Language Families and the Tower of Babel

Glottolog counts 421 families and isolates with no demonstrated links, as Babel predicts but also as the comparative method's horizon predicts on any chronology. Text-dated Indo-European and Sino-Tibetan are 6,000 to 8,000 years old. Under a Masoretic Babel, Hittite and Vedic need 2.5 to 4.8 times the fastest recorded rate; the Septuagint date with small founder groups nearly fits.

By AI (Claude Fable 5.1, xhigh thinking; sources, companion files and checks by Claude Opus 5.5, thinking level unavailable; split into parent and children by Claude Opus 5.5) Draft:  October 2026

"A prediction from all of this would be that, given the large number of languages in the world today, it should be possible to group them together into 'families' like the Indo-European family of languages. But there should be no links between one 'family' and another."  Carl Wieland, young-earth physician and writer, 19991

"while perhaps most linguists believe it is possible that there was a single original human language from which all others have developed, most further believe it is impossible on the basis of extant evidence ever to be able to show that this is the case."  Lyle Campbell and William Poser, historical linguists, 20082

"What is known with a fair degree of certainty is the time between proto-Indo-Aryan and the modern Indo-Aryan languages - something of the order of 3,000 years. But how can anyone tell that the development from proto-Indo-European to proto-Indo-Aryan took another 3,000 years?"  R. M. W. Dixon, field linguist, 19973

Summary

Genesis 11 says that one language became many at Babel, and that the people then scattered.  The young-earth writers who have taken this up say that Babel made language families, not languages:  a few dozen "stem" languages, one per clan, each of which then split by ordinary change into the family linguists see today.  That model makes four claims a skeptic can test.  There should be a few dozen families.  There should be no demonstrable links between them.  The diversity inside each family should fit the time since Babel at observed rates of change.  And no family should be older than Babel.  This article states the model and tests the first claim.  Three child articles test the rest:  Did Language Begin Once? the links, How Old Are the Language Families? the ages and the written record, and Is There Time Since Babel for the Language Families? the rates of change, with a computation of its own.  The sibling articles cover the text and date of Babel, which peoples spoke what, and the same founder-effect argument for genes.

How many families there are.  Glottolog 5.3, counted here from its own data tables, has 238 families of two or more languages and 183 isolates, 421 independent units in all.4  Lyle Campbell's independent 2018 count is about 406.5  The young-earth counts are far lower:  Carl Wieland's "8 to 20 (commonly 12 or 13)",1 Duursma's "twenty or so",6 Ken Ham's "about one hundred".7  Those figures come from the lumpers, Greenberg and Ruhlen, whose method also claims the links between families that the same writers deny.  Only John Oller's and Bodie Hodge's statements, which give the Genesis 10 names as a floor of 70 to 78 and not a total, survive the real count.89

No demonstrated links between families.  This is the young-earth side's genuine strength, and it is the mainstream position too.  Nostratic, Eurasiatic, Amerind, Altaic and Proto-World are all rejected or called unproven by the specialists, for statistical reasons rather than ideological ones.102  But the same specialists say one origin can never be shown by the comparative method, because basic vocabulary decays past recognition in about 6,000 to 10,000 years on the mainstream calibration.1112  So an old-earth history and a Babel history predict the same absence of links.  The observation is consistent with Babel.  It is not evidence for it.

How old the families are.  Indo-European and Sino-Tibetan, dated on written texts from the last 3,450 years with no radiocarbon in them, come out far older than any Babel date:  Sino-Tibetan 7,184 years on four dated texts, Indo-European 8,120 years before AD 2000 on 52 dated ancient languages.1314  Their lower 95% bounds are older than every Masoretic Babel in print, except Zhang and colleagues' Sino-Tibetan bound of about 2233 BC, which falls nine years after Ussher's 2242 BC and so takes in his date and the earlier Masoretic ones.15  Japonic and Turkic are younger than Babel on any account, Dravidian's median falls between the Masoretic and Septuagint dates, and the deep families without old texts rest on radiocarbon.

The divergence-time test.  Because the Bayesian dates rest on models a skeptic of both sides may distrust, Is There Time Since Babel for the Language Families? runs a simpler test a reader can check.  It measures shared basic-vocabulary cognates in the IE-CoR database and calibrates the rate of loss on 59 text-dated ancestor-to-descendant pairs within the last 2,450 years.16  Hittite, written about 1450 BC, and Vedic Sanskrit, about 1300 BC, share 21% of their meanings.17  Under a Babel at 2242 BC every Indo-European lineage would have had to replace its words 2.45 times as fast as the fastest lineage recorded over a millennium, and 3.5 to 4.8 times as fast once rate variation among meanings is allowed for;  under the Septuagint's 2767 BC, 1.5 and 2.2 to 3.0.18  Faster word loss in small post-Babel groups, computed from published young-earth founder sizes, halves the Masoretic gap and closes the Septuagint gap at a single rate.19  Austronesian cannot exclude a post-Babel origin on modern word lists.20

Families written down before Babel.  On conventional dates, Egyptian is written by rebus at Abydos about 3320 BC and Sumerian is securely identified by about 2800 BC.2122  Those two unrelated families precede every Babel date in print, including the Septuagint's, and Elamite and Hurrian bring the count to four before the later Masoretic dates.  The young-earth answer is the revised chronology of Ashton and Down or Osgood, in which all writing is post-Babel;  that answer costs what Egypt and the Flood date and Mesopotamian absolute chronology show it costs.

Verdict.  The count of families refutes the young-earth writers who gave a number and spares the two who gave a floor.  The absence of links between families is real, matches the young-earth prediction, and is equally what the mainstream expects.  The ages of the text-dated families, and this article's own divergence-time test, are the hard problem:  under the Masoretic chronology the first thousand years after Babel would have to have changed language two to five times faster than anything recorded in the next three thousand, in every Indo-European lineage at once.  The Septuagint chronology, with the small founder populations the young-earth literature already posits, comes within reach of the data.  The attestation record puts two unrelated written languages before any Babel date unless the whole ancient Near Eastern chronology is lowered.  The honest young-earth position reads Babel on the longer chronology, treats the family as the unit Babel made, and says plainly that Indo-European and Sino-Tibetan need a burst of early change that no document records.

Part 1: The young-earth model in its own words

The young-earth literature on language is small, and most of it is in Creation magazine, the Journal of Creation and Answers in Genesis publications.  Its position has to be stated before it is tested, because the writers do not all say the same thing, and the differences decide which tests apply.

Henry Morris, 1976.  The Genesis Record is the founding commentary.  Morris read "one language and one speech" as "one phonology and one vocabulary", and held that the confusion "applied only to the phonologies, not to the underlying thought processes which are part of man's uniqueness."2323  The mechanism was a miracle:  "another miracle of creation was required to create many new such phonologies all at once".23  He tied the number loosely to the seventy nations of Genesis 10, "distinct family units, distinct languages (therefore, subsequent to Babel), distinct regions, and, finally, actual nations", while allowing that "There may well have been others".2323  He put the dispersion at Peleg's birth, 101 years after the Flood, "about 2358 B.C." on Thiele's date for Abraham.2323  And he left a door open that later writers have not used:  "If gaps are allowed, however, these dates might conceivably be increased by, say, up to about one thousand years or more (possibly five thousand years at the very outside)."23  Morris cites no linguist and offers no linguistic evidence.

Charles Taylor, 1983 and 1997.  Taylor, writing "as a linguist", doubted that God created new languages at all:  "the Genesis text does not go as far as saying God gave the people new languages; it merely says that 'God confused their language'."24  His preferred mechanism was "an imposed deafness" lasting several years.24  In 1997 he separated two miracles, "the original creation of Adam as a talking and understanding being; and the subsequent division of humanity into language groups", and said of Nostratic that it was introduced "In trying to reduce the number of distinct language families (and so avoid the miracle of Babel)".2525  He gave no source for that motive.  His deafness mechanism matters because it predicts the opposite of what the other writers predict:  one shared vocabulary, re-sounded, would leave cognates across the whole world.

Carl Wieland, 1999.  Wieland's "Towering change" is the model most creationist readers know.  Babel did not make today's languages.  "At Babel, the family groups were each programmed suddenly and supernaturally with a different language. Each of these created languages has gone on to diversify into many separate languages, which are all related to each other—each group has a 'common ancestor' language."1  Because little time had passed since the Flood, "Perhaps only a few dozen separate languages needed to be created in order for each clan to have its own".1  "Thus, I think that the Indo-European 'family of languages' would have all originated from one 'stem' language at Babel."1  Then the prediction quoted at the head of this article:  families, and "no links between one 'family' and another".1  He put the number of families at "some 8 to 20 (commonly 12 or 13)" and said "That fits very comfortably with the descriptions in Genesis."1  He cited no linguist for the count.

