Speculation and error correction in Palaeolinguistics
There is hope that the advent of the Internet and the large number of people interested in the field, along with other events like the end of Communism and the maturing of social anthropology, can bring about a new era in ancient linguistic history and resolve some of the outstanding questions. At the same time, there is a large amount of uncontrolled speculation in the field, and my interest in the field at this point came partly from watching 'The X-Files' and having a hankering for weird ideas. 'Psychoceramics' is the scientific study of crackpots and this essay is partly in the psychoceramic field. It is all about Europe, and mainly about the northwest of that continent.
It was researched late 2008 to early 2009 and written up in April 2010.
In what follows IE stands for ‘Indo-European’, AS for 'Anglo-Saxon', PCT for 'Palaeolithic continuity Theory'. For bibliography see a posting elsewhere on this site.
1. Is English a 4th branch of the Germanic languages? English as an insular language. Did English reach England before Caesar?
2. the Palaeolithic Continuity Theory. Was Indo-European already spoken in Europe 10,000 years ago?
3. Was Northumbrian a 'creole language' and subject to 'premature loss of morphology' as a result of a heavy Celtic substrate influence? did a most-mortal Celtic element influence the development of middle English?
4. The question of substrates in Germanic. Is a large part of the common Germanic vocabulary borrowed from another language?
5. modifications to the Renfrew Thesis.
Does the theory (that the Neolithic immigrants to Europe were speakers of Indo-European) hold up?
If the area north of a certain line is the zone of 'contact diffusion' (as opposed to immigration) can we explain this by saying that Celtic and Germanic are 'broken European' and radically reshaped by substrate influence? The boundary between 'demic diffusion' and 'contact imitation' in the Neolithic spread of farming.
6. Is Theo Vennemann's theory that North-West Europe was Basque-speaking in early times (and possibly in the Palaeolithic?) correct? Is his other theory on the 'Semitidic' language affecting the Atlantic coast of Europe in Bronze Age times correct? are we all Basques?
7. The Morris-Jones theory of a Hamitic influence on Insular Celtic. Can we sustain it, has any new evidence been developed?
8. Nordwestblock theory. Was there an area (covering the Low Countries and northwest Germany) which in late prehistoric times was 'non Celtic non Germanic' and which spoke a different language ('Belgisch')? how far is this substrate represented in English? Was this the language spoken by the Belgae in England?
9. Schrijver's Minoan-Hattite-Celtic thesis. Is the Celtic verb related to the Hattic language spoken in Central Anatolia around 2000 BC?
1. Is English a 4th branch of the Germanic languages? or, English as an insular language.
Did the Belgae credited by Caesar with immigrating to Britain in the 2nd-1st Cs BC (roughly) speak a Germanic language and was this the ancestor of Anglo-Saxon?
summary of theory. There were no Anglo-Saxon invasions. The density of population reached by the 5th C AD in Britain was too high for significant amounts of new human material to fit in. The major change, from Celtic to Germanic language, therefore happened when the population was much less dense, before the Roman invasion and the prosperity it brought. The Belgae, reaching Britain from Northern France or Belgium in the last centuries before Caesar, spoke a Germanic language and this is the direct ancestor of modern English. There are massive resemblances between English genes and those of other Germanic areas around the North Sea, but the density of this genetic material argues against it arriving between 300-600 AD (roughly), so it must have already been there. The landing of a few Saxon pirates during the sub-Roman confusion is neither here nor there. English is a fourth branch of the Germanic languages because of this long separate development. Its resemblances to other north-west Germanic languages are deceiving and can be argued away.
Comments. This theory comes out of genetics and archaeology and its assaults on linguistics are purely attempts to bully the evidence into fitting a prejudged conclusion. The projection of genetic findings in 21st century populations onto the biological situation of 2000 years ago is highly inaccurate, not to say speculative. Dating the arrival of 'Germanic' genes in southern England is not convincing, although one day it may be. The theory appeals to a lot of people because of its rejection of foreigners - the Saxon invasions are abolished and the English have been English for 2000 years, or maybe 3000.
There was a migration of Belgae to Britain, in about the 2nd C BC, maybe even later, as identified by Caesar around 54 BC, in his surviving writings about Britain. The 'Belgic' area is in the 'nonCeltic nonGermanic' Nordwestblock area, as identified by Maurits Gysseling in the 1950s and Hans Kuhn somewhat later, so they could have spoken ‘Belgic’. (See article 5 below.) However, Gysseling also identifies a wave of germanisation of this area in the 3rd and 2nd Cs BC, so that it was 'stark germaniseerd' (heavily Germanised) at the time of their late migrations to Britain, and so those Belgae will have been partly of Germanic (or Germanic influenced?) speech. A lot depends on the dating of the Belgic invasion, which has wide acceptance among archaeologists as well as being attested to by Caesar. Possibly it means Germanic speech in England, possibly it meant an influx of speakers of a ‘third group’ or Nordwestblock language (see below article 8).
Further, there were certainly troops from Germanic Europe stationed in England as auxiliaries of the Roman Army, and some of these were probably given land when their time had expired, so that they settled and left offspring.
There are early Latin words in Anglo-Saxon, for example 'cheese' (caseum) 'church' 'street'. The important area is such words which exist in English but not in Continental Germanic dialects. There is such a stratum, about 200 words. However, the Saxons could still have acquired these words on the Continent, if they were living in the Seine-Maas area and in contact with speakers of Vulgar Latin.
The 4th branch theory requires the ancestor of English to have been spoken in England since the 2nd C BC at the very latest. This would mean roughly 400 years of Saxon speech living under Roman domination and in close contact with Latin speakers. Welsh has hundreds of Latin words from this phase of history. The ‘long’ history should also mean massive borrowing from British, agreed to be the main language of the island, and of the areas of the island which the Belgae dominated as an upper class. Both strata are positive evidence which would have survived, despite some attrition, to be recorded in early AS texts around the 7th C. But they are few in number and the Latin loans could have been acquired on the Continent. On this basis the 4th branch theory fails and is knocked out.
There is also the question of early Germanic words in Welsh. The count here may not be zero but there are certainly very few. You could argue that Welsh descends from the British of an area that had no Belgae, and the influenced areas lost their speech without leaving a record. This is still no evidence. If they did not acquire the words it is because a Germanic language was not being spoken in Roman Britain.
