RETURN TO CONTENTS

We have made a complete study of the information on the Indo European peoples that can be derived from the languages. As part of this study we have estimated the date at which the Indo European peoples began to disperse and the subsequent rate of change of Indo European languages. This study is summarized elsewhere in this site under the title "Proro Indo European Dating.

Here we summarize another aspect of the overall study, namely how language changes can be used to investigate the manner in which the European language groups separated from each other and when these separations occurred. The results are consistent and show a striking pattern of change that is consistent with current archaeological and historical evidence. No previous comprehensive study of this type has, we believe, been performed.

Our methodology for the study of the entry of Indo European peoples into Europe is similar to that adopted in our parallel study of Indo European Dating and this document should be consulted for the details. Here we will only summarize the methodology and the results.

We have shown that we may compare the survival of the Proto Indo European (PIE) derived by linguists to determine the date of the first dispersal of the Indo European peoples and estimate the rate of language change. The "Dispersal Date" we calculated is 4700 BC with an uncertainty of 400 years. The "Half-life" of the rate of change is calculated to be 2240 years with an uncertainty of 140 years. Both of these results differ significantly from other estimates of the date of Indo European origins and how rapidly these languages change. We argue that our unique study is proof of the principle of the validity of "glottochronology" and that, since our study is much more comprehensive than previous work and statistically more accurate, the results are superior to those of previously available. However, we point out that the entire process of glottochronology is empirical and caution that these methods may not apply to other language groups.

Written historical evidence of the activities of the Celtic "barbarians" who troubled the Greeks and other peoples and later the Romans, point to their prior presence in north-western Europe, in the region around the southern part of the Baltic Sea. And a little later the Germanic tribes moved south towards the lands north of the Black Sea as well as into the Roman Empire.

The origin of the peoples who occupied the lands around the Adriatic - Italy and the Balkans – is unclear but there is some evidence that the languages there have similarities to the two branches of Celtic and so they possibly had the same general origin. And the (later) movements of the Slavs were probably from the southern part of the Baltic Sea. The historical presence of the Balts in the same region is also good evidence of the significance of this region.

In conclusion, it seems plausible or even probable that all the European groups - Romance, Celtic, Germanic, Slavic, and Baltic – may well have once existed as a single, undifferentiated, Indo European group located around the lands at the south of the Baltic Sea. And in our study we look at the glottochronological evidence for the separation of every one of these five European language groups from each other.

In this study we use modern vocabularies of European languages and classical Latin to calculate when the language groups separated from each other. The circumstances are several thousands of years separated from the initial Proto European dispersal and the PIE vocabulary is not relevant to this study. In addition, the procedure is a little different when comparing contemporary vocabularies.

We will first discuss the appropriate half-life.

In the dispersal study we had an original, re-constructed, vocabulary and could look at the survival of these words in later, dated languages. This gave us a half-life for change of 4700 years. But now we will compare differences in the vocabularies of pairs of pairs of languages to find out how both have changed from a prior vocabulary. The mathematics involved shows that the half-life must be used in a slightly different way; since both later languages are changing from a single earlier one they are, in effect, moving away from each other at twice the rate they diverge from the original. The relative changes are twice as fast and it is as though the half-life is 4700/2 or 2350 years and we must expect to find fewer matches between the language pair than in the other circumstance.

There are five European language groups to consider, Germanic, Slavic, Baltic, Celtic and Latin. The number of currently spoken languages varies from group to group from seven or eight in Germanic and Slavic to two in Baltic; the numbers could be increased by including some recently extinct groups such as the Baltic Old Prussian. But the number of different languages is large so there is no lack of information to analyze. However, the large number of languages also means a large number of possible comparisons so that it also raises the difficulty of choosing what comparisons to make. Should we, for example, compare every language with every other? And, if we do this, how do we avoid possible confusion?

We have chosen a simple approach, which avoids ambiguity in cross comparisons and reduces the effort to a practicable level. For each pair of language groups, we compare one language in the first group with one in the second and do not use either language again in this particular comparison. For example, in comparing Germanic and Slavic we compare German and Croatian, Danish and Czech, Swedish and Polish, and Norwegian and Russian, four comparisons in all. We have also not tried to include every language simply because of the amount of work involved. We trust that we have included enough cases to give a representative answer for group-to-group comparison and the consistency we see gives us some confidence that this is so.