K. J. Duursma, 2002.  Duursma's Journal of Creation paper is the fullest young-earth linguistic case found.  "Secular linguists are puzzled by the existence of twenty or so language families in the world today. The languages within each family (and the people that speak them) have been shown to be genetically related, but few genetic links have been observed between families."6  "As the number of people alive at the time would not have been great, about a dozen or so languages would probably have been plenty."6  The confusion "is presented as a miraculous intervention by God. It is unlikely to have been acceleration of normal language changes".6  He took Greenberg's mass comparison as having "become the norm in comparative linguistics",6 used glottochronology's rate to argue that nothing can be traced beyond 10,000 years, so Proto-World is unprovable,6 and expected that on the same rate "we expect to see many similarities, still, between all languages in each family after only 5,000 years since Babel."6  The two sentences that follow are the only young-earth proposal found of faster change after Babel:  "the absence of any contact between languages in one family, over such a period, could result in greater word loss ... Also, some languages have a much quicker word loss, for cultural reasons."6  He read a textbook's suggested Proto-Indo-European date of 3000 BC as "consistent with the Biblical account".6

John Oller, 2008.  Oller is the most credentialed young-earth writer on the subject, a professional linguist.  In Answers in Genesis's Babel issue he gave 6,912 languages in 94 families, from the 2005 Ethnologue, and wrote of Proto-Indo-European and Proto-Sino-Tibetan:  "Neither of these distinct source languages, however, can reasonably be dated earlier than the miracle at Babel."8  He expected the Babel languages to be "radically distinct from each other".8  He held that observed rates suffice:  "The rate of change of known subgroups—notably those based on Latin, Sanskrit, Greek, and the Germanic and Slavic languages—shows that the whole number of languages in the world today could easily have been produced within a space of 4,000 years."8  And he made the one falsifiable prediction in this literature:  "the number of families will be reduced in the future to no fewer than the groups named in Genesis 10 where the Table of Nations appears."8  In the same article he reported that "secular linguists believe PIE was spoken in Europe sometime between the third and seventh millennium BC" and that Proto-Indo-European "is known to have split into the 449 Indo-European languages of today since approximately the time of Babel."88  The two fit together only at the young end of the range he reports:  a Masoretic Babel, which the young-earth writers found place between 2358 and 2008 BC, falls in the last centuries of the third millennium.

Bodie Hodge, 2013.  Hodge's Tower of Babel derives a floor from the text.  Counting the descendants of Noah's sons in Genesis 10, he arrives at 78 and calls it a "minimum number of language families (that is a tally of those listed in Genesis 10 subtracting Noah, Shem, Ham, Japheth, and Peleg)".9  Like Wieland, he makes the family, not the single language, the unit Babel created.

Ken Ham, 2022 and 2023.  The popular Answers in Genesis message equates the Babel families with the number linguists count.  2022:  "The Bible records about 70 or so 'language families' leaving Babel. Now, there are nearly seven thousand languages spoken today. ... Researchers also believe there's only about one hundred original language families!"26  2023:  "at Babel, he divided this language, creating about one hundred language families—and that's how many language families exist today!"7  No source is given for "about one hundred".

Murray Adamthwaite, 2019.  Adamthwaite, a PhD in ancient Near Eastern studies, is the one young-earth writer who uses the written record.  "all these languages appear on the scene at about the same time—the mid-third millennium BC on conventional chronology. For secular scholars, their origins remain obscure."27  "what is striking about the linguistic landscape of the Ancient Near East is the plethora of unrelated but highly complex languages."27  The confusion was "a sudden, supernatural act of God which created a whole suite of unrelated but highly complex languages."27  His footnote concedes that written attestation can lag a language's existence "by centuries".27  How Old Are the Language Families? tests his dating.

Jeffrey Tomkins, 2015.  The only young-earth use found of the phoneme-diversity debate.  Tomkins read Creanza and colleagues' 2015 finding, that phoneme inventory sizes do not follow a serial founder effect out of Africa, as showing that "multiple people groups have independent origins—a condition one would predict if the confusion of languages at the Tower of Babel happened as described in the Bible."28  His quotation of the paper is accurate.  His bracket "[single line of descent]", inserted into it, changes the meaning of "serial founder effect", and the inference is not the paper's.28  Did Language Begin Once? returns to it.

What the model claims, assembled.  Setting Taylor and Morris aside, the published model is consistent from Wieland to Ham:  Babel made a small number of unrelated languages, each the "stem" of one family;  each family then split by ordinary change;  families should show no links to one another;  and the time since Babel suffices, at observed rates, for the diversity inside each family.  Every writer found who gives a date for Babel puts it between 2767 BC and 2008 BC:  Morris's 2358 BC, the 2242 BC that Answers in Genesis prints, Andrew Sibley's 340 years after the Flood, which is 2008 BC on Ussher's Flood, Gavin Cox's "around 2,500 BC", and E. J. Tweedy's 2767 BC, Peleg's birth on the Septuagint, which he calls "an arbitrary assumption".293031  Most use the Masoretic dates.  Nobody takes Morris's thousand-year allowance.  Two claims differ in kind from the rest.  Wieland and Ham state the number of families as a fact that fits Genesis.  Oller and Hodge state it as a floor.  The first kind can be refuted by a count.  The second cannot, and Part 2 says so.

What no young-earth writer proposes.  No writer found argues that Babel produced languages that were already partly related to each other.  Oller's "radically distinct" says the opposite.  No writer mentions creoles, pidgins or mixed languages in connection with Babel.  No writer applies a revised ancient chronology to the dates of the Hittite, Linear B or Vedic texts.  No writer engages the Bayesian family dates, Bergsland and Vogt's critique of glottochronology, or Atkinson's phoneme gradient directly.  These absences matter in Is There Time Since Babel for the Language Families?, where some of the ways a young chronology could be rescued are tested and must be labelled as that article's hypotheticals rather than the model's claims.

Part 2: How many families there are

The count.  A language family is a set of languages shown by the comparative method to descend from one ancestor.  An isolate is a family with one member.  Glottolog, the Max Planck Institute's catalogue, classifies a language into a family only when "the best explanation for the similarities are inheritance from common ancestor", following the methods handbook of Campbell and Poser.32  It does not accept Altaic, Nostratic, Amerind, Nilo-Saharan or Khoisan.  Counted here from the Glottolog 5.3 data tables, there are 238 families of two or more languages and 183 isolates, 421 units in all.4  Of the 7,674 spoken first languages, 7,491 sit inside the 238 families.  Of the 421 units, 27 families and 80 isolates have no living member;  Sumerian, Elamite, Hattic and Etruscan are among them.  So 314 units have a living speaker.  Glottolog's own summary page says "246 families and 183 isolates";  the extra eight are its non-genealogical buckets for sign languages, pidgins, unclassifiable and unattested languages and the like, so the genealogical figure is 238.32

Two other counts agree.  Lyle Campbell, the leading skeptic of long-range proposals, listed region by region the families that approach consensus and arrived at "406 independent language families (including language isolates)", adding at once that "this number is far from straightforward or conclusive".5  Of the 406, 94 are extinct, "23.2% of the linguistic diversity of the world, calculated in terms of language families, is now lost", and 162 are isolates, of which 60 are extinct.55  Harald Hammarström, Glottolog's editor, counted 424 in 2016.12  Three independent counts within 5% of each other over a decade is as stable as this number gets.  Ethnologue's 143 is a different unit:  it keeps Niger-Congo, Nilo-Saharan, Australian and Trans-New Guinea as single families and leaves its 99 isolates out of the family count, so it must not be set beside Glottolog's 421 as if they measured the same thing.33

Where the families are.  The 421 units are not spread evenly.

Macroarea Families Isolates Units Spoken languages
Africa 36 15 51 2,212
Eurasia 23 12 35 1,740
Papunesia 72 52 124 2,124
Australia 23 9 32 355
North America 40 31 71 671
South America 44 64 108 572
Total 238 183 421 7,674

The Americas hold 179 of the 421 units but only 16% of the languages.  New Guinea and island Melanesia hold 124 more.  Africa and Eurasia, with more than half the world's languages, hold 86 units between them.4  Five families hold 55.7% of all family languages:  Atlantic-Congo with 1,408, Austronesian 1,276, Indo-European 586, Sino-Tibetan 519, Afro-Asiatic 381.  At the other end, 135 of the 238 families have five languages or fewer, and 61 have exactly two.

