Language Families and the Tower of Babel: the literature

This page lists every source worked up in the research for the article, with what it measured, who wrote it, which way it cuts, and how much of it was read. Sources the article does not cite are listed at the end with the reason.

How to read an entry

Direction. Whether the source supports, undercuts or complicates the young-earth model of Babel as the origin of language families, judged by what the source shows, not by what its author believes. Several mainstream sources are tagged supports, because the mainstream position that no family can be linked to another is the young-earth prediction too. Several young-earth sources are tagged undercuts.

Verified. full: the text was read directly, and every quotation the article takes from it was checked against that text with its page or section. partial: some of the text was read, as stated in the entry. abstract: only the abstract or record. citation: only the bibliographic entry.

Field	Values
`camp`	mainstream, creationist, old-earth-christian
`method`	measurement, model, dataset, inventory, review, commentary, handbook
`direction`	supports, undercuts, complicates, background
`verified`	full, partial, abstract, citation

The young-earth model in its own words

The Genesis Record

Morris, Henry M. The Genesis Record: A Scientific and Devotional Commentary on the Book of Beginnings. Creation-Life and Baker, 1976.

key: morris-1976c camp: creationist method: commentary direction: background verified: full

The founding young-earth commentary. The confusion was a miracle that made new phonologies, not new thought; the Table of Nations lists seventy nations with distinct languages "subsequent to Babel"; the dispersion is at Peleg's birth, about 2358 BC on Thiele's Abraham, with gaps allowed that could add "up to about one thousand years or more". Morris cites no linguist. Read by archive.org search-inside of genesisrecordsc00morr (printdisabled), printed page = leaf - 4, anchored on running heads four times.

Towering change

Wieland, Carl. "Towering change." Creation 22(1):22-26, December 1999. https://creation.com/towering-change

key: wieland-1999b camp: creationist method: commentary direction: complicates verified: full

The standard young-earth model: Babel made a few dozen "stem" languages, each the ancestor of one family, and there should be "no links between one 'family' and another". Its count of "some 8 to 20 (commonly 12 or 13)" families comes from the lumpers, not from any count the specialists accept, and it cites no linguist. Its no-links prediction matches the mainstream result. It also invents a "Sino-Asiatic" family of Chinese, Japanese and Korean that no classification has.

What happened at Babel?

Taylor, Charles V. "What happened at Babel?" Creation 6(2):20-23, November 1983. https://creation.com/babel-language

key: taylor-1983 camp: creationist method: commentary direction: complicates verified: full

A linguist's preference for "an imposed deafness" as the mechanism of the confusion, and a doubt that God made new languages at all. That mechanism predicts one re-sounded vocabulary, so cognates across the world, the opposite of the no-links prediction. His remark that small societies change more slowly is the reverse of the later findings of Nettle and Bromham. Locators by section heading of the web text.

The origin of language

Taylor, Charles V. "The origin of language." Journal of Creation 11(1):76-81, April 1997. https://creation.com/origin-of-language

key: taylor-1997 camp: creationist method: commentary direction: supports verified: full

Separates the creation of Adam as a speaker from the division of languages at Babel, and says Nostratic was introduced "to avoid the miracle of Babel", for which he gives no source; the label is older than the man he credits. His quotation of Roger Lewin's 1990 New Scientist piece was not checked at source and is in the quarantine file.

The Tower of Babel account affirmed by linguistics

Duursma, K. J. "The Tower of Babel account affirmed by linguistics." Journal of Creation 16(3):27-31, December 2002. https://creation.com/babel-linguistics

key: duursma-2002 camp: creationist method: commentary direction: complicates verified: full

The fullest young-earth linguistic case: about twenty families from Ruhlen, a dozen languages at Babel, a miracle rather than faster change, and glottochronology's rate used to set a 10,000-year horizon. Its two sentences on greater word loss in isolated post-Babel groups are the only young-earth proposal found of faster change, and the article's population test is built from them. Its citations were checked: the Ruhlen quotations are accurate but the same page predicts consolidation of the families; Crowley's rate is right but on p. 178, not p. 169; "His method has become the norm" is false. Robins's "3,000 BC" was not checked.

More Than PIE

Oller, John W. Jr. "More Than PIE." Answers 3(2), April-June 2008. https://answersingenesis.org/tower-of-babel/more-than-pie/

key: oller-2008 camp: creationist method: commentary direction: complicates verified: full

The most credentialed young-earth writer on the subject, a professional linguist. Gives 94 families from the 2005 Ethnologue, says the Babel languages will prove "radically distinct", that 4,000 years suffice at observed rates, and that families will never fall below the Genesis 10 names. The floor survives the real count; the timing claim contradicts his own report that secular linguists date Proto-Indo-European to the third to seventh millennium BC, and the divergence-time test refutes the rates claim for the Bronze Age pairs.

Tower of Babel: The Cultural History of Our Ancestors

Hodge, Bodie. Tower of Babel: The Cultural History of Our Ancestors. Master Books, 2013.

key: hodge-2013 camp: creationist method: commentary direction: complicates verified: partial

Babel made language families, at least 78 by the names of Genesis 10: the only young-earth count derived from the text, and a floor, not a total. Read only through Google Books search-inside snippets of volume sBE99WGQvpYC, which gave wording but no page numbers before rate-limiting; not on archive.org. The verifier (2026-10-10) could not reach the snippets again (Google's rate limit) and found a second copy of only one sentence, the 78-family floor, on Goodreads' quotation page for the book; the article now quotes only the words both routes share. The breakdown 14/39/25, the "grown", Greek-tongues and isolates passages are in the quarantine file (item 5), with the 2242 BC attribution (item 2).

How Many Languages? (2022)

Ham, Ken. "How Many Languages?" Answers with Ken Ham, volume 145, 20 April 2022. https://answersingenesis.org/media/audio/answers-with-ken-ham/volume-145/how-many-languages/

key: ham-2022b camp: creationist method: commentary direction: complicates verified: full

A radio transcript: about 70 families left Babel and researchers believe in "about one hundred original language families". No source is given for either number, and the hundred is wrong by a factor of about four.