2. The Palaeolithic Continuity Theory (PCT). Was Indo-European already spoken in Europe 10,000 years ago?
This takes the Indo-European continent which we find in 500 AD and projects it back onto the Old Stone Age. For these people English was spoken in England before the land link between England and France was flooded, in the Palaeolithic. Immigrations from Africa and Asia are thus discounted. Mario Alinei is the ideologue of Continuity. Some texts are here:
comment. I found this theory pointless because without written record there is no evidence of what languages were spoken 10,000 years ago. The basic problem with any Palaeolithic theory about language is that those people never used writing. We have no idea what languages they spoke in Europe or Asia. This theory has nothing to do with linguistics. The resistance to influences, and blood, coming into Europe from outside is suspect, and links with the beliefs of the Liga del Nord and other Italian groups opposed to the recent immigration into Italy (from North and East Africa and to a lesser extent from the Near East). The map of linguistic events in the New Stone Age will never be complete and thus lends itself to clever re-arrangements of speculations into new forms.
The Indo-European languages bear striking resemblances to each other. This is one of the key limiting factors in dating the IE Diaspora. It has to be early enough to get as far as Ireland and central-north India before history starts, but it has to be late enough for the amazing resemblances in morphology between Sanskrit, Greek, and Latin not to have been eroded by time. This constraint is fundamental to all building of models for early European language history, and PCT does not even address it.
If Alinei could find one tenable flaw in the "immigrant Indo-European" thesis then the music would start. But he does not.
Since most of Europe was under glaciers and uninhabited during the last Ice Age, the question of what language was spoken in those regions need not be answered.
The idea that artefact cultures didn't change between Palaeo- and Neolithic ignores the fact that they largely abandoned hunting-gathering and picked up on farming and stockbreeding. How much more discontinuous could anything be?
As has been pointed out (John Kozak, personal communication) there was an 'empty Europe' in the last Ice Age and a Mesolithic irradiation into a habitable post-glacial continent, (in millenia up to 8000 BC) and there is at least a possibility that this single point of origin homogeneity corresponds with the IE linguistic homogeneity. (Perhaps not single but few, perhaps from three or four entry points or refugia.) Proto-Indo-European was presumably spoken somewhere in the Mesolithic, but we do not know where that was. Wherever it was, it was a compact and coherent area. The genetic evidence points to a repopulation of Europe in Mesolithic times and puzzlingly limited 'replacement' of the population thereafter. Arguments about this may be largely to do with speculative models but deserve addressing. How does this sit with the hard linguistic evidence linking Europe to Anatolia, Iran, and North India?
The PCT stuff promises a delightfully pleasurable trip through thousands of years of drifting groups in a fabulously pristine landscape, but the game can't start unless someone produces some evidence. On the contrary, there is a kind of scientist who homes in on areas where there is no
evidence. A summary of early European linguistic diversity using modern theory based on findings in other continents is here:
This posting demolishes the PCT. It identifies a likely pattern of many (40?) non-IE languages in Europe which failed to reach written status (or, only through tenuous strata of loan words or perhaps structures of phonemes or intonation etc.). I find this intuitively attractive and I think most Indo-Europeanists would buy into that picture. I think the work of Ernst Lewy (1881-1966) sets out from that picture and has captured parts of it, as far as lost languages ever speak.
3. Was the Northumbrian dialect a 'creole language' in circa AD 600-900 and subject to 'premature loss of morphology' as a result of a heavy Celtic substrate influence? Did the Celtic majority population of the early decades after the Anglo-Saxon Settlement influence the Anglo-Saxon language before it was finally recorded in writing? Did ‘Celtic speech habits’ continue to shape the development of English?
This theory of a Brittonic-Northumbrian mixed language is reported in Katy Wales’ book on Northern English, as it was worked out in a paper by three Finnish scholars: Marku Filppula, Juhani Klemola, and Heli Pitkänen (and now in their whole book on Celtic influences).
For a discussion of the term 'creole' see article 5 below.
To assess this theory we would have to answer (at least) four questions for any 'loan' feature:
(a) Was this feature missing from Anglo-Saxon in early times
(b) was this feature present in Welsh in early times
(c) is Welsh the only candidate source
(d) Is this a loan feature or an "efficiency" feature related to the general loss of morphology
It is agreed that English has evolved through progressive loss, over a very long period, of word endings, tools essential to the language in its beginning and which there were other very good reasons for not losing. In 450 AD England must have been full of Celtic speakers, using a language we can call 'British', which was undoubtedly related to Cornish, Welsh, and a language later spoken in south-west Scotland and NW England ('Cumbrian'). They were an advanced agricultural population who must have reached a high population density (between four and seven million people according to a recent estimate). The Celtic loanwords in Old English are very rare and the disappearance of the Romano-British and their speech is a major puzzle. It is one where the evidence disappoints expectations, and where numerous writers have distorted the evidence to fit comforting preconceptions about 'English continuity' and tranquillity.
The laws of King Ine of Kent distinguish between English and Welsh (wylisc) population goups in the early 7th C. Presumably the ‘wylisc’ population went through a bilingual phase after the conquest, perhaps including a phase of speaking ‘broken English’ along with P-Celtic. By 700 AD there is general agreement that these speakers had disappeared, being assimilated to the Anglo-Saxon speech community. They cease to be identified as a separate social group (of low status) in the law codes. The proposal is that the legal disadvantages waned and vanished between Ine and the laws of Alfred dated circa 890 which no longer have that status distinction. The Celts in England all merge with the Saxons by 700, or else we would hear more about them. This process was taking place during the formation of Old English and indeed can be defined as a part of that process. The question is whether Old English, as finally written down when literacy arrived, had absorbed some abiding influences from the bilingual Celtic under-class. The Finns say that this influence showed itself in much later linguistic changes, in the transition from Old English to Middle English and in changes in late Middle English (so 12th to 15th centuries AD). These changes are mainly the simplification of verb and noun endings and the rise of the periphrastic verb (so ‘Did you speak?’ for ‘spakest thou?’).
I think we can enjoy the hypothesis of a speech community of say AD 450 to 700, divided by race but living in the same settlements. In the model, the bilinguals are during this transition period the majority; in some districts they are a huge majority, say 90%. Only a few thousand Saxons have crossed the sea in their wooden boats. The lower status group is bilingual but the Angles and Saxons, the high status group, only speak their own language. The sociological gap between the two languages is sensitive and so there is no casual drift of vocabulary from one to the other. So, speaking P-Celtic was a sign of low status, and using Celtic words when speaking English was stigmatised. This would explain the amazingly low percentage of Celtic loanwords, while failing to explain the supposed Celtic influence on sentence structure. The model has to add a further ramification, namely that the speech community lacked the linguistic insight to recognise these patterns as Celtic, and so had no defence mechanism which could reject them. The speaker is less accurate about purging more abstract structures and so Celtic enters English, before the arrival of writing, in elements of syntax, which are the ones which the Three Finns identify as Celtic derived. This is unconscious and it does evade the censor.