Johanna Nichols and Colin Renfrew explain the pattern as spread zones and residual zones.  A spread zone, such as the Eurasian steppe or most of Australia, has few families over a large area and shallow time depth, because one language expanded and wiped out its predecessors;  a residual zone, such as the Caucasus, northern Australia or New Guinea, keeps many families and deep diversity.34  Renfrew argued that most spread-zone families are farming dispersals of the last 10,000 years.34  The mechanism is real and mapped.  It is also neutral on absolute time:  a Babel model can use the same mechanism, and Renfrew's "at least 10,000 years" for the residual zones is the radiocarbon chronology of settlement, not a linguistic measurement.34  Hammarström adds a point that cuts against anyone, on either side, who reads "more families" as "more time":  South America, settled about 11,000 years ago on the mainstream chronology, is as diverse as New Guinea at about 49,000, and "Africa, presumably the continent of origin, exhibits much less diversity."12

The creationist counts against the real one.  Wieland's "8 to 20 (commonly 12 or 13)" and Duursma's "twenty or so" are both refuted by a factor of twenty or more.  Where did they come from?  Duursma cites Merritt Ruhlen's 1987 Guide to the World's Languages, and the count there is Greenberg's:  "Greenberg (1987) classifies the world's languages into the 15 families shown in List 7.3".35  Greenberg's families are the products of mass comparison, the method Campbell, Ringe and Glottolog reject, and Ruhlen uses it on the next page to argue that even those fifteen will merge:  "future consolidation of some of the groups in List 7.3 is likely".35  So the young-earth side took the lumpers' count and rejected the lumpers' conclusion.  Duursma's claim that Greenberg's method "has become the norm in comparative linguistics" is false;  Campbell and Poser's handbook, the one Glottolog follows, treats it as the method to avoid.2  Ham's "about one hundred" is wrong by a factor of four and cites nothing.  Oller's 94 was the 2005 Ethnologue's count, a lumped figure of the same kind as its 143 today, not a count of independent units.

Oller's and Hodge's statements fare differently.  Oller predicted that the number of families "will be reduced in the future to no fewer than the groups named in Genesis 10".8  Hodge calls his 78 the "minimum number of language families".9  A floor of 70 to 78 is safe with 421 units counted, and will stay safe:  Campbell expects the number to fall as comparative work joins families and to rise as uncontacted groups are documented, but no realistic programme of joining gets from 406 to 78.5  What the floor does not do is explain the excess:  no young-earth writer found says why there should be 421 units rather than 78 or any other number.

What the count does and does not measure.  Campbell's definition is the key fact for both sides:  the world's families are "the set of independent families (including isolates) that cannot reliably be shown to be related to any other language family."5  The comparative method can join two languages.  It can never prove two languages unrelated.  So 421 is an upper bound on the number of separate origins, not a measurement of it, and Glottolog's editors say their classification is "the best guess" and "subject to change as scholarship and interest in those languages increase."32  Dene-Yeniseian, discussed in Did Language Begin Once?, is a live case of two of Glottolog's families that specialists suspect are one.  A young-earth model that takes 421 as the number of languages created at Babel would be committing itself to a number the data cannot fix.  A Babel model needs the number of families to be at least the number of Babel languages, and the data give it that;  it needs nothing more from the count, and the count gives nothing more.

Did language begin once?

Linguistics cannot say.  Donald Ringe, Johanna Nichols, and Lyle Campbell and William Poser all hold that one origin is likely or possible and that it cannot be demonstrated, because basic vocabulary decays past the comparative method's reach after 6,000 to 10,000 years.10112  So the absence of links between the 421 units, which Wieland predicted, is what any history older than that horizon predicts too.  It is consistent with Babel and is not evidence for it.

The proposals of one origin or of deep links, Ruhlen's Proto-World, Nostratic, Eurasiatic, Altaic and Pagel's ultraconserved words, fail statistical tests, and Atkinson's phoneme gradient out of Africa did not survive reanalysis.  A creationist who cites Proto-World inherits a rejected method and contradicts the young-earth model of unrelated Babel languages.  The sign languages of a Negev clan and of Nicaragua show that new languages can arise within a generation among people who already have the faculty.  The evidence is in Did Language Begin Once?.

How old are the language families?

The families dated on written texts alone are the oldest.  Every text their dating runs use is within the last 3,450 years, yet Sino-Tibetan comes out at 7,184 years and Indo-European at 8,120.1314  The model-dependence is real, from Chang's 5,950 years to Heggarty's 8,120, but the disagreement is between 6,000 and 8,000 years, not between 4,000 and 8,000.  Only Zhang and colleagues' Sino-Tibetan interval reaches a Masoretic Babel:  its lower bound, about 2233 BC, is nine years after Ussher's 2242 BC.3615  Japonic and Turkic are younger than Babel, Dravidian's median lies between the Masoretic and Septuagint dates, and the deep families without old texts rest on radiocarbon.

On conventional dates two unrelated families, Egyptian and Sumerian, are written down before every Babel date in print, and four before the later Masoretic dates.  Adamthwaite's claim that the Near Eastern languages appear at the same time fails, since their first texts span about 2,500 years.  The young-earth revised chronologies remove the problem by making all writing post-Babel, at the cost set out in the Egypt and Mesopotamia articles.  The evidence is in How Old Are the Language Families?.

Is there time since Babel for the language families?

Not on the Masoretic date.  Documented lineages keep 70 to 85% of their basic vocabulary per millennium, and in 1,000 to 2,200 years Latin, Old Norse and Old Chinese each became a group of plainly related languages, not unrelated families.  The divergence-time test finds that Hittite and Vedic, under a Babel at 2242 BC, need every Indo-European lineage to have changed 2.45 times as fast as the fastest lineage recorded over a millennium, and 3.5 to 4.8 times as fast once rate variation is allowed for.18

The Septuagint's 2767 BC lowers the factors to 1.5 and 2.2 to 3.0.  Faster word loss in small founder groups, the one young-earth proposal of faster change, halves the Masoretic gap and closes the Septuagint one at a single rate, when computed from published young-earth founder sizes;  with the Late Bronze Age lowered by 550 years, on the scale of Habermehl's chronology though she does not apply it to language, the Septuagint case fits even with rate variation.19  The Masoretic date then fits at a single rate, and with rate variation only by stacking every favourable choice.  Austronesian, with no Bronze Age texts, cannot exclude a post-Babel origin.20  The rates, the code and every escape route are in Is There Time Since Babel for the Language Families?.

Objections and responses

"This article leans on the comparative method's horizon to say the no-links result is not evidence for Babel, then leans on the same method's dates to say the families are too old.  Which is it?"  Both, because they are different measurements.  The horizon is about detecting a relationship across the chance floor.  The family ages are about how much change has accumulated inside a demonstrated family, where cognacy is not in doubt.  The first says nothing about Babel either way.  The second is a measurement of elapsed time, with its rate the only unknown, and Is There Time Since Babel for the Language Families? bounds the rate from the documented cases.

Issues and unanswered questions

  • Hodge's book has not been read with page numbers. Only his statement of the 78-family floor could be checked twice; his breakdown of the 78 by son of Noah and his remarks on Greek and on the isolates are held in the quarantine file until a copy is read.
  • Dixon's Woodlark Island case is cited by him from Lithgow 1973, which has not been read at source; the Muyuw 13%-in-50-years figure comes through Nettle.
  • Which young-earth model is the model? Taylor's deafness mechanism and Morris's gap allowance each contradict the Wieland-Hodge model on a point that matters here. The organisations publish all of them. A reader is entitled to ask which one is being defended before any test is run.
  • The 421 count's upper-bound status cuts against this article's own strongest finding. If further comparative work joined Indo-European to Uralic, or Dene-Yeniseian were demonstrated, the young-earth no-links prediction would fail for that pair and the family ages would get older, not younger.

Verdict

On the count.  There are about 406 to 424 independent families and isolates, and the number has been stable for a decade.  Wieland's dozen and Ham's hundred are refuted;  Oller's and Hodge's floor of 70 to 78 is safe and will stay safe.  Nothing in the young-earth model explains why there are 421 units rather than 78, and nothing in the data fixes the number of separate origins, which runs from 1 to 421.

On links between families.  None is demonstrated, and the specialists reject every proposed macro-family on statistical grounds.  This is the young-earth side's genuine strength, and it should be stated plainly:  the family map looks the way separate creation at Babel would make it look.  It also looks the way 100,000 years of decay past the comparative method's horizon would make it look, and the linguists who say so are the same ones who reject Proto-World.  The observation does not discriminate.  The creationist who cites Ruhlen for one original language has it backwards.