How Many Languages? (2023)

Ham, Ken. "How Many Languages?" Answers with Ken Ham, volume 149, 26 April 2023. https://answersingenesis.org/media/audio/answers-with-ken-ham/volume-149/how-many-languages/

key: ham-2023d camp: creationist method: commentary direction: complicates verified: full

The same feature a year later: God created "about one hundred language families" at Babel, "and that's how many language families exist today". Inconsistent with the 2022 script and with Hodge's floor of 78, and wrong against every count.

The languages of Babel

Adamthwaite, Murray. "The languages of Babel." Creation 42(1):52-55, January 2020; web 11 November 2019. https://creation.com/babel-languages

key: adamthwaite-2019 camp: creationist method: commentary direction: complicates verified: full

The only young-earth writer found who uses the written record: many unrelated, complex languages "appear on the scene at about the same time", the mid-third millennium BC. On its own conventional chronology the first Egyptian and Sumerian texts are 600 to 800 years earlier and Hattic, Kassite and Etruscan much later, so the "same time" spans about 2,500 years. It also relies on Barry Fell's Indus decipherment, which has no standing, and presents Hittite wātar and English water as a word that passed from Hittite into Greek and Latin, when the two are cognates inherited from Proto-Indo-European. Its footnote that attestation can lag a language by centuries is a fair point.

Out of Babel, Not Africa

Tomkins, Jeffrey P. "Out of Babel, Not Africa." Institute for Creation Research, 16 February 2015. https://www.icr.org/article/out-babel-not-africa

key: tomkins-2015e camp: creationist method: commentary direction: complicates verified: full

The only young-earth use found of the phoneme-diversity debate. It quotes Creanza and colleagues accurately but inserts the gloss "[single line of descent]" into their sentence, which changes what "serial founder effect" means, and draws an inference about independent origins that the paper does not make.

Speaking of Adam and Eve

Rana, Fazale. "Speaking of Adam and Eve: Study of Languages Supports Biblical Account of Human Origins." Reasons to Believe, 23 August 2011.

key: rana-2011 camp: old-earth-christian method: commentary direction: undercuts verified: full

An old-earth reading of Atkinson's phoneme gradient as confirming an African origin of humanity about 100,000 years ago, near Eden. It shows that the gradient's direction, if real, suits an old-earth model and not a Mesopotamian Babel. Written before the 2012 critiques.

A Better Model for the Stone Age, Part 2

Osgood, A. J. M. "A Better Model for the Stone Age, Part 2." Journal of Creation (then EN Tech. J.), 1988. https://creation.com/a-better-model-for-the-stone-age-part-2

key: osgood-1988 camp: creationist method: model direction: background verified: full

A revised ancient chronology that puts every culture outside lower Mesopotamia after Babel, about 2200 BC, and estimates about 25,000 people at Babel. The article uses that figure as one input to its founder-population test. Read through the sibling article neolithic-and-the-flood-date; creation.com prints no volume or pages for Part 2.

Modelling biblical human population growth

Carter, Robert; Hardy, Chris. "Modelling biblical human population growth." Journal of Creation 29(1):72-79, 2015.

key: carter-2015 camp: creationist method: model direction: background verified: full

An agent-based model of growth from Noah's family: more than 1,000 people a century after the Flood is possible under some settings, and up to 40,000 only in scenarios the authors call not likely. The lower figure is the smallest founder input in the population test. Read through the sibling article neolithic-and-the-flood-date.

Synchronization of the Biblical and Egyptian Timelines

Habermehl, Anne. "Synchronization of the Biblical and Egyptian Timelines." Answers Research Journal 16:561-576, 2023.

key: habermehl-2023 camp: creationist method: model direction: background verified: full

A young-earth Egyptian chronology that lowers the New Kingdom and the Late Bronze Age by about 450 to 550 years. It is the published young-earth scale for the article's 550-year shift of the Bronze Age texts; it does not discuss language, so that application is the article's own. Read through the sibling article egypt-and-the-flood-date and its dossier.

Dating the Tower of Babel events with reference to Peleg and Joktan

Sibley, Andrew. "Dating the Tower of Babel events with reference to Peleg and Joktan." Journal of Creation 31(1):80-87, 2017.

key: sibley-2017 camp: creationist method: commentary direction: background verified: full

The one young-earth journal paper on the date of Babel. It prefers the Masoretic Text with Seder Olam Rabbah, Babel 340 years after the Flood at Peleg's death, which is 2008 BC on Ussher's Flood. Cited in the parent's Part 1 to show the young-earth Babel dates run from about 2500 to 2008 BC, not only 2358 to 2242. Read in the library copy (sources/web/sibley-2017_dating-tower-of-babel-peleg-joktan.html), Abstract and Summary.

Do claimed dates for Gobekli Tepe and the Lost City of Dwarka outdate the Bible's timeline?

Cox, Gavin. Creation Ministries International, 23 December 2021.

key: cox-2021b camp: creationist method: commentary direction: background verified: full

A reply to a reader that places Babel both "c. 140–200 years after the Flood" and "around 2,500 BC", the earliest young-earth Babel date found in print. Cited in the parent's Part 1 for the range of Babel dates. Read in the library copy, section Dwarka dating issues.

Oard's Ice Age and Settlement of Northern Europe on Masoretic and Septuagint Timelines

Tweedy, E. J. Answers Research Journal 17:763-783, 2024.

key: tweedy-2024 camp: creationist method: model direction: background verified: partial

Sets Oard's post-Flood Ice Age against both chronologies and takes Peleg's birth on the Septuagint, 2767 BC, as the dispersion date, calling it "an arbitrary assumption" (p. 766). Cited in the parent's Part 1 as the young-earth writer who prints a Septuagint Babel. Read in the library PDF, pp. 765-766 (folios checked on pdf pages 1, 3, 5).

The Ark Landing, the Migration, and the Early Settling of Noah's Sons, Part 1

Osgood, A. John M. Answers Research Journal 16:501-528, 2023.

key: osgood-2023 camp: creationist method: model direction: background verified: full

A young-earth Mesopotamian chronology: the dispersion at the end of Ubaid 2, the Uruk culture as Cush and Nimrod (Enmerkar), Uruk IV writing as a post-Babel "reinvention", Hammurabi at "circa 1450 B.C.-revised". Cited in how-old-are-the-language-families Part 2 as one of the two revised chronologies that make all writing post-Babel. Read in the library copy, sections A Detailed Look at the Sons of Shem and The Uruk Period.