The theory says further that Celtic habits continued to influence the development of English even centuries after the use of Celtic had died out. Thus, if the bilingual Celtic/English phase ended around 700 AD, the transition to modern English during the 14th C was still dominantly influenced by these 'Celtic speech habits', lurking especially among the lower classes. This argues for a kind of linguistic unconscious, where repressed memories of the past (pre-natal, in fact) continue to steer behaviour in the present. For them, you can inherit propensities to change the code as well as the code.
This allows the scholars to attribute every change in English to substrate influence even when the language contact had ceased centuries before. Although Celtic words are almost completely missing from the English vocabulary, this is to be compensated for by defining syntactic innovations, and so the very structure of the language and the majority of utterances, as Celtic influenced.
P-Celtic continued to be spoken in north-west England and south-west Scotland until perhaps the 12th century (hence Cumberland = 'Cymru', Welsh people.) But the expansion of English within Britain continued, so that even in 1200 AD there were border strips (in Devon, in Cumbria, perhaps in Strathclyde, and perhaps in Shropshire and Herefordshire) where P-Celtic speakers had recently acquired English. It is hard to see these areas as being linguistically or culturally influential.
Digging has found very few traces of Angles arriving in Northern England, so scholars are puzzled at the complete replacement of the Celtic language of sub-Roman north-east England by an Anglian dialect ('Northumbrian') in the period circa AD 500-800. It is agreed that the revolutionary simplification of word endings (morphology) which gave rise to Middle English began in the North, although this simplification is usually ascribed to the arrival of great numbers of Danes who learnt English easily but also simplified it where it differed from Danish. This is already pointed out in the Cambridge Hist of English and American literature (1907-21) on line. It says 'These changes naturally began where the evil was greatest, in the Northumbrian dialect' and makes a different point, that different classes of nouns were levelled so that the endings in -s were used throughout (to replace the lost ones in n). So they had -es for genitive and -as for the plural. This simplification is compatible with language mixing, with many people acquiring something as a second language. The rules which are typically acquired latest by children may be, collectively, rejected and not acquired at all. The Anglo-Scandinavian simplified language is known as ‘Angle-Mangle’. Signs of the change appear in 10th century manuscripts in Northumbrian (Gospel translations), later than the first large-scale Danish invasion (AD 866-7) but much earlier than the emergence of Middle English in the 12th century. This early dating is tantalising, even irritating, and unfixable. The new theory is that the code breakdown was four centuries older, and due to the presence of very many Celtic people in Northumbria, who learnt English inefficiently and so produced a creolised language which had eliminated the difficult morphology and had acquired some Celtic features. It was adopted by a later mixed Anglo-Danish population and spread over the entire country because it was easier to learn, this was its key design feature. (However, no pre-Danish texts have been produced to support this.)
The transition to Modern English was again revolutionary but bears very strong resemblances to the Old English/Middle English transition, in that it meant a further abandonment of morphology and the development of other features, such as word order, to express essential distinctions of meaning.
comments. The record of Northumbrian is strikingly thin on the ground and has internal problems which make my hair stand on end. I can see that classical language history did the detailed work while leaving the generalisations well alone, so that there is a vacuum, but you can't look at the evidence without thinking that the stuff you really want is missing. My impression (and I could not find any statements on this in print) is that the various theories of how Northumbrian could have come to take the forms it did, in the surviving manuscripts, all have fatal flaws, so that none of the specialists has any theory which claims to explain the facts. If that is not so, still I could not find any articles which did present an explanation.
The idea of a 'linguistic unconscious' is not just novel but completely unacceptable. (If I am not mistaken, Heinrich Wagner thought of such a thing in the history of Insular Celtic.)
Any feature which makes a language easier to learn for adults makes it easier to learn for children. The 'language efficiency' argument does not point us decisively at an ethnically different population group rising, submerging, underlying, etc. Languages have to survive being learnt by children.
Both Dutch and Danish underwent changes extremely similar to the transition between Old English and Middle English. We should bear in mind that the collapse of complex morphology is a feature of all Western European languages. In fact, the geographical boundary where this fact ceases to apply is of the highest interest. Baltic and Slavonic languages are very conservative on this point, (as is Finnish, I believe), although modern Greek has undertaken major simplifications. If this path has been followed by Italian and Danish, we must suspect that English has followed it for similar reasons and not, it follows, for reasons confined to this island, such as a Celtic stratum of more or less subdued churls in the area between Forth and Humber. It would be risky to develop a theory of English development which did not also account for the similarity to the other languages of Western and Southern Europe.
As for the amazing lack of Welsh/British words in early English, explanations of things that did not happen will always resemble descriptions of things that did not happen.
4. The question of substrates in Germanic
In 1995, FBJ Kuiper, a Dutch linguist mainly occupied with the history of Sanskrit, wrote an article in North West European Language Evolution (NOWELE) which has become a classic. ('Gothic bagms and Old Icelandic ylgr'.) It concerns, in detail, only two words (ylgr 'she-wolf' and bagms, 'tree') but its implications go much further and imply the existence of a lost substrate language with internal sound alternations which do not fit into the history of Germanic. FBJ Kuiper wrote what may be the classic article on this topic, where he identifies four substrate languages:
a1 Old European river names (as collected by Hans Krahe)
a2 the so-called language of geminates
a3 the “European” (or “Atlantic,” or “North Balkan”) substrate, characterized by the prefix *a (probably stressed) and the frequent occurrence of the vowel *a; a stop system in which labial and velar stops alternated
a4 Mediterranean not of use for Germanic
RSP Beekes’ 1999 lecture on the subject finds that of about 7600 root words in the Germanic component of Dutch, about half do not have Indo-European etymologies. The proposal is simply that these come from substrate languages and the project is to find any patterns in these words and to disengage something of the source languages and of the history of their contact with Germanic.
This project is associated with the ‘Leiden School’ and a new etymological dictionary of the Dutch language (edited by Dr Marlies Philippa, 2003-9) now gives full word histories of the substrate words, something which no European etymological dictionary has seen fit to do up till now.
The core of the search is to find patterns within many thousand individual words and a lot of the action has been to do with the geminates language, where the density and cohesion of the recovered material leads it away from orphan status and towards credibility as a fragment of a lost language. The scale of the loans leads us towards the conclusion that the Germanic peoples include a large proportion of people who were not Germanic and possibly not Indo-European.