On the ages.  The families dated on written texts alone, with no radiocarbon anywhere in them, come out at 6,000 to 8,000 years for Indo-European and Sino-Tibetan, with lower bounds older than every Masoretic Babel in print except one Sino-Tibetan bound, which falls nine years after Ussher's date.  This article's own test, with no Bayesian model in it, gives the same answer from the Bronze Age texts:  Hittite and Vedic share a fifth of their basic vocabulary at 1450 and 1300 BC, and a Masoretic Babel requires every Indo-European lineage to have changed 2.45 times faster than the fastest lineage recorded over a millennium, 3.5 to 4.8 times faster once rate variation is counted, for the first eight to nine centuries after the dispersion.  The two escape routes with young-earth sources, Duursma's faster loss in small groups and Habermehl's lowered Late Bronze Age, which she does not apply to language, are computed here from the movement's own numbers.  Faster loss alone halves the Masoretic gap and closes the Septuagint one at a single rate.  With the lowered dates as well, the Masoretic single-rate case fits too, but the Masoretic gap with rate variation closes only if every favourable assumption, Atkinson's bursts included, is taken at once.  Austronesian, with no old texts, fits a post-Babel origin at the fastest Polynesian rate, and that is a real point for the young-earth side.

On attestation.  Egyptian and Sumerian are written before every Babel date in print on conventional chronology, and Elamite and Hurrian before the later Masoretic ones.  The young-earth answer is to lower the whole ancient chronology, and the cost of that is settled in the Egypt and Mesopotamia articles, not here.

What an honest young-earth position looks like.  Read Babel on the Septuagint chronology or with Morris's gap, which the movement's founder allowed and his successors have not taken up.  Take the family as the unit Babel made, as Wieland and Hodge do, and give the Genesis 10 names as a floor, not a count.  Claim the absence of links as consistent with Babel and no more.  Say plainly that Indo-European requires a burst of early change, two to three times anything recorded, in small founder groups whose later states were not small, that the text-dated Sino-Tibetan ages point the same way, and that no document records it.  Do not cite Proto-World, do not cite Wang's Babel sentence, do not cite the ASJP dates without Bostoen's reply, and do not say the ancient Near Eastern languages appear at the same time.  That position loses the Masoretic date on the language evidence and keeps the rest.

Companion pages

Sources

  1. 1.

    Wieland, Carl.  “Towering change.”  Creation 22(1):22–26.  Dec 1999.  A popular magazine article that states the young-earth "stem language" model most creationist readers know:  Babel made a few dozen unrelated languages, one per clan, and each then diversified into a family by ordinary change.  It predicts about 8 to 20 families with no links between them, and cites no linguist for the count.

    1. 1a.

      Section What about Babel?:  “A prediction from all of this would be that, given the large number of languages in the world today, it should be possible to group them together into 'families' like the Indo-European family of languages. But there should be no links between one 'family' and another.” ↩ ↩

    2. 1b.

      Section What about Babel?:  “Thus, estimates of the number of different 'language families' vary, and are difficult. But they are generally in the vicinity of some 8 to 20 (commonly 12 or 13). That fits very comfortably with the descriptions in Genesis.” ↩ ↩

    3. 1c.

      Section Diagram caption, (C) Genesis Orchard:  “At Babel, the family groups were each programmed suddenly and supernaturally with a different language. Each of these created languages has gone on to diversify into many separate languages, which are all related to each other—each group has a 'common ancestor' language.” ↩

    4. 1d.

      Section What about Babel?:  “Since very little time had elapsed since the Flood, there would not have been hundreds of clans yet. Perhaps only a few dozen separate languages needed to be created in order for each clan to have its own”. ↩

    5. 1e.

      Section What about Babel?:  “Thus, I think that the Indo-European 'family of languages' would have all originated from one 'stem' language at Babel.” ↩

  2. 2.

    Campbell, Lyle, and William J. Poser.  Language Classification:  History and Method.  Cambridge University Press.  2008.  The standard handbook on how language relationships are proved, by two historical linguists, and the classification standard Glottolog names as its own.  It accepts the comparative method and rejects multilateral comparison, glottochronology as a test of relationship, and global etymologies, on statistical grounds.  It is the strongest mainstream witness that no family older than about 8,000 years is demonstrated and that one origin of language can never be shown.

    1. 2a.

      Page 366, note 3:  “while perhaps most linguists believe it is possible that there was a single original human language from which all others have developed, most further believe it is impossible on the basis of extant evidence ever to be able to show that this is the case. This means for some scholars that not only is monogenesis a possibility, so is polygenesis, but that we will never know for sure which is right.” ↩ ↩

    2. 2b.

      Page 243, section 9.4:  “The Nostratic hypothesis has received considerable attention in recent years … It is better known than most other proposed macro-families, but it remains controversial. Most historical linguists either reject it (see Campbell 1998, 1999; Ringe 1995a; Trask 1999; Vine 1991) or ignore it.” ↩

    3. 2c.

      Page 171:  “Ringe (1999:219) points out a fatal flaw in the logic of multilateral comparison”.  Pages 171–172:  Quoting Ringe 1999, p. 229, on Greenberg's claim.  “as one compares an increasing number of languages the probability of a match in sound and meaning quickly falls to the point where it can be ignored … is false, indeed, its falsity ought to have been obvious”. ↩

  3. 3.

    Dixon, R. M. W.  The Rise and Fall of Languages.  Cambridge University Press.  1997.  A short book on how language families arise and how far they can be dated, by a field linguist of Australian languages who holds that whether all Australian languages share one ancestor cannot be answered.  Its pages 47 and 48 are a mainstream linguist's plain statement that the rate of change measured over the documented last 3,000 years of Indo-Aryan says nothing certain about the earlier, undocumented part.

    1. 3a.

      Page 47:  “What has always filled me with wonder is the assurance with which many historical linguists assign a date to their reconstructed proto-language. … We are told that proto-Indo-European was spoken about 6,000 years ago. What is known with a fair degree of certainty is the time between proto-Indo-Aryan and the modern Indo-Aryan languages - something of the order of 3,000 years. But how can anyone tell that the development from proto-Indo-European to proto-Indo-Aryan took another 3,000 years?” ↩

  4. 4.

    Counted from the Glottolog 5.3 data tables32 with glottolog-counts.ts;  the output is glottolog-counts.md.  The tables hold 246 top-level family languoids, but 8 of them are not genealogical:  Bookkeeping, Sign Language, Unclassifiable, Pidgin, Unattested, Artificial Language, Speech Register and Mixed Language.  That leaves 238 genealogical families, every one with two or more languages, plus 183 isolates, spoken first languages that belong to no family:  238 + 183 = 421 units.  27 families and 80 isolates have no living member, so 421 - 27 - 80 = 314 have one.  Glottolog's own page says "246 families", which appears to count the 8 buckets.  Assigning each family to the macro-area that holds most of its languages gives Africa 51 units, Eurasia 35, Papunesia 124, Australia 32, North America 71 and South America 108. ↩ ↩ ↩

  5. 5.

    Campbell, Lyle.  “How Many Language Families are there in the World?”  Anuario del Seminario de Filología Vasca "Julio de Urquijo" 52(1–2):133–152.  2018.  A region-by-region list of the language families that approach a consensus among specialists, by the leading skeptic of long-range groupings.  It finds about 406 independent families, 162 of them isolates and 94 extinct, and leaves out unclassified languages, pidgins, creoles and sign languages.  By its own definition it counts what cannot be shown to be related, not what is known to be unrelated.

    1. 5a.

      Page 133:  “How many language families are there in the world? Surprisingly, most linguists do not know. Estimates range from one (according to supporters of Proto-World) to as many as about 500. … let me report the answer here: 406 independent language families (including language isolates). However, this number is far from straightforward or conclusive, as we shall see.” ↩ ↩

    2. 5b.

      Page 141:  “Thus, the total of language families in the world is c. 406.  Of these 406 language families, 94 are extinct … This means that 23.2% of the linguistic diversity of the world, calculated in terms of language families, is now lost.” ↩

    3. 5c.

      Page 143:  “there are many more language isolates than most people, including most linguists, are aware of – 162 total. That is 39.9% of the world's c.406 language families. … Of these language isolates, 60 are extinct (37%).” ↩

    4. 5d.

      Page 149:  “The imprecise number of approximately 406 independent language families (including 162 language isolates) is as close as we can come now to answering the question … Some of the known families will probably be adequately demonstrated to be related to other known families through additional comparative work … reducing the total number of language families. Other discoveries may increase the number of language families.” ↩

    5. 5e.