Unwrapping the Pharaohs

Ashton, John; Down, David. Unwrapping the Pharaohs: How Egyptian Archaeology Confirms the Biblical Timeline. Master Books, 2006.

key: ashton-2006-unwrapping-the-pharaohs camp: creationist method: model direction: background verified: full

A popular revised chronology that begins Egyptian history after Babel, about 2100 BC, with Menes in the 21st century BC. It is the young-earth answer to Egyptian writing before Babel; what that answer costs is argued in egypt-and-the-flood-date, which read the book by archive.org search-inside (unwrappingpharao0000asht, printdisabled, printed page = leaf - 4).

How many families there are

Glottolog 5.3

Hammarström, Harald; Forkel, Robert; Haspelmath, Martin; Bank, Sebastian. Glottolog 5.3. Max Planck Institute for Evolutionary Anthropology, 2026. doi:10.5281/zenodo.18840935. CLDF release v5.3, commit db1d557.

key: hammarstrom-2026 camp: mainstream method: dataset direction: complicates verified: full

The conservative, comparative-method-only classification. Counted from its tables with research/code/glottolog-counts.ts: 238 families of two or more languages and 183 isolates, 421 units, 314 with a living member. The site's "246 families" appears to include eight non-genealogical buckets. 421 is an upper bound on the number of origins, not a measurement of it, because the method can join languages but cannot prove them unrelated.

What are the largest language families?

Eberhard, David M.; Simons, Gary F.; Robinson, Alison J. (eds.). Ethnologue: Languages of the World, 29th edition. SIL Global, 2026. https://www.ethnologue.com/insights/largest-families/

key: eberhard-2026c camp: mainstream method: inventory direction: background verified: full

Gives 143 families. Its family browser shows why the number is not comparable with Glottolog's: it keeps Niger-Congo, Nilo-Saharan, Australian and Trans-New Guinea as single families and lists 99 isolates in a separate bucket. The article cites it only to say so.

How Many Language Families are there in the World?

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. doi:10.1387/asju.20195.

key: campbell-2018 camp: mainstream method: inventory direction: complicates verified: full

An independent region-by-region count by the leading skeptic of long-range groupings: about 406 families, 162 of them isolates, 94 extinct. It agrees with Glottolog within 4%. Its definition makes the count one of what "cannot reliably be shown to be related", and it expects the number to move both ways. Printed page = PDF page + 131.

Linguistic diversity and language evolution

Hammarström, Harald. "Linguistic diversity and language evolution." Journal of Language Evolution 1(1):19-29, 2016. doi:10.1093/jole/lzw002.

key: hammarstrom-2016 camp: mainstream method: review direction: complicates verified: full

Glottolog's editor counts 424 families and isolates by macro-area and states the comparative method's horizon, "expected not to exceed 10,000 years", and why it cannot be pushed back with basic vocabulary. He also notes that family density does not follow the age of settlement, which cuts against any "more families means more time" argument on either side.

The Origin and Dispersal of Languages: Linguistic Evidence

Nichols, Johanna. In Jablonski, Nina G. and Leslie C. Aiello (eds.), The Origin and Diversification of Language, Memoirs of the California Academy of Sciences 24, pp. 127-170, 1998.

key: nichols-1998 camp: mainstream method: model direction: undercuts verified: full

The strongest mainstream argument from the count: oldest stocks about 6,000 years, 7% shared cognates after 6,000 years as the floor of significance, and a branching model in which one ancestor needs at least 72,000 to 132,000 years to yield the world's stocks. Every input is calibrated on the mainstream ages of known families, so the model is consistent but not independent of the dating question, and Nichols herself reads its 48,000-year age for the Americas as evidence of several entries rather than as a date. Read by archive.org search-inside (origindiversific0000paul), printed page = leaf - 12; Table 3 checked against the page image.

The Origins of World Linguistic Diversity: An Archaeological Perspective

Renfrew, Colin. In Jablonski and Aiello (eds.), The Origin and Diversification of Language, pp. 171-192, 1998.

key: renfrew-1998 camp: mainstream method: review direction: complicates verified: full

Spread zones and residual zones: most spread-zone families are farming dispersals of the last 10,000 years, and residual zones keep many families. The pattern is real and visible in the Glottolog count, but it is neutral on absolute time, and its "at least 10,000 years" for residual zones comes from the radiocarbon chronology of settlement.

A Guide to the World's Languages, Volume 1: Classification

Ruhlen, Merritt. A Guide to the World's Languages, Volume 1: Classification. Stanford University Press, 1987.

key: ruhlen-1987 camp: mainstream method: inventory direction: complicates verified: partial

The source of Duursma's family count: Greenberg's 15 families in List 7.3, and on the next page the prediction that "future consolidation of some of the groups in List 7.3 is likely". The low young-earth counts come from the same lumping method that claims links between families. Read by archive.org search-inside of guidetoworldslan0000ruhl, pp. 260-261 only.

How old the families are

Language trees with sampled ancestors

Heggarty, Paul, et al. "Language trees with sampled ancestors support a hybrid model for the origin of Indo-European languages." Science 381:eabg0818, 2023. doi:10.1126/science.abg0818.

key: heggarty-2023b camp: mainstream method: model direction: undercuts verified: full

Indo-European root 8,120 years before AD 2000 on 161 languages and 52 text-dated ancient languages, with the data rather than the analyst deciding which ancient languages are ancestors. Its own sensitivity runs give the young-earth reader's best mainstream number, 4,997 years on the older IELex data with forced ancestry, which the authors attribute to excess synonyms inflating rates. Read first in the author manuscript, then in the published version with the project's AAAS account; the article quotes the published wording by section.

Mapping the origins and expansion of the Indo-European language family

Bouckaert, Remco, et al. Science 337:957-960, 2012. doi:10.1126/science.1219669.

key: bouckaert-2012 camp: mainstream method: model direction: undercuts verified: full

Root 7,116 to 10,410 years ago on 103 languages, calibrated on 20 dated ancient languages and 14 dated splits, all from historical sources, the oldest Hittite at about 3,450 years. The 2013 correction was not read. Read in the PMC author manuscript and its supplement.