Beekes says that the ‘Loanwords are often terms for local phenomena such as landscape features (various sorts of bodies of water: pools, puddles, runnels; hills, knolls, etc.) and terms for twigs, branches boughs, etc.; terms for ... tools; for household equipment such as pots, pans, forks, spoons; names for animals and plants when these were unknown to the newly arrived speakers’. The semantic cohesion of the group, already identified by phonetic considerations, is a powerful argument in favour of its identity as a genuine substrate block.
If you find doublets of words which have been borrowed twice, the alternations in the recorded forms allow you to hypothesize about the phonemic structure of the source language. The more regularity of patterning you find, the more the idea that the pattern exists is confirmed.
Take the following English words: chieftain captain chattel cattle. They all come from a French word meaning 'head'. The fact that the same alternation appears twice indicate that there was an alternation within the source language. We know in fact that Old French had a dialect distinction such that Norman French had a hard c and other regions had 'ch'. It is only when the four words are taken in aggregate that it becomes credible that all four come from the same language (in this case French). Their cohesion at semantic level is also a vital piece of evidence. (The idea that you would refer to cattle as ‘heads' and that this would become a collective term for them is not obvious from the word go.) It follows that the real identity of sets of words may only become apparent after considerable work, and it follows from this, further, that results from an earlier time may have to be abandoned. The evidence for substrates has to be considered as an entity, where repetitive sets of features have a weight in their entirety. This cumulative force is by now irresistible. The balance of power in Germanic language history has altered. Where we see alternations in the loan material which are impossible or unknown in the 'legitimate' stock, that is another indication that we are looking at another language. One feature of the loan material is seen in the stubb/stump group. These two words come from the same word in the source language but one has an m in it. This is the so-called prenasalisation. (It has to be stubb originally because otherwise the -u- would be long.) As a hypothesis, it allows the capture of numerous words which had been a source of frustration to earlier scholarship. The link between ‘sleep’ and ‘slumber’ is presumably attractive and easy to accept. Tracing a few dozen of these strong links allows us to be happier with less obvious and less direct links.
I have the very strong impression that language historians have generally regarded finding a good IE etymology for any word as success and failing to find one as failure. They have regarded themselves as ‘Indo-European’ scholars and not as general language historians. Recognition has an emotional value, a poetic taste of fulfilment. I think the IE share in the European languages which reached written record may have been over-estimated, perhaps quite seriously.
There is a second factor, to do with the development of laryngeal theory. This has led to a new idea of the nature of the IE root, so that it can only be CeC or eC. (C stands for ‘a consonant’.) That is, there was only one vowel in the parent language. This eliminates a large number of proposed IE roots which do not follow this pattern and so could not have been part of the IE language as reconstructed. It is based on the prevalence of ablaut in Indo-European, wherbey vowels alternate in a single root, as if they were a single vowel being modified by something unamed, a ‘thing which modifies’. This one vowel theory goes back to de Saussure in 1878; however the whole laryngeal theory was not available until the late 1920s, after the decipherment of Hittite with its clear evidence for the former existence of laryngeal consonants, identified quite soon as the things which modified. Its consequences have been fascinating to all but also far-reaching and so slow to work out. The standard dictionaries of etymology are not in line with this ‘one vowel’ theory. Szemerenyi has suggested that the idea of a language with only one vowel is contradicted by the evidence of the world’s languages as they have been captured by 20th C linguists. It is so marginal in the distribution field as to be subject to very severe doubts as a postulate. Thus the reconstruction with a single vowel may be wrong. There is a link between the development of laryngeal theory in Indo-European and a revived interest in non-IE word histories and phonological shapes. For example, Beekes wrote a book on laryngeals in Greek. The 'new' substrate words are partly new inhabitants of the gap made by wiping out all IE roots that do not have the vowel e.
Beekes remarks that the first wave (actually the first century or 150 years) of IE studies was so occupied with phonology that they neglected semantics; that is, many accepted derivations are wrong because the meaning of words was not considered and the acoustic resemblance is a coincidence. IE scholars have shown ingenuity and collegiate solidarity rather than pure objectivity. If we find an etymology of the word ‘ship’ which links it to a root meaning ‘cut’, this may be ingenuity rather than a good etymology. Why would the ‘cut’ word give you the name for something which is made of wood, floats, goes on the water, brings fish, etc.? Is it not better to give up the ‘noble genealogy’ and accept that this is a substrate word acquired on arrival on the shores of the Baltic? The root is presumably the one seen in Scheibe, German for ‘a slice’ of bread etc. Do we think of a ship as a floating slice?
Beekes suggests that the rule whereby a vowel is short before a double consonant but short before a single one is a feature of the substrate language. This is related to the observation that many words suspected of being substratic include double, i.e. geminated, consonants, which (according to orthodox etymological theory) do not exist in proto-Germanic. It is important to note that geminates can re-emerge in the later history of the Germanic languages and that they remained in the demarcated territory of pet-names ('hypocoristics'), so that Uffa is a by-form of Uhtferth. It is known from 20th C observations that reduplication is a feature of child language, and evidently the pet-names belong with child language even though they can continue to be used in adult language. Thus many names in mediaeval documents that list the servile population, those who owe labour obligations and rent, are recorded in pet-name form, while this is much rarer for kings and bishops. It is interesting that King Offa is recorded in this form rather than as *Ohtferth. (Actually there were two King Offas.) The geminate words are not personal names.
Another proposed substrate feature is the p/k alternation (or labial-velar alternation). This has a history, as Kluge had noticed, already in 1913, that there was a problem with a group of words which he had isolated, all too clearly. If we look at the German words streifen and streichen, it seems likely that they are a doublet, deriving from a single word in some speech universe where p could alternate with k (these being the earlier versions of the f and ch). (English strip, strike.) If we look at the English word slack, its exact semantic correspondence in Low German is schlapp (High German schlaff). Again p and k alternate - the same word appears with both variants. Another example is in stronk and stromp (Dutch words both meaning 'stump‘). English words scrape and scratch (original consonant a ‘k’) are another example. The link between English creep and German kriechen is another. (The Nordwestblock word for 'daybreak' alternates, thus 'krieken' in Dutch and 'creep of day' in older Scots.)
(Another way of looking at this is to say that these are the only alternations recoverable - a huge range of other possible alternations are not found in any identified doublets.)
Another series shows an alternation between k and kn (connected to a theoretical kn-gn alternation where one side evolved through hn to simple n). Thus Dutch knauwen matches German nagen 'gnaw'. There is an internal doublet in Dutch of knauwen and knagen - which shows a velar-labial alternation.