      Page 137:  “A language family is the set of languages for which there is sufficient evidence to show that they descend from a common ancestor. Language isolates are also language families, but have only one member language, not related to any other language or language family. The ultimate number of language families in the world is the set of independent families (including isolates) that cannot reliably be shown to be related to any other language family.” ↩

  6. 6.

    Duursma, K. J.  “The Tower of Babel account affirmed by linguistics.”  Journal of Creation 16(3):27–31.  Dec 2002.  The fullest young-earth linguistic case found.  It takes about twenty families from the lumpers' classification, holds that Babel was a miracle making about a dozen languages, and uses glottochronology's rate to set a 10,000-year horizon on reconstruction.  It is also the only young-earth source found that proposes faster word loss in isolated post-Babel groups.

    1. 6a.

      Section Opening paragraphs:  “Secular linguists are puzzled by the existence of twenty or so language families in the world today. The languages within each family (and the people that speak them) have been shown to be genetically related, but few genetic links have been observed between families.” ↩ ↩

    2. 6b.

      Section Language families:  “As the number of people alive at the time would not have been great, about a dozen or so languages would probably have been plenty.” ↩

    3. 6c.

      Section Language families:  “The Babel account suggests that several languages came into existence on that day. It is presented as a miraculous intervention by God. It is unlikely to have been acceleration of normal language changes”. ↩

    4. 6d.

      Section Language classification:  “A dramatic breakthrough came in the person of Joseph Greenberg in the middle of last century. … His method has become the norm in comparative linguistics.” ↩

    5. 6e.

      Section Evidence for one single proto language?:  “In some languages this change has been measured for more than 2,000 years. The result shows that 19.5% of the core vocabulary changes every 1,000 years. If this is the same for all languages, it means that statistically all words in a language should be replaced within a period of about 10,000 years. That would make any research beyond that period of time impossible. This, in turn, makes it impossible to prove that all language families are ultimately related.” ↩

    6. 6f.

      Section Language families:  “If the percentage of word loss described above is correct, we expect to see many similarities, still, between all languages in each family after only 5,000 years since Babel. However, the absence of any contact between languages in one family, over such a period, could result in greater word loss … Also, some languages have a much quicker word loss, for cultural reasons.” ↩ ↩

    7. 6g.

      Section Timing consistent:  “The 3,000 years BC, which Robins mentions, is significant, as such a time span is consistent with the Biblical account.” ↩

  7. 7.

    Ham, Ken.  “How Many Languages?”  Answers with Ken Ham, volume 149.  Answers in Genesis.  Apr 26, 2023.  The transcript of a short radio feature by the founder of Answers in Genesis, the popular form of the young-earth message on languages.  It equates the families made at Babel with the number linguists count today, about one hundred, and gives no source for that number.

    1. 7a.

      Unpaginated radio transcript.  “Well, language scientists say they came from fewer than one hundred original language families. … But, at Babel, he divided this language, creating about one hundred language families—and that's how many language families exist today!” ↩ ↩

  8. 8.

    Oller, John W. Jr.  “More Than PIE.”  Answers 3(2).  2008.  An article in Answers in Genesis's Babel issue by a professional linguist, the most credentialed young-earth writer on the subject.  It gives 6,912 languages in 94 families from the 2005 Ethnologue, holds that Babel made radically distinct languages and that 4,000 years suffice for today's diversity, and predicts that the count of families will never fall below the Genesis 10 names.

    1. 8a.

      Section Babel—Key to Unlocking Language Families:  “Also, we can make a testable prediction: the number of families will be reduced in the future to no fewer than the groups named in Genesis 10 where the Table of Nations appears.” ↩ ↩ ↩

    2. 8b.

      Section A Patchwork of Languages:  “That language family, too, is believed to have come from a common source, but different from PIE. Neither of these distinct source languages, however, can reasonably be dated earlier than the miracle at Babel.” ↩

    3. 8c.

      Section Babel—Key to Unlocking Language Families:  “Because God does all things well, nothing half-heartedly or without complete effectiveness, the languages created at Babel will almost certainly turn out to be radically distinct from each other.” ↩

    4. 8d.

      Section Babel—Key to Unlocking Language Families:  “The rate of change of known subgroups—notably those based on Latin, Sanskrit, Greek, and the Germanic and Slavic languages—shows that the whole number of languages in the world today could easily have been produced within a space of 4,000 years.” ↩

    5. 8e.

      Section Reconstruction of Early Languages Hits a Wall:  “Based on the scant historical evidence, secular linguists believe PIE was spoken in Europe sometime between the third and seventh millennium BC, give or take a couple thousand years.” ↩

    6. 8f.

      Section Babel—Key to Unlocking Language Families:  “For example, Proto-Indo-European is known to have split into the 449 Indo-European languages of today since approximately the time of Babel.” ↩

  9. 9.

    Hodge, Bodie.  Tower of Babel:  The Cultural History of Our Ancestors.  Master Books.  2013.  A book-length young-earth treatment of Babel by an Answers in Genesis writer.  It holds that Babel made language families, not languages, and derives a floor of at least 78 families from the names in Genesis 10;  it is the only young-earth count derived from the text.

    1. 9a.

      Page number not available.  “minimum number of language families (that is a tally of those listed in Genesis 10 subtracting Noah, Shem, Ham, Japheth, and Peleg)”. ↩ ↩ ↩

  10. 10.

    Ringe, Donald A. Jr.  “On Calculating the Factor of Chance in Language Comparison.”  Transactions of the American Philosophical Society 82(1).  American Philosophical Society.  1992.  A monograph that computes with binomial probabilities how many sound-and-meaning matches two unrelated languages will show by chance in a 100-word list.  Real families clearly exceed chance and languages not believed related do not, while comparing many lists at once finds systematic resemblances even among artificial ones.  It is the statistical core of the mainstream rejection of Amerind, Eurasiatic and Proto-World.

    1. 10a.

      Page 80, section 11:  “(6) Comparison of the vocabularies of several languages at once normally yields a pervasive pattern of systematic similarities, even when the languages in question are artificial constructs (and thus necessarily unrelated). … of all the possible forms of comparison considered in this paper, multilateral comparison is the least reliable and the most likely to be positively misleading.” ↩

    2. 10b.

      Page 1:  “The consensus of opinion among mainstream historical linguists is that while all human languages are likely to be genetically related, the remoter relationships cannot be demonstrated by reliable linguistic methods because the languages in question have diverged too much.” ↩

  11. 11.

    Nichols, Johanna.  “The Origin and Dispersal of Languages:  Linguistic Evidence.”  The Origin and Diversification of Language, Memoirs of the California Academy of Sciences 24, 127–170.  California Academy of Sciences.  Edited by Nina G. Jablonski and Leslie C. Aiello.  1998.  A chapter by a leading typologist that defines the "stock" as the oldest grouping the comparative method can prove, puts the oldest stocks at about 6,000 years, and explains that ceiling by vocabulary loss.  It then models how fast stocks multiply and finds that one ancestor would need at least 72,000 to 132,000 years to yield the world's stocks.  Every input of that model is calibrated on the mainstream ages of known families.

    1. 11a.

      Page 128:  “After 6000 years of separation, two languages are expected to exhibit only 7% shared cognates; and 7% represents the lowest number of resemblant items that can safely be considered distinct from chance … Hence the 6000-year age of the oldest securely reconstructable language families is also the age after which, on binary tests, the incidence of cognates slips below the level of significance.” ↩

    2. 11b.

      Page 141:  “The two polygenetic scenarios both imply multiple ancestors to modern languages, and the two monogenetic scenarios both imply a single ancestor.  There is of course no way of establishing the number of ultimate ancestral languages by tracing descent …” ↩

  12. 12.

    Hammarström, Harald.  “Linguistic diversity and language evolution.”  Journal of Language Evolution 1(1):19–29.  2016.  A review by the editor of Glottolog that counts 424 families including isolates and tabulates them by continent.  It explains why the comparative method cannot see past about 10,000 years and notes that the density of families does not follow the age of settlement:  South America is as diverse as New Guinea, and Africa much less so.

    1. 12a.

      Page 23:  “An estimate of time-depth is possible thanks to a number of calibration points, either historical, inscriptional or archaeological (in the rare event that one can convincingly argue a link between an archaeological entity and a (proto-)language). The deepest families recognized as 'demonstrated' in, for example, Hammarström et al. (2015) are expected not to exceed 10,000 years. Unless there is a breakthrough in the way language families are demonstrated—that goes beyond basic vocabulary—this time limit cannot be improved upon: if languages share too little vocabulary we would not accept they are related while if languages share a lot of vocabulary, we do not think the relation is old.” ↩

    2. 12b.