Language-tree divergence times support the Anatolian theory of Indo-European origin

Gray, Russell D.; Atkinson, Quentin D. Nature 426:435-439, 2003. doi:10.1038/nature02029.

key: gray-2003 camp: mainstream method: model direction: undercuts verified: full

Root 7,800 to 9,800 years ago, 14 historical calibrations, and a leave-one-out test that recovers 13 of 14 within 390 years: the best evidence that rates are stable inside the window the texts cover, though not beyond it. Its opening quotation of Dixon is misapplied to Indo-European; see the quarantine file. Read in the accepted manuscript; the article cites it by section.

Ancestry-constrained phylogenetic analysis supports the Indo-European steppe hypothesis

Chang, Will; Cathcart, Chundra; Hall, David; Garrett, Andrew. Language 91(1):194-244, 2015. doi:10.1353/lan.2015.0005.

key: chang-2015 camp: mainstream method: model direction: undercuts verified: full

Root 5,950 years with eight attested languages forced to be ancestors, against 7,580 on the same kind of data without that constraint: one modelling choice moves the root 1,600 years. Its lower bound is still about 600 years before a Masoretic Babel. Its ancestry constraints are exactly the calibrations a young-earth chronologist accepts.

Dated language phylogenies shed light on the ancestry of Sino-Tibetan

Sagart, Laurent, et al. PNAS 116(21):10317-10322, 2019. doi:10.1073/pnas.1817972116.

key: sagart-2019 camp: mainstream method: model direction: undercuts verified: full

Root 7,184 years on four text calibrations no younger chronology disputes, all within 2,800 years; the lower bound, 5,093, is older than any Masoretic Babel. The cleanest case of a text-only extrapolation, so the young-earth reply has to be about the model, not radiocarbon.

Phylogenetic evidence for Sino-Tibetan origin in northern China: Supplementary Information

Zhang, Menghan; Yan, Shi; Pan, Wuyun; Jin, Li. Nature 569:112-115, 2019, Supplementary Information. doi:10.1038/s41586-019-1153-z.

key: zhang-2019b camp: mainstream method: model direction: complicates verified: partial

Root 5,871 years before 1950 on eight calibrations from written history and recorded migrations. Its lower bound, 4,183 years before 1950, is the only dated family root found that reaches into a Masoretic window, by nine years. The article is paywalled; the free supplement was read and is the key cited.

Dated phylogeny suggests early Neolithic origin of Sino-Tibetan languages

Zhang, Hanzhi; Ji, Ting; Pagel, Mark; Mace, Ruth. Scientific Reports 10:20792, 2020. doi:10.1038/s41598-020-77404-4.

key: zhang-2020d camp: mainstream method: model direction: undercuts verified: full

A third Sino-Tibetan dating, 7,983 years on 131 languages and six text calibrations. The three Sino-Tibetan studies differ by 2,100 years in the mean, which measures their model and sample dependence.

Bayesian phylogenetic analysis of Semitic languages

Kitchen, Andrew; Ehret, Christopher; Assefa, Shiferaw; Mulligan, Connie J. Proceedings of the Royal Society B 276:2703-2710, 2009. doi:10.1098/rspb.2009.0408.

key: kitchen-2009 camp: mainstream method: model direction: undercuts verified: full

Semitic root 5,750 years, with its floor fixed at the first Akkadian writing, 4,350 years. Semitic is one branch of Afroasiatic, so this is a minimum for the family. The Akkadian anchor itself sits before or at a 2242 BC dispersion.

The Civilizations of Africa

Ehret, Christopher. The Civilizations of Africa: A History to 1800. University Press of Virginia, 2002.

key: ehret-2002 camp: mainstream method: review direction: complicates verified: full

Afroasiatic's first split "sometime before 11,000 BCE", tied by reconstructed grass-processing words to the radiocarbon-dated Cataract Tradition. The date is an archaeological correlation, so it carries no weight against Babel independent of radiocarbon. Read by archive.org search-inside (civilizationsofa0000ehre), printed page = leaf - 16.

The Prehistory of a Dispersal: the Proto-Afrasian (Afroasiatic) Farming Lexicon

Militarev, Alexander. In Bellwood, Peter and Colin Renfrew (eds.), Examining the Farming/Language Dispersal Hypothesis, pp. 135-150. McDonald Institute, 2002.

key: militarev-2002-prehistory-of-a-dispersal-proto-afrasian camp: mainstream method: model direction: complicates verified: full

Proto-Afroasiatic in the ninth millennium BC by lexicostatistics, tied to the Natufian. The method is glottochronological and the homeland date rests on radiocarbon; the author belongs to the Moscow long-range school. Read by archive.org search-inside (examiningfarming0000unse), printed page = leaf - 16.

Cultural and climatic changes shape the evolutionary history of the Uralic languages

Honkola, T., et al. Journal of Evolutionary Biology 26(6):1244-1253, 2013. doi:10.1111/jeb.12107.

key: honkola-2013 camp: mainstream method: model direction: complicates verified: full

Uralic root about 5,300 years on three calibrations from loanword strata, history and archaeology, none a text of the family; the lower bound, 3,330 years, is well after any Babel date.

A Bayesian phylogenetic study of the Dravidian language family

Kolipakam, Vishnupriya, et al. Royal Society Open Science 5:171504, 2018. doi:10.1098/rsos.171504.

key: kolipakam-2018 camp: mainstream method: model direction: supports verified: full

Dravidian root a median 4,433 years before AD 2000 on inscriptional and literary calibrations, overlapping the Masoretic window better than any other family. One family younger than Babel says nothing about how the 421 arose.

Bayesian phylogenetic analysis supports an agricultural origin of Japonic languages

Lee, Sean; Hasegawa, Toshikazu. Proceedings of the Royal Society B 278:3662-3669, 2011. doi:10.1098/rspb.2011.0518.

key: lee-2011 camp: mainstream method: model direction: supports verified: full

Japonic root about 2,182 years on historical calibrations, matching Yayoi farming. A useful control: the same method that makes Sino-Tibetan 7,000 years old makes Japonic young.