As Gysseling is clear that the ‘Belgisch’ language was Indo-European, it is not the ‘geminate language’. To confirm this, Beekes positions the latter in ‘the west and north of central Europe’, an area which does not include any district within the Nordwestblock. Evidently there is a paranoia of substrates in which someone loses restraint and makes all substrates converge. Nonetheless the cover of the published version of his lecture reproduces an altar to the goddess Nehalennia, because of her non-Germanic name. Nehalennia is however one of the names which are identified as belonging to the Nordwestblock.
Archaeologists resist any idea of migration and strive to justify ‘local development’ of changes even in the most unpromising cases. Linguists strive to find an Indo-European etymology for every item of lexicon and regard a local origin (substrate) as a failure. The contrast is very bizarre when after all it was the same people who used both artefacts and words. It cannot be founded in a sound philosophy of science. It must connect to the traditions of the professions - transmitted by unconscious imitation and thriving in the form of anxiety.
A Boutkan essay attributes slakot' (Ru.) to a North European substrate and links it to Swedish slage, which is swamp (cf. slack in Northern England). He lists other Low German words about wet mud, splashes, etc. He says schlagregen may not be 'schlag' but 'rain that makes you wet’ as slakot' is 'snow that makes you wet'. schlackern. There is a wonderful array of Scots words from this root unknown to Boutkan, with and without the initial s. One is slaicher. And the word 'slake' I guess. This is a whole essay on one word but it does point to a whole world of possibilities. This is the sludge-slush root. The original North Europeans were world experts on mud.
5. Modifications to the Renfrew thesis
Does the distribution of features in which languages deviate from the pristine Indo-European construction record a geographical distinction in the balance between inward migration and Mesolithic population continuity as the first farmers spread across Europe?
summary. The front runner of theories about early language history in Europe is presented in Colin Renfrew’s Language and History (1984) which equates the spread of the IE languages across Europe with the spread of farming. Thus if we find related languages in all parts of Europe, resembling each other evidently because they spread from the same point (or area), this matches with the spread of farming to all points of the continent, also from a fixed area (or perhaps from two areas, via Greece and Spain, respectively). In fact, where we find non-farming models of subsistence continuing, for climatic reasons, we also find non-IE languages, for example in Scandinavia and northern provinces of Russia. It is obvious that almost all IE peoples in early times are also farming peoples. In Europe, the eastern languages are archaic and the Western ones highly innovatory or broken up. With respect to the reconstructed ancestral language, Slavic is highly archaic and Baltic even more so. There are striking resemblances between the changes away from the source language, with its amazing wealth of verb and noun endings, followed by Insular Celtic and Germanic before the arrival of writing, and the development of the Romance languages in their journey away from Latin.
Renfrew has been extremely active in this area of theory since 1984, so that there are now many Renfrew essays which substantially modify the model in his classic 1984 book. He has now specified that he sees the earliest Indo-Europeans as living in the Konya Plain, in south-central Anatolia, during the Mesolithic, and acquiring knowledge of farming there. He proposes (this is before the evidence can answer us) that they were the late Palaeolithic population of that region, so that if we could go back to 8000 BC we would find a tribe (living by collecting food, not cultivating it, but already sedentary) there who were already speaking a proto-Indo-European. (The famous urban site of Catal Hoyuk or Huyuk, dug by Mellaart and Hodder, is on the Konya plain.)
5a differences in IE languages reflecting the boundary between 'demic
diffusion' and 'contact imitation' in the Neolithic
summary. In a later essay, Renfrew retracts the idea that all of Europe was populated by migrants from Anatolia, the primary farmers, spreading at very much the rate that farming spread. Instead he now draws a contrast between demic diffusion vs contact imitation 'availability'. The terms are drawn from the work of population geneticists and the adaptation reflects the findings of these geneticists (Ammerman and Cavalli-Sforza).
Demic means 'Neolithic' gene flow into Europe and imitation means Mesolithic genetic continuity - Mesolithic populations imitating farming practices and holding on to their territory to give rise to modern Europeans. The realm of imitation is essentially Europe outside the Balkan 'Old Europe' core, which possibly had outliers in Ukraine and South Italy. ‘Demic’ is an adjective for the word ‘population’ -‘demic diffusion’ means diffusion (of farming essentially) by a flow of people who knew how to do it. There is another possibility, which is of non-farming populations observing farmers on their borders and acquiring the new techniques and food species without romantically disappearing from history.
Does the theory (that the Neolithic immigrants to Europe were speakers of Indo-European) hold up if the area north of a certain line is the zone of 'contact diffusion' (as opposed to immigration)? But Europe North and West of that line is just as Indo-European as the Mediterranean, so can we explain this by saying that Celtic and Germanic are 'broken European' and radically reshaped by substrate influence? Can the Indo-Europeanisation of NW Europe be described to any extent as creolisation? What is the linguistic correlate for the alleged demic/imitation split?
In these later essays Renfrew describes monothematic/ polythematic states of Indo-European, Hittite (Anatolian?) being monothematic. (see another essay on this site for details) He suggests that North languages separated first and there was a Greek -Sanskrit phase of shared innovations after that separation. The intense language contact which produced the polythematic type took place in the Balkans and regions now in the Ukraine; Hittite, Germanic, and Celtic were excluded from them because they are not the product of that intensely creative phase, which fundamentally changed the nature of 'Indo-European'.
comment. The Mesolithic continuity idea is inevitably tempered by the demonstrated IE nature of the languages of Northern and Western Europe, so that by this account the Mesolithics must have imitated their richer neighbours enough to take their language over. This has hardly begun to be explored, but in this account the original features of Celtic and Germanic vis-à-vis Greek, Slavic, etc. go back to the original acculturation and contact phase and to a stratum of the European past which is far beyond the earliest horizon of written history.
There is however another view of the ‘northern peripheral languages’, which is that their unusual features are due to the way in which they were acquired by pre-existing populations, and in fact to simplifications akin to creoles.
5b We now have a new question, roughly: do the distinctive features of Celtic and Germanic, in deviating from classical IE, represent (a) the effect of language mixing with a Mesolithic population outside the core area of early farming or (b) archaic IE which ‘missed out on’ the advergent language development taking place in the south-east, as the core area of early European farming and technological change? I cannot think of any way of testing this.
