim that this evidence is malleable, and therefore inconclusive when it comes to trying to establish the indigenous or intrusive character of Old Indo-Aryan... ...other scholars claim that the palaeo-linguistic evidence clearly supports the indigenist / ‘Out-of-India’ model...
...it is worth mentioning that recent genetic evidence (pre-2009) (whatever its relevance may be in this context) appears to support the indigenist hypothesis of the origin of the Sanskrit speaking peoples and culture...
4. 4. The conventionalist approach
Those scholars who adopt the conventionalist approach to reconstruction claim that linguistic
classifications do not necessarily imply and guarantee the existence of the corresponding speech
community. Therefore, searching for the original IE community, and the pre-historical, highly
intertwined linguistic and extra-linguistic processes that would have brought about the IE
languages, is a pointless task. This remains the case even if linguistics is assisted by other
disciplines whose methods of analysis are claimed to have recently achieved high levels of
reliability, such as archaeology and genetics. In other words, comparative historical linguistics
cannot in any way shed light onto pre-historical ‘facts’ (homelands, migrations, institutions etc.).
Historical linguistics was not meant to be, it is not and cannot be used as a ‘branch of
5. Introduction to the chapters (in alphabetical order)
Marcantonio also argues that adding laryngeal segments to the process of reconstruction (whatever
the rights and wrongs of the laryngeal theory may be), dangerously increases the explanatory power
of the comparative method, in this way further contributing to the flexibility of the overall
The reader has seen... a variety of views about IE, ranging from the belief that it represents the language of a real pre-historical community; through the thesis that it is only a model to embody linguistic correlations; all the way to statistical evidence that (many) linguistic correlations themselves may be merely an artefact of the method of analysis. In fact, when the various components of the theory are brought together so that they can be seen holistically, it is hard to pin down what the foundations of the theory are actually supposed to be.
For example, one of the founding principles of the traditional version of the theory was the
assumption that morphological paradigms cannot be borrowed, and therefore it is possible to trace
genetic inheritance through them. However, we have seen evidence of wholesale paradigm borrowing,
based on studies of languages in contact. In any case, some scholars now hold that morphology is
less relevant than other factors – but it is at present unclear whether, or how, these other factors
may be verified or falsified...
In Russian (later)
Contents Türkic languages
Contents Türkic in English
Classification of Türkic languages
G. Ekholm Germananic Ethnology
C. Stevens Gmn.-Türkic traits
A. Toth German Lexicon
A. Toth Turkic and English
R. Mc Callister Non-IE in Gmc. languages
Türkic borrowings in English
Türkic in Romance
Alans in Pyrenees
Türkic in Greek