Bayesian phylolinguistics infers the internal structure and the time-depth of the Turkic language family

Savelyev, Alexander; Robbeets, Martine. Journal of Language Evolution 5(1):39-53, 2020. doi:10.1093/jole/lzz010.

key: savelyev-2020 camp: mainstream method: model direction: supports verified: full

Turkic root about 2,066 years on two text tips, where Turkologists put it on other grounds: the method works over short extrapolations, which cuts both ways.

Triangulation supports agricultural spread of the Transeurasian languages

Robbeets, Martine, et al. Nature 599:616-621, 2021. doi:10.1038/s41586-021-04108-8.

key: robbeets-2021 camp: mainstream method: model direction: complicates verified: full

A 9,181-year root for a "Transeurasian" grouping that Glottolog and most specialists do not accept, calibrated on linguists' estimates of proto-language ages rather than texts. Weak evidence either way.

Language phylogenies reveal expansion pulses and pauses in Pacific settlement

Gray, Russell D.; Drummond, Alexei J.; Greenhill, Simon J. Science 323:479-483, 2009. doi:10.1126/science.1166858.

key: gray-2009 camp: mainstream method: model direction: complicates verified: full

Austronesian root about 5,230 years on 400 languages, calibrated on settlement archaeology and four dated languages. The supplement with the full calibration list was not obtained.

How accurate and robust are the phylogenetic estimates of Austronesian language relationships?

Greenhill, Simon J.; Drummond, Alexei J.; Gray, Russell D. PLoS ONE 5(3):e9573, 2010. doi:10.1371/journal.pone.0009573.

key: greenhill-2010b camp: mainstream method: model direction: complicates verified: full

Re-dated on the five least contested calibrations, the root barely moves, to 5,117 years. Its one deep calibration, Proto-Oceanic, is the radiocarbon-dated Lapita horizon.

Bantu expansion shows that habitat alters the route and pace of human dispersals

Grollemund, Rebecca, et al. PNAS 112(43):13296-13301, 2015. doi:10.1073/pnas.1503793112.

key: grollemund-2015 camp: mainstream method: model direction: background verified: full

Bantu root about 6,900 years, all four calibrations radiocarbon-dated archaeological sites: language dates that are radiocarbon dates by another name.

Phylogeographic analysis of the Bantu language expansion supports a rainforest route

Koile, Ezequiel, et al. PNAS 119(32):e2112853119, 2022. doi:10.1073/pnas.2112853119.

key: koile-2022 camp: mainstream method: model direction: background verified: full

Bantoid root a median 5,110 years, reusing Grollemund's calibrations, so not an independent test.

The origin and expansion of Pama-Nyungan languages across Australia

Bouckaert, Remco R.; Bowern, Claire; Atkinson, Quentin D. Nature Ecology & Evolution 2:741-749, 2018. doi:10.1038/s41559-018-0489-3.

key: bouckaert-2018 camp: mainstream method: model direction: background verified: full

Pama-Nyungan, about 250 of Australia's 355 languages, is mid-Holocene, mean 5,671 years, on one archaeological calibration that the authors say the dates scale with. Most of Australia's present diversity is far younger than its settlement, which shows how spreads erase older families.

A Recent Northern Origin for the Uto-Aztecan Family

Greenhill, Simon J., et al. SocArXiv preprint, 2022. doi:10.31235/osf.io/k598j. Published in Language 99(1), 2023.

key: greenhill-2022 camp: mainstream method: model direction: supports verified: full

Uto-Aztecan about 4,100 years, within or below a Masoretic window, on calibrations that are almost all radiocarbon archaeology and plant distributions. The preprint was read; the published version sat behind a captcha on a site that is not one of the project's accounts, so the preprint pages are cited.

Automated Dating of the World's Language Families Based on Lexical Similarity

Holman, Eric W., et al. Current Anthropology 52(6):841-875, 2011. doi:10.1086/662127.

key: holman-2011 camp: mainstream method: model direction: complicates verified: full

ASJP dates for nearly every family from 40-word lists on 52 calibrations, with a mean error of 29%. Every calibration older than 2,500 years is archaeological, so the deep part of the clock is not independent of radiocarbon. Its Indo-European, 4,348 years, is thousands of years younger than the Bayesian figures, and Koen Bostoen's published comment says its Niger-Congo date is too shallow. Read from a Wayback copy of the Zurich repository PDF.

The Rise and Fall of Languages

Dixon, R. M. W. The Rise and Fall of Languages. Cambridge University Press, 1997.

key: dixon-1997 camp: mainstream method: commentary direction: supports verified: full

A field linguist's statement that the rate measured over the documented 3,000 years of Indo-Aryan says nothing certain about the undocumented part, and that the change "could have been rapid ... or slow". It cuts both ways: Dixon's own example is an older Proto-Indo-European, 10,500 years. His "4,000 years BP or 40,000 years BP" is about Australian, not Indo-European. Read by archive.org search-inside (risefalloflangua0000dixo), leaf = page + 8.

One origin or several

On Calculating the Factor of Chance in Language Comparison

Ringe, Donald A. Jr. Transactions of the American Philosophical Society 82(1), 1992. doi:10.2307/1006563.

key: ringe-1992 camp: mainstream method: model direction: complicates verified: full

Binomial computation of chance resemblance between unrelated word lists. Real families exceed chance and unrelated languages do not, and multilateral comparison finds systematic resemblances even among artificial lists. Its first sentence states the mainstream position: all languages are likely related, but it cannot be shown. Statistics of this kind cannot tell separate creation from undetectable common descent, because both predict chance-level resemblance between families.

Language Classification: History and Method

Campbell, Lyle; Poser, William J. Language Classification: History and Method. Cambridge University Press, 2008.

key: campbell-2008 camp: mainstream method: handbook direction: complicates verified: full

The methods handbook Glottolog follows. It rejects Nostratic, Eurasiatic, Amerind, Altaic and Proto-World, quotes Ringe 1999 on the chance matches multilateral comparison produces, puts the oldest established family at about 8,000 years, and says one origin is possible but can never be shown. Its note 2 on p. 365 records creationists citing Proto-World as support for Babel, which leans on the very claims the mainstream rejects. Read by archive.org search-inside (languageclassifi0000camp), printed page = leaf - 14, anchored on running heads four times.