      Page 22, section 4:  “The current understanding of demonstrated families admits no less than 424 families in the count of Hammarström et al. (2015). The genealogical diversity is unevenly distributed across continents, see Table 2.” ↩

    3. 12c.

      Page 23:  “Also the distribution of genealogical diversity is not directly indicative of time of original settlement by Homo sapiens of different continents. As per Table 2, the most diverse areas are South America and Greater New Guinea which differ dramatically in the age of settlement—11,000 for South America (Waters and Stafford 2007) vs 49,000 years ago for New Guinea (Summerhayes et al. 2010) and, Africa, presumably the continent of origin, exhibits much less diversity.” ↩

  13. 13.

    Sagart, Laurent et al.  “Dated language phylogenies shed light on the ancestry of Sino-Tibetan.”  Proceedings of the National Academy of Sciences 116(21):10317–10322.  2019.  A Bayesian dating of 50 Sino-Tibetan languages on cognate-coded basic vocabulary.  Its four calibrations are all written texts no younger chronology disputes, Old Chinese, Old Tibetan, Tangut and Old Burmese, yet the root comes out about 7,184 years old.  The date is an extrapolation of a rate learned on branches no older than 2,800 years.  Mirror:  PMC

    1. 13a.

      Section Results, Tree Topologies and Dating:  “The mean root age estimated with the relaxed-clock model is at 7184 B.P., with 95% highest posterior density interval (HPD) [5093–9568] B.P.” ↩ ↩

  14. 14.

    Heggarty, Paul et al.  “Language trees with sampled ancestors support a hybrid model for the origin of Indo-European languages.”  Science 381:eabg0818.  2023.  A Bayesian dating of 161 Indo-European languages on the IE-CoR cognate database, coded by more than 80 specialists, with 52 ancient languages dated from their texts as calibrations and a model that lets the data decide which ancient languages are ancestors.  It finds a root about 8,120 years before AD 2000, and reports its own sensitivity runs, one of which, on older data with forced ancestry, gives 4,997.

    1. 14a.

      Section Results:  “Our main analysis (Fig. 2) produced an estimated date for the root of the Indo-European language family that is too early to be compatible with the Steppe hypothesis: ~8120 yr B.P., with a 95% credible region of 6740 to 9610 yr B.P. [Date estimates are reported here as a median date before present, followed by the 95% credible region (highest posterior density, or HPD), all rounded to the nearest decade, and taking the "present" for modern languages as 2000 CE.]” ↩ ↩

  15. 15.

    Zhang, Menghan et al.  “Phylogenetic evidence for Sino-Tibetan origin in northern China in the Late Neolithic:  Supplementary Information.”  Nature 569.  2019.  The supplement to a Bayesian dating of 109 Sino-Tibetan languages on 949 root-meanings.  Its eight calibrations all come from written history or recorded migrations, none from radiocarbon, and the root comes out about 5,871 years before 1950.  Its lower 95% bound is the only dated family root found that reaches into a Masoretic Babel window.

    1. 15a.

      Page 9, section 2.4:  “The average time estimation for the divergence of the Sinitic and TB languages was 5,871 years BP (95% Highest posterior density (HPD): 4,183 – 7,817 years BP). In all the examined models, their estimations including average root ages and 95% HDP ranges were below 9,000 years (see Extended Data Fig. 5).” ↩ ↩

  16. 16.

    calibrate-ie.ts on IE-CoR37, with the attestation dates and their textual anchors in ie-dates.ts;  the full table, 113 rows, is ie-calibrations.tsv.  For a documented ancestor and descendant that share a fraction S of their meanings over t millennia, the retention per lineage per millennium is r = S^(1/t).  The 59 strict pairs, where the old language is the recorded ancestor, give, with loans counted as replacements:  fastest on any interval 0.556, Middle Cornish of 1300 to Late Cornish of 1700;  fastest on an interval of a millennium or more 0.694, New Testament Koine of AD 100 to Cypriot Greek, with Old English to English at 0.695;  10th percentile 0.705;  median 0.779;  75th percentile 0.802;  slowest 0.894, Old Czech to Czech.  With loans set aside as missing:  fastest 0.726, median 0.821, slowest 0.919.  Pairs whose old end is a close sister of the true ancestor, such as Old Icelandic to Norwegian, are kept out of the strict set, because they include extra divergence and so would make the rates look faster. ↩

  17. 17.

    Hittite, dated 1450 BC, and Early Vedic, 1300 BC, share S = 0.212 of 132 meanings.  At the fastest millennium-scale rate, r = 0.694, L = ln 0.212 / ln 0.694 = 4.25 millennia, and the common ancestor sits at (-1450 - 1300 - 4250) / 2 = -3500, that is 3500 BC, with a bootstrap range of 3118 to 3961 BC.  At the Cornish rate, 0.556, it is 2695 BC;  at the median, 0.779, 4486 BC;  at the slowest, 0.894, 8322 BC.  With loans set aside as missing and the matching fastest rate, 0.726, it is 3777 BC.  The modern comparisons, from the same code:  Spanish and Russian 21.8%, English and Latin 21.8%, Attic Greek and Latin 26.5%, Lithuanian and Russian 35.3%. ↩

  18. 18.

    If Indo-European began at Babel in year TB, two lineages attested in years dX and dY have had Lb = (dX - TB) + (dY - TB) years.  The speed-up k is the lineage length the stated rate needs to produce the observed S, divided by Lb;  below 1 the pair fits.  For Hittite and Vedic under a Babel at 2242 BC, Lb = 792 + 942 = 1,734 years.  At the fastest millennium rate the needed length is 4.25 millennia, so k = 4.25 / 1.734 = 2.45;  at the Cornish rate it is ln 0.212 / ln 0.556 = 2.64 millennia, k = 1.52.  For the gamma columns the needed length is beta x (S^(-1/alpha) - 1), with the fitted shape alpha and scale beta, divided by a factor that makes the gamma model reproduce the faster of the two fastest millennium-scale calibrations:  on the disjoint fit 10.53 / (1.757 x 1.734) = 3.45, and on the all-pairs fit 4.80.  Under the Septuagint's 2767 BC, Lb = 2,784 years and the factors are 0.95, 1.53, 2.15 and 2.99.  Code babel-ie.ts;  output ie-babel-speedup.tsv. ↩ ↩

  19. 19.

    This is the article's construction from young-earth inputs;  no young-earth writer has made it.  The inputs:  people at Babel, about 25,000 from Osgood or 1,000 from Carter and Hardy;  Babel languages, about a dozen from Duursma or at least 78 from Hodge;  founders per lineage = people / languages / 2, because splitting Proto-Indo-European in two at Babel gives the root pair the most time, which gives 6.4, 160, 1,042 or 1,667.  Bromham and colleagues' loss slope, b = 0.12, makes the loss rate at population N, relative to a calibration lineage at Nc, (N / Nc)^-b;  the upper 95% bound, 0.184, was also run.  Each lineage grows exponentially from N0 at Babel to Na at its texts, with Na of 10,000, 100,000 or a million and Nc of 100,000, 360,000 or a million, and the multiplier is capped at Bromham's smallest population, 200.  For exponential growth over T years the effective length is (N0/Nc)^-b x T x (1 - (Na/N0)^-b) / (b ln(Na/N0)), and the code checks this closed form against numerical integration to one part in a million.  The central case, Na = 100,000 and Nc = 360,000, gives a mean multiplier of 1.76 for 160 founders, 1.99 for 6.4 and 1.55 for 1,042.  Hittite and Vedic then need, under the Masoretic date, 2.45 / 1.76 = 1.39 at one rate and 1.97 with gamma;  under the Septuagint, 0.87 and 1.22;  with the 550-year shift under the Masoretic date, 0.85 and 1.20, and under the Septuagint 0.62 and 0.88.  If 21% of divergence comes in bursts at splits, the time-dependent part falls by up to a factor of 0.79, giving 1.10 and 1.56 for the Masoretic case and 0.69 and 0.96 for the Septuagint.  Over all the scale assumptions, with 160 founders and the Masoretic date, the residual runs 1.09 to 1.83 at one rate and 1.54 to 2.58 with gamma;  holding every lineage at 160 people for the whole interval gives 0.86 to 1.13 and 1.21 to 1.60.  Code babel-ie.ts;  output ie-babel-popgrowth.tsv, 5,040 rows. ↩ ↩

  20. 20.

    abvd.ts on the ABVD38;  output abvd-results.tsv.  Calibration:  the 28 sister pairs among eight East Polynesian lists, dated AD 1950, with each split put at AD 1250, the later settlement bound, so that each lineage has run 0.7 millennia and the rates come out as fast as possible.  The fastest pair, Tahitian and Marquesan, shares S = 0.575, so r = 0.575^(1/1.4) = 0.673;  the median of the 28 is 0.795.  Root:  in Glottolog 5.3, Atayalic and Malayo-Polynesian are both primary branches of Austronesian, and the 56 Atayalic and Malayo-Polynesian pairs have a median S of 0.098.  At the fastest rate the two lineages need ln 0.098 / ln 0.673 = 5.9 millennia, which puts Proto-Austronesian at about 990 BC;  a Masoretic Babel allows 2 x (1950 + 2242) = 8,384 years, so k = 0.70.  At the median rate it is 10.1 millennia, 3110 BC, k = 1.21.  With the gamma correction it is 4140 BC on the disjoint shape and 10550 BC on the all-pairs shape.  Three 18th-century lists from Cook's first voyage, which give absurd rates of 0.03 to 0.22 a millennium against the modern lists of the same languages, were excluded as list noise. ↩ ↩

  21. 21.