Has there been a traumatic suppression of awareness about the degree of decay/ barbarism of the Germanic group, connected with the nationalist commitment of the German scholars involved? And an indifference to the Celtic group? No one doubts that Celtic is a highly atypical IE language and the theory that its original features represent the adaptations of a Mesolithic Western population learning a foreign tongue is attractive and at least gets us off the hook of trying to explain it any other way. The idea that Germanic is the other half of the same category is unattractive to many people but has some intellectual history behind it.
comment on the creole theory. Sigmund Feist started the 'Germanic - creolised' hare in 1932. The term creolised derives from certain languages mainly of Negro and mixed populations in the New World, which radically simplified the structures of (conditionally) 'parental' languages because of the way in which they were acquired. The 'parental' languages included Spanish, Portuguese, English and Dutch, but the 'child' languages had compelling resemblances to each other. This was an ‘unrespectable’ zone of language and scholars saw it as beneath their dignity. The great Austrian linguist Hugo Schuchhardt reached, towards 1900, a fundamental theoretical understanding of Creole and so provided a theoretical basis, not existing until that time, for understanding how a language can be simplified. Scholars were not attracted by the human subjects (Black people, mostly) and were very slow to see in creoles a source of ideas on radical language change.
Feist explained the changes, occurring before the written record, which gave rise to the Germanic family as we find it in that record, as being due to a creolisation in which the speakers of a non-IE language learnt IE and radically changed it in order to make it easier. Antoine Meillet did something similar in 1922, although not using the Creole label. I think the resistance of German scholars to the idea that German is a ‘post-creole’ language is not altogether intellectual and rational. Having said that, Feist’s theory may be wrong. Supposing it to be right, it may really connect with the distinction between Core/Old Europe and periphery/Outer Europe/ contact borrowing.
Other scholars have discussed the resemblances between Hittite and Germanic. The overall shape of the Gmc tense system is similar to that in Hittite and so consistent with being archaic-faithful and so not creolised.
6. Vennemann's theory on two prehistoric non-IE languages affecting NW Europe in later prehistory. Was Western Europe Basque-speaking in early times (and possibly in the Palaeolithic?)? was there a Semitic or Hamitic language which was spoken on the coastline of Western Europe in the last millennia before the birth of Christ?
summary. Theo Vennemann's theory is that there are two key substrate languages in Western Europe, one Basque, present from earliest times, and one 'Semitidic' (by which he means a language which is like a Semitic language but is perhaps only a cousin of them), present in 600-300 BC. There are toponyms from these languages all over Western Europe. The Indo-Europeans were thus preceded in the far West by two main non-IE populations.
His views are expressed in a huge volume, Europa vasconica - Europa Semitica, 977 pp. (These are the two peoples he describes.) This is TV's collected articles 1984-2000 so not a connected argument. It is in German. He is also, with Elisabeth Hampel, one of the two authors of an article saying that 'we're all Basques'. He claims that Raetic (the ancient language of the Austrian Alps, roughly) is related to Basque, and claims Phoenician as a superstrate in Baltic Europe. He formerly thought it was an anonymous Semitic language but changed the theory to be Punic. He also finds 'Atlantic' people on the littoral: they are NOT Basques, but supply toponyms on the coast, moving by ship and equated with a 'semitidic' language (i.e. like Hamitic but more like Semitic). He links the latter language with the Carthaginians and says it is maritime and arrives much later in history than Basque. Names like Solent, Solund, Tay, Tawe, Uist, Scilly, come from the language of this group. (Another paper says that there was an ‘Afroasiatic’ language in W Europe from the 5th or 4th millennium BC. Afroasiatic is a term for the vast language group which includes Berber in North Africa, Ancient Egyptian, and the Semitic languages.)
A longer summary is this, from a review by P. Baldi, and B. Richard Page: (where V means Vennemann)
"After the last ice age, which ended about 11,000 years ago, Indo-European agriculturists, possibly originating in the Pannonian Basin of central Europe, migrated further into Europe in the sixth millennium BCE, arriving in Scandinavia beginning around the fourth millennium BCE. The migrating Indo-Europeans encountered other, non-IE people, who had started to settle there already in the eighth millennium BCE, i.e. several millennia after the last ice age, and had already named the European rivers, lakes, mountains and settlements. Thus the oldest water names are probably the oldest ‘‘linguistic documents’’ in Europe north of the Alps. The structure of these names betrays an agglutinating language with initial accent, no vowel quantity and a predominant vowel a. The language family responsible for these names is called by V ‘‘Vasconic’’, whose only surviving descendant is the Basque language of the Pyrenees. Additionally, there are toponyms on the Atlantic littoral which are neither Vasconic nor Indo-European. The prehistoric language responsible for these names (and other linguistic effects) is called by V the ‘‘Semitidic’’ (also ‘‘Atlantic’’), group of languages, i.e. languages related to the Mediterranean Hamito-Semitic languages, which were spoken along the European Atlantic seaboard from the fifth millennium BCE until the first millennium CE. These languages are held to have influenced the Indo-European languages of the northwest littoral from the fifth millennium BCE onward."
The Indo-Europeans were the second group to move in but won because they had agriculture - the Vasks were cattle breeders and apple growers.
Vennemann says that the etymon of vecher/feasgar/ vesper/ucher (words for evening in languages from Russian to Welsh) is EUSKARA the Basque name for Basques, viz. West people. West=evening and the Basques (once) occupied the West of the continent.
comments. I find this highly unconvincing. Perusing the review and the abstracts makes the whole thing seem vacuous. I can't see any valid etymologies, not one. So it was a theory which after collecting evidence has collapsed. There are thousands of words from the 'apple/plough/ pig' substrate and none of them are Semitic. If they were someone else would have noticed. There are two key papers by specialists (one by Terry Langedoen, on-line in American Linguist) which demolish Vennemann's claims about the 'source languages' in which he is not a specialist. The individual etymologies, the ‘engine room’ of the theory, don't hold up. The question is not whether he is right, since he may well be right, but whether there is any recoverable evidence which speaks in his favour. I do not see any. Possible links (like Uibhist = Ibiza) are compatible with sailor-names fixed by maritime passers-by whose affinities say nothing about the language of the major part of the land population.
Beekes says that this theory has scarcely been investigated yet and so that it may develop into something.
The idea that Phoenicians had an influential role in the early history of Northwestern Europe goes back to Samuel Bochart in 1646 and Aylett Sammes ‘whose enchanting but bafflingly muddle-headed Britannia Antiqua Illustrata was first published in 1676' (Kendrick). Given that the Phoenicians got as far as Spain, and were expert at sailing, the speculation that they went up the Atlantic littoral was bound to be made by many antiquaries over the centuries. However, the concrete product of these enjoyable speculations is nil. Limited coastal contacts would not always have left traces that could be picked up 3000 years later.
This theory overlaps a great deal with the Morris-Jones/ Pokorny theory (below).