An Introduction to Historical Linguistics

Crowley, Terry. An Introduction to Historical Linguistics, 2nd edition. Oxford University Press, 1992.

key: crowley-1992 camp: mainstream method: handbook direction: complicates verified: full

The textbook Duursma cites. It puts the comparative method's horizon at 5,000 to 10,000 years because shared vocabulary falls to a small proportion, and its exercise on Tok Pisin, 146 of 200 items still English after about a century, shows how fast a creole replaces vocabulary and how much of it survives. The Tok Pisin count has no stated source. Read by archive.org search-inside, printed page = leaf - 2.

The Origin of Language: Tracing the Evolution of the Mother Tongue

Ruhlen, Merritt. Wiley, 1994.

key: ruhlen-1994-origin-of-language camp: mainstream method: commentary direction: complicates verified: full

The main argument for Proto-World, on evolutionary premises, from multilateral comparison. Ruhlen admits the proposals are "almost universally condemned". A young-earth writer who cites him for one original language inherits his method and contradicts the no-links prediction. Read by archive.org search-inside (originoflanguage00merr), printed page = leaf - 14.

Global Etymologies

Bengtson, John D.; Ruhlen, Merritt. In Ruhlen, On the Origin of Languages: Studies in Linguistic Taxonomy, pp. 277 ff. Stanford University Press, 1994.

key: bengtson-1994 camp: mainstream method: commentary direction: complicates verified: full

Proto-World in its proponents' words: resemblances across six or more of 32 groupings, whose chance probability they call "close to zero". Ringe's figure, quoted by Campbell and Poser, gives a better than 90% chance of such a match by chance on 23 lists. Read by archive.org search-inside (onoriginoflangua0000ruhl), leaf = page + 18; the chapter's last page was not checked.

Ultraconserved words point to deep language ancestry across Eurasia

Pagel, Mark; Atkinson, Quentin D.; Calude, Andreea S.; Meade, Andrew. PNAS 110(21):8471-8476, 2013. doi:10.1073/pnas.1218726110.

key: pagel-2013 camp: mainstream method: model direction: complicates verified: full

The strongest statistical attempt at a macro-family, Eurasiatic at about 14,450 years, using the Moscow "Tower of Babel" project's reconstructions. Also the citable source for word half-lives of 2,000 to 4,000 years and 10,000 to 20,000 for the slowest words.

Ultraconserved words and Eurasiatic? The 'faces in the fire' of language prehistory

Heggarty, Paul. PNAS 110(35):E3254, 2013. doi:10.1073/pnas.1309114110.

key: heggarty-2013 camp: mainstream method: commentary direction: supports verified: full

The reply: the reconstructions are "not actually data", the correlation is built in, and sound change limits detection. It supports the narrow young-earth point that no macro-family is demonstrated. Mahowald and Gibson's letter and the authors' reply in the same issue were not read.

Linguistic Phylogenies Support Back-Migration from Beringia to Asia

Sicoli, Mark A.; Holton, Gary. PLoS ONE 9(3):e91722, 2014. doi:10.1371/journal.pone.0091722.

key: sicoli-2014 camp: mainstream method: model direction: complicates verified: full

Takes Vajda's Dene-Yeniseian link as given and models the migration it implies. The link is the one cross-continental proposal specialists take seriously, received with "cautious optimism"; it shows the method can still join families, so 421 is not fixed.

Phonemic diversity supports a serial founder effect model of language expansion from Africa

Atkinson, Quentin D. Science 332(6027):346-349, 2011. doi:10.1126/science.1199295.

key: atkinson-2011 camp: mainstream method: model direction: complicates verified: full

Phoneme inventories in 504 languages shrink with distance from a best-fit origin in southern Africa. It argues for one origin, or a main one, which Babel also predicts; the dispute with a young-earth reader is about place and time, and the time depth comes from genetics, not phonemes.

Comment on Atkinson (Jaeger, Pontillo and Graff)

Jaeger, T. Florian; Pontillo, Daniel; Graff, Peter. Science 335(6072):1042-a, 2012. doi:10.1126/science.1215107.

key: jaeger-2012 camp: mainstream method: model direction: complicates verified: full

Shuffling simulations show the cline is carried by where families sit, not individual languages, and that the method's false-positive rate is inflated. The family as the unit of dispersal is what a Babel model also predicts.

Comment on Atkinson (Cysouw, Dediu and Moran)

Cysouw, Michael; Dediu, Dan; Moran, Steven. Science 335(6069):657-b, 2012. doi:10.1126/science.1208841.

key: cysouw-2012 camp: mainstream method: model direction: complicates verified: full

With actual phoneme counts the best-fit origins are eastern Africa and the Caucasus, with a credible area including the Middle East, and phoneme inventories carry no reliable deep signal at all. Read from a Wayback snapshot of the Science page.

Comment on Atkinson (Wang, Ding, Tao and Li)

Wang, Chuan-Chao; Ding, Qi-Liang; Tao, Huan; Li, Hui. Science 335(6069):657-c, 2012. doi:10.1126/science.1207846.

key: wang-2012b camp: mainstream method: model direction: complicates verified: full

Exact counts give a marginally better fit from central Asia, "where the Babel was supposed to be". The margin is in the third decimal place, Atkinson showed it rests on four related outliers, and their own data with the original binning put the origin back in Africa. The Babel phrase is a geographic aside, not an endorsement.

Comment on Atkinson (Van Tuyl and Pereltsvaig)

Van Tuyl, Rory; Pereltsvaig, Asya. Science 335(6069):657-d, 2012. doi:10.1126/science.1209176.

key: vantuyl-2012 camp: mainstream method: measurement direction: complicates verified: full

No phoneme decline with distance on any continent but Africa, and none there after ten languages were rechecked. So the phoneme data say nothing about the route through Eurasia, for or against a Near Eastern start.

Response to comments (Atkinson 2012)

Atkinson, Quentin D. Science 335(6069):657-e, 2012. doi:10.1126/science.1210005.

key: atkinson-2012 camp: mainstream method: model direction: complicates verified: full

Restores Africa as the best fit on Wang's data by removing outliers, by a margin over Europe the author calls "not decisive".