    The Cambridge Encyclopedia of the World's Ancient Languages.  Cambridge University Press.  Edited by Roger D. Woodard.  2004.  The standard reference on the languages of the ancient world, one chapter per language by a specialist:  Piotr Michalowski on Sumerian, Gernot Wilhelm on Hurrian, John Huehnergard and Christopher Woods on Akkadian and Eblaite, Calvert Watkins on Hittite, Woodard on Greek.  Each states when its language is first attested and how securely.

    1. 21a.

      Page 20:  “We have to wait for about four hundred years for our next archaic texts from Ur, dated approximately 2800 BC … the notation of a few morphological elements and phonetic glosses provides convincing evidence that the language of the texts is indeed Sumerian.”  “The next larger groups of texts from Sumer are the Early Dynastic III texts from Fara (ancient Shuruppak), Abu Salabikh, Nippur, and Adab from around 2500 BC. … now for the first time we have evidence for an extensive written poetic tradition.”  From Michalowski's chapter on Sumerian. ↩

  22. 22.

    “Visible Language:  Inventions of Writing in the Ancient Middle East and Beyond.”  Oriental Institute Museum Publications 32.  Oriental Institute of the University of Chicago.  Edited by Christopher Woods.  2010.  The catalogue of a museum exhibition on the first writing systems, with chapters by specialists:  Woods on Mesopotamia, Elise V. MacArthur on Egypt, Edward L. Shaughnessy on China.  It separates the first writing from the first text securely in a known language, which is the distinction the attestation count turns on.

    1. 22a.

      Page 119:  “Tomb U-j, which dates to ca. 3320 bc, is a rectangular, mudbrick tomb with twelve interconnected chambers”.  From MacArthur's chapter on the Egyptian writing system. ↩

  23. 23.

    Morris, Henry M.  The Genesis Record:  A Scientific and Devotional Commentary on the Book of Beginnings.  Creation-Life and Baker.  1976.  The founding young-earth commentary on Genesis.  It treats the confusion of tongues as a miracle that made new phonologies, ties their number loosely to the seventy nations of Genesis 10, and dates the dispersion to about 2358 BC while allowing that gaps in Genesis 11 could add a thousand years or more.  It cites no linguist.

    1. 23a.

      Page 267:  “The whole population was said originally to have been 'of one language and one speech' (or, literally, 'of one lip and one set of words,' apparently a reference to one phonology and one vocabulary).” ↩

    2. 23b.

      Page 276:  “The confounding of languages applied only to the phonologies, not to the underlying thought processes which are part of man's uniqueness.” ↩

    3. 23c.

      Page 277:  “Likewise, another miracle of creation was required to create many new such phonologies all at once (presumably leaving the old intact, in the brain-nerve-tongue complex of those who did not participate in the Babel rebellion).” ↩

    4. 23d.

      Page 262:  “distinct family units, distinct languages (therefore, subsequent to Babel), distinct regions, and, finally, actual nations. A total of twenty-six such 'nations' is listed as coming from Shem, as compared with thirty listed from Ham … and only fourteen from Japheth. Thus a total of seventy such primeval nations is listed”. ↩

    5. 23e.

      Page 263:  “There may well have been others, as the listings seem quite uneven in a number of cases”. ↩

    6. 23f.

      Page 282:  “Adding up the numbers in these verses, therefore, the record indicates that Peleg was born 101 years after the Flood. This probably was the year of the Babel Dispersion.” ↩

    7. 23g.

      Page 309:  “If Thiele's analysis is correct, this would place the Flood (assuming no gaps in Genesis 11) at 2459 B.C. and the Dispersion at about 2358 B.C.” ↩

    8. 23h.

      Pages 309–310:  “If gaps are allowed, however, these dates might conceivably be increased by, say, up to about one thousand years or more (possibly five thousand years at the very outside).” ↩

  24. 24.

    Taylor, Charles V.  “What happened at Babel?”  Creation 6(2):20–23.  Nov 1983.  A popular article by a young-earth linguist that weighs four ways God might have confused language at Babel and prefers a temporary imposed deafness.  It doubts that God created new languages, which sets it apart from every other young-earth writer here.

    1. 24a.

      Section Distorted receiver:  “However, the Genesis text does not go as far as saying God gave the people new languages; it merely says that 'God confused their language'. So I suspect that this suggestion goes a little too far to be correct.” ↩

    2. 24b.

      Section Distorted receiver:  “The third possibility is that God rendered the people at Babel temporarily deaf (partially or totally) for several years.”  “An imposed deafness is my choice, as a linguist, of the most likely natural mechanism which God could have used to cause the confusion at Babel.” ↩

  25. 25.

    Taylor, Charles V.  “The origin of language.”  Journal of Creation 11(1):76–81.  Apr 1997.  A young-earth linguist's paper on the origin of language in Adam, which separates that miracle from the division of languages at Babel.  On Babel it says only that Nostratic was proposed to avoid the miracle, a motive it gives no source for.

    1. 25a.

      Section Opening paragraph:  “the Bible indicates that there were two distinct miraculous events: the original creation of Adam as a talking and understanding being; and the subsequent division of humanity into language groups as a judgment on the rebellion of the descendants of Noah.” ↩

    2. 25b.

      Section From animals to humans?:  “In trying to reduce the number of distinct language families (and so avoid the miracle of Babel), Shevoroshkin introduced the label 'Nostratic' for the 'reconstruction' of a protolanguage linking five or six major language families. … But this is extremely speculative, and depends on the researcher's individual semantic interpretations.” ↩

  26. 26.

    Ham, Ken.  “How Many Languages?”  Answers with Ken Ham, volume 145.  Answers in Genesis.  Apr 20, 2022.  The transcript of a short radio feature by the founder of Answers in Genesis.  It sets about 70 families leaving Babel beside about one hundred original families counted by researchers, and gives no source for either number.

    1. 26a.

      Unpaginated radio transcript.  “The Bible records about 70 or so 'language families' leaving Babel. Now, there are nearly seven thousand languages spoken today. … Researchers also believe there's only about one hundred original language families!” ↩

  27. 27.

    Adamthwaite, Murray.  “The languages of Babel.”  Creation 42(1):52–55.  Nov 11, 2019.  A popular article by a young-earth writer with a doctorate in ancient Near Eastern studies, the only young-earth piece found that uses the written record of languages.  It argues that many unrelated, complex languages appear in the Near East at about the same time, the mid-third millennium BC, as Babel predicts.

    1. 27a.

      Section The first phenomenon:  “Two phenomena stand out; first, all these languages appear on the scene at about the same time—the mid-third millennium BC on conventional chronology. For secular scholars, their origins remain obscure.” ↩

    2. 27b.

      Section A polyglot of unrelated languages:  “Indeed, what is striking about the linguistic landscape of the Ancient Near East is the plethora of unrelated but highly complex languages. … Rather, what we find is precisely what we would expect from the story of the Tower of Babel.” ↩

    3. 27c.

      Section From Shin'ar to the whole world:  “the evidence is instead consistent with what Genesis describes: a sudden, supernatural act of God which created a whole suite of unrelated but highly complex languages. The result was the array of languages in the Near East, as cited above, which all appear at the same time—the second half of the third millennium BC.” ↩

    4. 27d.

      Section References and Notes, note 15:  “Care must be taken to distinguish the (later) written attestation of a language from its actual (earlier) existence as a lingua franca, which can be a gap of centuries. This is particularly so with Etruscan, and possibly Elamite.” ↩

  28. 28.