7. In 1900, John Morris-Jones proposed that a Hamitic substrate (from a language resembling the Berber languages and Ancient Egyptian) influenced Insular Celtic and accounted for its differences from classical Indo-European (and Continental Celtic). This theory has been around for a century. Can we sustain it, has any new evidence been developed?
“Following the hypothesis initially proposed by J. Morris-Jones (1900), several scholars have assumed that typologically unusual features found in Insular Celtic are borrowed from some unknown substratum, presumably belonging to the Afro-Asiatic family, where such features have also been attested. This hypothesis found some adherents in the following decades, such as Julius Pokorny, Heinrich Wagner and Orin Gensler, but it has been vigorously and convincingly
criticized in recent works by G. Isaac and K. McCone.” (Matasovic) By Insular Celtic we mean languages deriving from the British isles, such as Gaelic, Welsh, and Breton, as distinct from rather shadowy languages known as ‘Continental Celtic’. The Insular group has a number of very distinctive features which have not yet been found at all in the scraps of Continental Celtic.
comments. The 1900 theory has recently been developed and added to. Another suggestion (by Stephen Hewitt I think) is that there may have been a language C: massively present in NW Europe and influencing Celtic, and less present in N Africa and influencing Berber as a substrate. This illuminates the space of possibilities and is attractive.
I found on the Internet Graham Isaac's very convincing paper attacking the whole Morris-Jones line. However there is considerable recent activity on the other side and the outcome is far from clear. The lack of evidence is a problem - if we had a good picture of linguistic conditions in Spain in early times (say 1000 BC to 800 AD) then the truth would emerge one way or the other. As it is the recording of the Berber languages, for example, is so late in history that getting back to Bronze Age conditions calls for an activity close to delusion. There are early Berber texts, perhaps of the 13th C AD although I could find few details, but this is just too late. I am sceptical that we know what the Hamitic languages of the Atlantic Coast were like in 2000 BC. I doubt then that we can connect them to a substrate language in the British Isles.
Morris-Jones was, we can say, the first scholar to direct German linguistic method to the history of Welsh, and as a consequence had immense prestige in Wales throughout the 20th C. He also wrote prize-winning poetry in a rigid, classicising style. His definition of Welsh linguistic purity was so powerful that it was intimidating. When TJ Morgan wrote that writing Welsh was like writing ‘under the eyes of the Gestapo’ he was referring to the purism of Morris-Jones and his followers. (This phrase was the title of a recent book by Simon Brooks about the history of literary criticism in Wales.)
"One may of course hypothesise anything. But then, in the case of hypothesising the existence of a language in Dark-Age Britain and Ireland, that language must be presented." (Graham Isaac, Galway) Isaac's intent in this paper was to disprove the theory of a 'Hamitic' substrate influencing Insular Celtic. I can’t find any decisive evidence either way but in that absence denying any connection between the Irish Sea Province and Morocco-Algeria is the neutral or minimal position.
8. The Nordwestblock. Was there an area (covering the Low Countries and northwest Germany) which in late prehistoric times was 'non Celtic non Germanic' and which spoke a different language ('Belgisch')? How far is this substrate represented in English? was this the language spoken by the Belgae who crossed the Channel to settle England?
This is the thesis of the book Völker Zwischen Kelten und Germanen, by Rolf Hachmann, Georg Kossack, and Hans Kuhn (1962), which approached the issue from linguistic, archaeological, and historical angles. They used the word ‘Nordwestblock’ (NWB). This separate people was submerged by the wars of resistance which were a reaction to major Roman invasions of 'Germany' in the first century AD, and the culture of the area was Germanic from that time on. The initial conquest may have involved only an ‘exchange of elites’ so that the mass of the population went on speaking the old language for a few generations.
Caesar, in the mid-1st century BC, divides Gaul between the Aquitani, Gauls and Belgae, and we may ascribe these groups to the Basque, Celtic and Nordwestblock peoples.
Kuhn developed the thesis greatly in later papers and found elements of 'NWB language' also further up the Rhine, for example around Trier.
Gysseling probably wrote an article identifying a non-Germanic non-Celtic language in the Netherlands already in 1952 (not available to me). It must be the same language; Kuhn added evidence for a wider regional extent, on the coast between Holland and Denmark and further up the Rhine. I am assuming that Gysseling and Kuhn are describing the same theory, although they use different terminology.
Up until the arrival of writing, especially writing which survives until today, it is hard to allocate peoples to any linguistic group. The NWB theory relies on a belief that the limited number of non-classical languages which survived long enough to be written down roughly 600-1000 AD were not the complete set which had been thriving up till say 0 BC, and that the dynamics of the violence during the Volkerwanderung had allowed strong groups to expand but also caused weak ones to go under. So rather than define everything in NW Europe as tidily Celtic or Germanic you look for evidence on the linguistic affinities of any tribe or region. The result is that a whole area on the Lower Rhine and the North Sea coast emerges as (nonCeltic nonGermanic) and needs a new name. The distinction between Gmc and NonGmc is only of great interest if you subscribe to an ethnocentric view of later prehistory. The idea that all the historical agency of population groups depends on their ethnicity sounds distinctly 19th century and nationalistic. This idea was in part a reaction to the shortage of information in the relics of the past, of the Iron Age and early historical period. A sort of operatic way of making dumb artefacts sing. For most students, the domain of the Gmc languages is not a unity and they would not suppose that customs, laws, etc. obtain over the whole domain, without further proof. The unit of customs was presumably the tribe, of which there were hundreds; inscriptions and narrative texts very carefully record the names of tribes. The postulate of a 'third language group' is of interest to etymologists and there may in fact be an appreciable amount of lexical material for which an 'Indo-European' derivation is likely but which has unclear relations with Celtic and Germanic in their classical conceptions.
The NWB theory is as robust as it can be in the absence of any written text in this profoundly missing language. As Kuhn identifies Belgish as Indo-European it cannot be the substrate language we are looking for and it has nothing new to offer on the ‘demic diffusion’ theory in article 5.
Caesar decribes large scale invasions of eastern England by the Belgae, who were there when he landed on the island. If we follow Gysseling then the speech of these Belgae may have included Belgic, Nordwestblock, elements, which were non-Celtic. These may have included both Germanic and non-Germanic elements. If the Anglo-Saxons spent a century or two in the South Netherlands, they lived in an area teeming with Belgic words (possibly even population groups) and may have picked up a wealth of such vocabulary there before crossing the Channel. That is two 'ports of entry' and the evidence should be examined (Gysseling was not interested in languages in Britain).