Rejection of a serial founder effects model of genetic and linguistic coevolution

Hunley, Keith; Bowern, Claire; Healy, Meghan. Proceedings of the Royal Society B 279(1736):2281-2288, 2012. doi:10.1098/rspb.2011.2296.

key: hunley-2012 camp: mainstream method: measurement direction: complicates verified: full

Every genetic prediction of a serial founder process holds and most phonemic ones fail; phonemes record recent contact, not the ancient expansion. That removes the phoneme gradient as evidence for an African origin and for a Babel origin alike.

A comparison of worldwide phonemic and genetic variation in human populations

Creanza, Nicole, et al. PNAS 112(5):1265-1272, 2015. doi:10.1073/pnas.1424033112.

key: creanza-2015 camp: mainstream method: measurement direction: complicates verified: full

2,082 phoneme inventories against genetic data from 246 populations. Inventory size does not fall with distance from Africa; the best-fit centre is in the far north of Europe, which the authors say does not mark an origin. The paper Tomkins misreads.

Dating the origin of language using phonemic diversity

Perreault, Charles; Mathew, Sarah. PLoS ONE 7(4):e35289, 2012. doi:10.1371/journal.pone.0035289.

key: perreault-2012 camp: mainstream method: model direction: complicates verified: full

A phoneme clock calibrated on an assumed 45,000 to 65,000-year settlement of South-East Asia and the Andamans, which a young-earth reader rejects. The ratio of its own figures survives any calibration, about 3.5, and the article prices it as a scale argument of its own. The authors say its premise is not widely accepted and its rate rests on one case.

The emergence of grammar: systematic structure in a new language

Sandler, Wendy; Meir, Irit; Padden, Carol; Aronoff, Mark. PNAS 102(7):2661-2665, 2005. doi:10.1073/pnas.0405448102.

key: sandler-2005 camp: mainstream method: measurement direction: supports verified: full

Al-Sayyid Bedouin Sign Language developed a fixed word order found in no language around it within one generation. A new language can arise fast among people who have the faculty; it says nothing about how fast spoken languages split.

Children creating core properties of language

Senghas, Ann; Kita, Sotaro; Özyürek, Aslı. Science 305(5691):1779-1782, 2004. doi:10.1126/science.1100199.

key: senghas-2004 camp: mainstream method: measurement direction: supports verified: full

Nicaraguan Sign Language gained discrete, combinatorial structure from children within about two decades of a school opening in 1977. Read in the accepted manuscript; the article quotes only the published abstract.

How fast languages change

The basis of glottochronology

Lees, Robert B. Language 29:113-127, 1953. doi:10.2307/410164.

key: lees-1953 camp: mainstream method: measurement direction: complicates verified: full

Thirteen documented ancestor and descendant pairs on a 215-word list keep 76% to 85% a millennium, mean 81%. Nothing in them is faster than 76%, which is the quantity a post-Babel compression needs exceeded. Lees concedes the constant rate is "clearly false" if word classes decay at different speeds. Read in the archive.org microfilm of Language.

Towards greater accuracy in lexicostatistic dating

Swadesh, Morris. International Journal of American Linguistics 21(2):121-137, 1955. doi:10.1086/464321.

key: swadesh-1955 camp: mainstream method: measurement direction: complicates verified: full

The 100-word list keeps about 86% a millennium. Swadesh's reason for the revision, that unstable items go first so the observed rate rises with depth, is the rate-heterogeneity point that makes a constant-rate date a floor. He drops the fast Coptic case as hidden dialect divergence.

On the validity of glottochronology

Bergsland, Knut; Vogt, Hans. Current Anthropology 3(2):115-153, 1962. doi:10.1086/200264.

key: bergsland-1962 camp: mainstream method: measurement direction: complicates verified: full

The classic disproof of a constant rate: Icelandic keeps nearly everything, East Greenlandic in a group of about 400 with a name taboo changes several times faster. It is a real mechanism for fast change in small groups, and an equally real case of very slow change. One commentator's remark on p. 147 is not used; see the quarantine file.

Frequency of word-use predicts rates of lexical evolution throughout Indo-European history

Pagel, Mark; Atkinson, Quentin D.; Meade, Andrew. Nature 449:717-720, 2007. doi:10.1038/nature06176.

key: pagel-2007 camp: mainstream method: model direction: undercuts verified: abstract

Frequency of use predicts about half the variation in replacement rates among 200 meanings. Only the abstract was read; the article cites only the abstract, and takes the half-lives from Pagel 2013.

Evolutionary dynamics of language systems

Greenhill, Simon J., et al. PNAS 114(42):E8822-E8829, 2017. doi:10.1073/pnas.1700388114.

key: greenhill-2017 camp: mainstream method: model direction: complicates verified: full

Basic vocabulary is the slowest and most tree-like part of language in 81 Austronesian languages, grammar five times faster, and 29% of lexical change comes in bursts at language births. The bursts help a young chronology a little; the absence of any faster subsystem that could hide relationships does not.

Languages evolve in punctuational bursts

Atkinson, Quentin D., et al. Science 319(5863):588, 2008. doi:10.1126/science.1149683.

key: atkinson-2008 camp: mainstream method: model direction: complicates verified: full

10% to 33% of lexical divergence happens at splits, 21% in Indo-European. Bursts are tied to the number of splits, not elapsed time, so they reduce the time-dependent part needed by at most a fifth.

Rate of language evolution is affected by population size

Bromham, Lindell; Hua, Xia; Fitzpatrick, Thomas G.; Greenhill, Simon J. PNAS 112(7):2097-2102, 2015. doi:10.1073/pnas.1419704112.

key: bromham-2015 camp: mainstream method: measurement direction: supports verified: full

Small populations lose basic words faster, slope -0.12 on log-log scales, a factor of about 2.5 over its range. It is the measured basis of the article's strongest escape route. Against it: total change did not depend on population, founder effects did not improve the fit, and the loss effect was not significant on two of three population measures.

Is the rate of linguistic change constant?