    Tomkins, Jeffrey P.  “Out of Babel, Not Africa.”  Institute for Creation Research.  Feb 16, 2015.  A short news piece from the Institute for Creation Research on Creanza and colleagues' 2015 comparison of phoneme and genetic variation.  It is the only young-earth use found of the phoneme-diversity debate, and it reads the absence of a phoneme founder effect as evidence for separate origins at Babel.

    1. 28a.

      Paragraph 1:  “Newly published research combining genetic, language, and demographic data challenges the idea of a single lineage of languages and human populations evolving out of Africa. Instead, the data supports the idea that multiple people groups have independent origins—a condition one would predict if the confusion of languages at the Tower of Babel happened as described in the Bible.” ↩

    2. 28b.

      Paragraph 5:  “The researchers noted, "However, the geographic distribution of phoneme inventory sizes does not follow the predictions of a serial founder effect [single line of descent] during human expansion out of Africa."”  The bracketed words are Tomkins's insertion into Creanza and colleagues' sentence. ↩

  29. 29.

    Sibley, Andrew.  “Dating the Tower of Babel events with reference to Peleg and Joktan.”  Journal of Creation 31(1):80–87.  2017.  The one young-earth journal paper devoted to the date of Babel.  It weighs the Masoretic, Samaritan, Septuagint and Jubilees chronologies against the early Jewish commentaries and prefers a Babel 340 years after the Flood, at Peleg's death, against the popular 101 years.

    1. 29a.

      Section Abstract:  “The preferred solution of this paper is to follow the Masoretic Text and the Seder Olam Rabbah commentary that places the Babel event 340 years post-Flood at Peleg's death.” ↩

  30. 30.

    Cox, Gavin.  “Do claimed dates for Gobekli Tepe and the Lost City of Dwarka outdate the Bible's timeline?”  Creation Ministries International.  Dec 23, 2021.  A reply by a Creation Ministries International writer to a reader's letter about two sites dated older than the biblical timeline.  It places Babel both "c. 140–200 years after the Flood" and "around 2,500 BC".

    1. 30a.

      Section Dwarka dating issues:  “We would place the rise of these nomadic Steppe peoples several centuries to a millennium after the Tower of Babel, which occurred around 2,500 BC”. ↩

  31. 31.

    Tweedy, E. J.  “Oard's Ice Age and Settlement of Northern Europe on Masoretic and Septuagint Timelines.”  Answers Research Journal 17:763–783.  2024.  A young-earth paper that sets Michael Oard's post-Flood Ice Age model against both the Masoretic and the Septuagint chronologies of Genesis 11, and so dates the dispersion on each.

    1. 31a.

      Page 766:  “For simplicity, we will use the LXX date of Peleg's birth (2767 BC) to represent the Dispersion on the LXX timeline, which is an arbitrary assumption based on Genesis 10:25.” ↩

  32. 32.

    Hammarström, Harald et al.  Glottolog 5.3.  Max Planck Institute for Evolutionary Anthropology.  2026.  The standard open catalogue and classification of the world's languages, released with its data tables.  It joins languages into a family only where the comparative method shows common descent, following Campbell and Poser's handbook, and counts everything else as an isolate, so it is the conservative count.  The family and isolate totals in this article are counted from its tables.  Mirror:  Github.com ↩

    1. 32a.

      Section Glottolog information, Languoids:  “Currently 2026-03-02 there are 7674 spoken L1 languages (i.e. spoken languages traditionally used by a community of speakers as their first language).  Languages are classified (see below) into 246 families and 183 isolates, i.e., one-member families.” ↩

    2. 32b.

      Section Glottolog information, Languoids:  “This classification is the best guess by the Glottolog editors … Users should be aware that for many groups of languages, there is little available historical-comparative research, so the classifications are subject to change as scholarship and interest in those languages increase.” ↩

    3. 32c.

      Section Glottolog information, Classification, step 6:  “If there are similarities to other language(s) that can be shown exceed chance, there are three possible kinds of explanations: universals, contact or inheritance from a common ancestor (Campbell, Lyle and Poser, William J. 2008). If the best explanation for the similarities are inheritance from common ancestor, languages are classified as belonging to the same family. A language which, by this principle, does not belong to the same family as any other language is also called an isolate.” ↩

  33. 33.

    “What are the largest language families?”  Ethnologue:  Languages of the World, 29th edition.  SIL Global.  Edited by David M. Eberhard, Gary F. Simons and Alison J. Robinson.  2026.  A short summary page of the most widely used catalogue of living languages.  Its family count keeps large groupings that Glottolog does not accept as demonstrated, such as Niger-Congo and Trans-New Guinea, and lists isolates apart from families, so its number is a different unit from Glottolog's.

    1. 33a.

      Paragraph 1–2:  “The six largest language families by language count are Niger-Congo, Austronesian, Trans-New Guinea, Sino-Tibetan, Indo-European, and Afro-Asiatic.  Of the 143 different language families, these six stand out as the major language families of the world.” ↩

  34. 34.

    Renfrew, Colin.  “The Origins of World Linguistic Diversity:  An Archaeological Perspective.”  The Origin and Diversification of Language, Memoirs of the California Academy of Sciences 24, 171–192.  California Academy of Sciences.  Edited by Nina G. Jablonski and Leslie C. Aiello.  1998.  An archaeologist's account of why language families cluster where they do.  It takes over Nichols's contrast between spread zones, where one family covers a large area, and residual zones, where many families survive, and argues that most spread-zone families are farming dispersals of the last 10,000 years.  The ages it gives residual zones come from the radiocarbon chronology of settlement, not from linguistic measurement.

    1. 34a.

      Page 182:  “When a language family displays the 'spread zone' pattern it displays what the linguist may term a low genetic density (i.e., a limited number of linguistic units unrelated by descent from a common language ancestor) over a sizable area, and a relatively shallow time depth. The 'residual zone' pattern shows more language families, greater linguistic diversity within the language family, and greater antiquity of the linguistic stocks there. The Caucasus is a good example of a 'residual zone,' and is sometimes regarded as a linguistic refugium.” ↩

    2. 34b.

      Page 171:  “It is argued, however, that the most significant process underlying the 'spread zone' versus 'residual zone' pattern evident in the modern geographical distribution of language families was agricultural dispersal, with a time depth no greater than about ten thousand years.”  Page 183:  “All the 'spread zone' distributions are the product of dispersals within the past 10,000 years (Class B in Table 1). The most significant cases … are indeed farming dispersals.” ↩

    3. 34c.

      Page 183:  “All the 'residual zone' language distributions have been in place for at least 10,000 years, since before the end of the last glaciation (Class A in Table 1).” ↩

  35. 35.

    Ruhlen, Merritt.  A Guide to the World's Languages, Volume 1:  Classification.  Stanford University Press.  1987.  A survey of the world's languages by Greenberg's principal collaborator, which sets out Greenberg's lumping classification of all languages into about fifteen families.  It is the source of the low family counts the young-earth writers use, and it argues for links between those families.

    1. 35a.

      Page 260:  “Greenberg (1987) classifies the world's languages into the 15 families shown in List 7.3”. ↩

    2. 35b.

      Page 261:  “there are indications that this stone wall of opposition is beginning to crumble, and future consolidation of some of the groups in List 7.3 is likely”. ↩

  36. 36.

    Chang, Will et al.  “Ancestry-constrained phylogenetic analysis supports the Indo-European steppe hypothesis.”  Language 91(1):194–244.  2015.  A Bayesian dating of Indo-European that forces eight attested ancient languages, such as Latin and Vedic Sanskrit, to be the ancestors of their 39 modern descendants, so the rate of change is learned from observed change.  Calibrated on dated texts only, it gets a root about 5,950 years before AD 2000;  on Bouckaert's data without the constraints it reproduces 7,580.  It is the clearest case of one modelling choice moving the root by 1,600 years.  Mirror:  Escholarship.org

    1. 36a.

      Page 220, figure 5:  “A1: Narrow dataset 5950 7190–4870”.  The median root age, then the 95% HPD interval, in years before AD 2000. ↩

  37. 37.

    Heggarty, Paul et al.  IE-CoR:  Indo-European Cognate Relationships database, CLDF release.  2023.  The dataset behind Heggarty and colleagues' 2023 dating:  160 languages, 170 meanings and 25,741 cognate judgements by specialists, with 52 ancient languages dated by their texts and 1,036 cognate sets flagged as loans.  It is the database this article's divergence-time test measures. ↩

  38. 38.

    Greenhill, Simon J. et al.  Austronesian Basic Vocabulary Database, CLDF release.  2020.  The largest comparative word-list database for one family, 2,040 word lists over 210 meanings, with cognate judgements.  It is the database for this article's Austronesian test. ↩