I surmise that the place to look for older Belgic words is in Welsh. The search for Belgic words is on. If we accept Dutch knauwen/ knagen as a Belgic root then Welsh cnoi 'chew' is presumably from the same root. (but cf. Irish cognaim?)
Estimates of the population of the Netherlands in prehistory do not point to a very large number of speakers of this lost language (or group of languages). We are looking at a population which combined pastoralism with fishing and seafaring, with a minimal input of field crops. The nature of the terrain in the NWB would seem to point against the presence of a rich agricultural people capable of transmitting so many important words to the uncouth Germani. Even if we double the numbers to allow for a much larger area being included in the NWB than just the Netherlands, the total does not sound very impressive.
The topic opens onto the idea that the boundaries of linguistic groups may not be the significant boundaries in the ethnographic history of Europe, and that real 'cultural regions' may be smaller and more persistent than the whole areas covered by languages and states. Emmanuel Todd has given gripping examples of this in his analysis of European family structure as the basis for political cultures, finding in particular that different parts of France have quite different family structures in modern times. He sites these in cohesive 'culture provinces' which would therefore be the real units of history, with States as small, elite, conspiratorial blocks travelling above the population as a whole, whose history was unrecovered until recent decades. Thus, the Low Countries are evidently distinct from the neighbouring regions without this identity being reflected by major linguistic differences. Moreover, French-speaking Belgians are like other Netherlanders and not like the French. All this can be connected to a continuity with an much earlier state of affairs: where the Netherlands were ethnically distinct from the regions around them. Naturally the borders of the 'Lower Rhine province' have shifted over the centuries.
The 'Belgic' group produced the feminine -stjo- ending, so that 'spinster' and 'baxter' are feminines. And the word 'brook'. A search for further traces in England would be of great interest. It is quite unclear to me whether this NWB language was also spoken across the sea, here in England. I am not aware of any scan of the Welsh vocabulary.
An interesting example of a Belgic word analysed by Gysseling is pinky, Dutch pink, the little finger. Gysseling says that this word means ‘five’, as the thumb is the first digit. The word is only known in the Netherlands (the American word is a loan from Dutch settlers) and in Scotland. It bears a very satisfying resemblance to the agreed IE word for five, ‘penkwe’. ‘Pink’ compares with Greek pente and Slavic pyat. It should mean ‘five’. It is not Celtic (Welsh ‘pump’, Irish ‘coich’). It must be IE and so is further evidence that Belgish was an Indo-European language.
9. The Minoan-Hattite-Celtic thesis
I was searching for stuff from the Leiden Substrate Group and came across this:
- an inaugural from the professor of Celtic languages at Leiden. He writes about a Celtic substrate in the formation of Dutch/ Old Saxon/ Northern French and it is all going swimmingly. He talks about north and south Celtic and the complex verb in Irish. Then he starts to explain that the complex verb is a sign of relationship to Minoan and Hattite and that the original farmers were Minoan-Hattites. Even though we can't actually read Minoan texts. Schrijver says that the compound verb of Insular Celtic reveals the affinity of that language with two languages of the second millennium BC, Minoan and also Hattite, a non-IE language recorded in the archives at Boghazkoy. He connects this also with the North Caucasus. I have not examined the question, but since we cannot read either Minoan or Hattite it is far from certain that the mentioned words are verbs or that their form shows a multiple compound construction.
Schrijver says that the Celtic languages are the spider at the centre of the web of early European linguistic relations.
comment. I think comment is unnecessary. I will merely wave towards a spider in the corner.
A note on the move away from word endings as bearers of meaning
It seems deeply unreasonable to declare that the processes affecting French, German, Spanish, Welsh and English (less morphology, more auxiliaries, roughly) are completely independent phenomena. Clearly the languages sharing this process are geographically contiguous and the ones impervious to it are outside that area, also forming a grouping. It is possible that in some ways Gaelic, the most peripheral of these languages, has gone furthest along the road. It seems equally irrational to state that the processes leading to the origin of Germanic, with its radical loss of noun cases and so on, were really unrelated to the process which led to the shift from Middle English to modern English, in the 14th century - and more than 2000 years later. This was a change which continued along its own axis - outdistancing itself, you could say.
The move of European languages (mainly Western European?) from declension morphology to the use of auxiliaries and word order (and restricted morphology) is a problem which linguists may have no defensible and encompassing answer to.
If it was already well under way in 1000 BC, why was it still advancing quite rapidly in the 18th C AD?
If it had such an impetus behind it, what was resisting that impetus for most of those 3000 years? Why was the war unwinnable? If morphology was so strong, why did it lose its empire -and where did the anti-morphology impulse or principle come from?
If it was 'modernisation' in Sweden or England, why did it not occur at all in Russia or Lithuania (or in India)?
The change has limits in space as well as time. It has spread from the archaic West to Italy and Greece. This distribution has led to its being associated with the Roman Empire, but it has made progress in areas where the Empire never reached or among peoples (the English) who were never subject to the Empire. It is as conspicuous in Rumanian as in French.
Surely we are not going not say that speaking Russian is awkward and difficult? Russians do it quite fluently.
When something is insatiable, does that mean it is also frustrating? Not
fulfilling itself? Or just that this is Progress?
It is normal to learn surviving old languages and to read the texts written in them, for example the Mabinogion, Beowulf, Heimskringla, or the Tain. One cannot go beyond without accepting certain conditions and becoming in some way like a phantom, living on air. It is a mirage to think that you can know a language without learning it.
The recovery of languages which do not have any considerable texts is an experience of poverty, abnormal in comparison to the study of recoverable texts and languages. Where evidence is incomplete, speculation is a commendable act. The lit space of written record is in Europe surrounded on all sides by the darkness of the unrecorded and the penumbra of what is fitfully splashed by flickers from the lit zone. Those who choose to go out into the darkness with its fragments of light do so presumably because they enjoy the speculation - something close to emptiness. Speculation is closely related to ignorance but can be the stepfather of knowledge. Before reaching written form, speculation itself must be subjected to the tests of reason, including comparison with all available sets of real evidence.
Reading a text involves having committed say 5000 words to memory (counts vary, it may be more) and being involved with lost languages may appeal partly because that phase of memorisation can be skipped. This however gives linguistics with the language taken out. The introduction of wrong hypotheses to their mortality clears the air and lets the truth peep out. The lit zone of language history is surrounded by phantoms claiming to be real parts of history. Some are not even phantoms but stray noises making themselves out to be phantoms.
The supplier of this blog changed her software and as a reuslt odler posts, like this one, lost their formatting. I apologise for the interval during which this post was an undifferentiated flow.