In the case of Latin we have adopted a slightly different procedure. While Italian, French, and others are clearly in the same language group, there is certainly some admixture of other groups, and so we have chosen to simply compare classical Latin with the languages of the other groups. This means that the Latin vocabulary is not modern and so we reduce the number of Latin comparisons to allow for the additional change to Latin to that which would be expected if its vocabulary had survived to 1950 AD.

Our main criterion in the choice of vocabulary is similar to that in selecting a PIE list in the sense we must use names of objects etc. that are likely to be used in the past and in the present. However, we see no particular virtue in building upon the actual PIE list itself, and, to try to be objective, we have adopted the following procedure.

We have taken a modern dictionary in a language with which we are unfamiliar, specifically English-Lithuanian, and simply scanned it alphabetically to see what appropriate names it contains. And we have restricted our selection to names of familiar objects. This produced a list of 124 names that ranged from "adze" and "amber" to "wheel" and "yoke". We believe that this list is non-controversial and that a similar list would result from looking at any other modern dictionary.

We trust that the number of words is large enough to give adequate statistical accuracy.

We have made 36 separate calculations of the common words in pairs of languages; these calculations are then collected to show the same common data for the language groups. There are ten such group-to-group comparisons.

We shall illustrate the results by one example, Germanic and Baltic.

Swedish/Lithuanian 17/124

German/Latvian 19/123

Danish/Old Prussian 7/68

Taking all of the results we find the percentage of group-to-group survivors varies. The percentage varies from a high value of about 13% to 14% to a low of about 3%. These results show (as expected) that the languages separated from each other at different times, the oldest time of separation corresponding to the lowest percentage and the most recent to the largest. And, using the half-life we calculated for change of Proto Indo European, we calculate the times at which the separations occurred.

We find that the ten calculated times of separation fall into four different groups, that is, they occurred at four different times, namely 3625± 190 BC, 2559± 147 BC, 1914± 185 BC, and 1275± 176 BC. These data are shown in the following table.

This series of dates can be used to construct a consistent and plausible pattern of separations, but it is not simple since it involves a "stretching" of language group separations prior to complete separation. The figure below shows this pattern. Here the languages within a given ellipse are assumed to be mutually intelligible and those outside this boundary are not. For simplicity, we do not include the uncertainty in the dates in this discussion but the reader should keep them in mind.

Initially, up to some unknown date, we assume that the five groups have not separated at all. That is, all the languages are mutually intelligible.

The first movement towards separation, in 3625 BC, is one where Latin, Celtic, and Germanic are mutually intelligible and the same is true for Slavic, Baltic, and Germanic. But Latin and Celtic have diverged sufficiently that the speakers of these tongues can no longer communicate with Slavic and Baltic. We assume that these changes were the results of drifting of peoples within the area they occupy. We might also guess that the common tongue, Germanic, would presumably have regional distinctions, dialects, depending upon where the speakers were adjacent to the other groups.

The second separation, about 1000 years later in 2559 BC, shows the complete separation of Latin from the other groups, perhaps the beginning of a migration south towards the Adriatic. The other groups presumably remained in place since there is no evidence that their relationships change at this time.

The third separation shows the movement of the Celts, presumably towards the South. We also see the partial separation of the Slavs so that they can no longer communicate with the Germanic peoples but are still intelligible to the Balts. The Celtic movement is consistent with what we know of the locations of the Celts in historical times, and that of the Slavs is not inconsistent with our (limited) knowledge of their history.

The final, fourth, separation, in 1275 BC, shows the separation of Germanic and Baltic. The Germans, at least in part, probably moved south from the Baltic region since we know from history that the Germans appeared on the northern borders of the Roman Empire. But not all the Germans moved away from the Baltic since the Swedes are found there later, and this separation may have been the result of the movement of the Balts.

We believe that this study demonstrates the usefulness of glottochronology under favorable circumstances. It is possible that there is sufficient information available to apply our methods in other areas, for example, to the separation of Indian languages.