Nettle, Daniel. Lingua 108:119-136, 1999. doi:10.1016/S0024-3841(98)00047-3.

key: nettle-1999 camp: mainstream method: model direction: supports verified: full

A simulation, not data: change is much faster below about 200 to 250 speakers. It reports Lithgow's Muyuw case, 13% replacement in 50 years, which was not read at source.

The World's Major Languages

Comrie, Bernard (ed.). The World's Major Languages. Croom Helm and Oxford University Press, 1987.

key: comrie-1987 camp: mainstream method: handbook direction: complicates verified: full

Specialists on Latin to Romance, Arabic to its colloquials and Maltese, and Malagasy: in 1,000 to 2,000 years one language becomes a group of mutually unintelligible but plainly related languages. Read by archive.org search-inside; chapter openings checked against the copyright page.

Chinese

Norman, Jerry. Chinese. Cambridge Language Surveys. Cambridge University Press, 1988.

key: norman-1988 camp: mainstream method: handbook direction: complicates verified: full

The Sinitic varieties as a family as diverse as Romance from the Qin-Han expansion, about 2,200 years: fast diversification outside Indo-European, and still one obvious family.

A linguist's perspective on the settlement history of Madagascar

Adelaar, Alexander. NUSA 61:69-88, 2016.

key: adelaar-2016 camp: mainstream method: review direction: undercuts verified: full

About 1,300 years on a large island with heavy contact produced one relatively homogeneous dialect complex, still plainly Southeast Barito.

Language Contact, Creolization, and Genetic Linguistics

Thomason, Sarah Grey; Kaufman, Terrence. University of California Press, 1988.

key: thomason-1988 camp: mainstream method: handbook direction: undercuts verified: full

Creoles and mixed languages take most of their basic vocabulary from one source, so they look closely related to it, not unrelated. No documented contact process turns one language into several whose basic vocabularies share nothing. It gives no creole-against-lexifier percentage, so the article gives none beyond Crowley's.

The data of the divergence-time test

An Indoeuropean Classification: A Lexicostatistical Experiment

Dyen, Isidore; Kruskal, Joseph B.; Black, Paul. Transactions of the American Philosophical Society 82(5), 1992.

key: dyen-1992-indoeuropean-classification camp: mainstream method: measurement direction: background verified: full

Pairwise cognate percentages for 84 Indo-European lists, the validation targets for the article's code, and the statement that similarity decays like a mixture of radioactive elements. Its treatment of creole cognates as "pseudo-cognates" is a definition, not a measurement.

Comparative Indoeuropean Database, File IE-DATA1

Dyen, Isidore; Kruskal, Joseph; Black, Paul. 1997.

key: dyen-1997 camp: mainstream method: dataset direction: background verified: full

The raw cognation classes behind the monograph. The code reproduced 11 of 13 printed percentages from it to the decimal.

IE-CoR

Heggarty, Paul, et al. Indo-European Cognate Relationships database, CLDF release, 2023. https://github.com/lexibank/iecor

key: heggarty-2023 camp: mainstream method: dataset direction: background verified: full

160 languages, 170 meanings, 25,741 cognate judgements and 52 text-dated ancient languages. The data of the Indo-European test. Its list was chosen to include less stable meanings, so its rates are calibrated on itself.

Austronesian Basic Vocabulary Database

Greenhill, Simon J.; Blust, Robert; Gray, Russell D.; Forkel, Robert. CLDF release, 2020. https://github.com/lexibank/abvd

key: greenhill-2020 camp: mainstream method: dataset direction: background verified: full

2,040 word lists over 210 meanings. The data of the Austronesian test.

High-precision radiocarbon dating shows recent and rapid initial human colonization of East Polynesia

Wilmshurst, Janet M.; Hunt, Terry L.; Lipo, Carl P.; Anderson, Atholl J. PNAS 108(5):1815-1820, 2011. doi:10.1073/pnas.1015876108.

key: wilmshurst-2011 camp: mainstream method: measurement direction: supports verified: full

East Polynesia settled after AD 1000 in two pulses, on 1,434 screened dates. Its late chronology makes the Polynesian calibration rates faster, which favours a short chronology.

Families written down before Babel

Visible Language

Woods, Christopher (ed.). Visible Language: Inventions of Writing in the Ancient Middle East and Beyond. Oriental Institute Museum Publications 32, 2010.

key: woods-2010 camp: mainstream method: review direction: undercuts verified: full

Egyptian written by rebus at Abydos Tomb U-j about 3320 BC; Sumerian behind the Uruk tablets "quite slim", certain by the mid-third millennium; Chinese from the oracle bones about 1220 BC. Printed page = PDF page - 2.

The Cambridge Encyclopedia of the World's Ancient Languages

Woodard, Roger D. (ed.). Cambridge University Press, 2004.

key: woodard-2004 camp: mainstream method: handbook direction: undercuts verified: full

Specialists' first attestations: Sumerian convincing at Ur about 2800 BC, Eblaite and Akkadian by the mid-third millennium, Hurrian names in the Akkade period, Hattic only in Hittite archives, Linear B from perhaps the fifteenth century BC. Read by archive.org search-inside; the leaf offset drifts, so each page was read from the scan data and checked against the book's index.

The Decipherment of Linear Elamite Writing

Desset, François, et al. Zeitschrift für Assyriologie 112(1):11-60, 2022. doi:10.1515/za-2022-0003.

key: desset-2022 camp: mainstream method: measurement direction: undercuts verified: full

Elamite, an isolate, written by about 2240 BC in the Treaty of Naram-Sin, before the decipherment and independent of it. How specialists received the decipherment was not checked. Printed page = PDF page + 10.

Kanišite Hittite

Kloekhorst, Alwin. Kanišite Hittite: The Earliest Attested Record of Indo-European. Brill, 2019.

key: kloekhorst-2019 camp: mainstream method: measurement direction: complicates verified: full

Hittite names from about 1930 BC, the oldest Indo-European words, after every Masoretic Babel date; Indo-European's problem is its internal diversity, not its attestation. Declines to rely on the Armi names from Ebla. Read in the author-shared PDF, printed page = PDF page - 13.

An absolute chronology for early Egypt

Dee, Michael, et al. Proceedings of the Royal Society A 469:20130395, 2013. doi:10.1098/rspa.2013.0395.

key: dee-2013 camp: mainstream method: measurement direction: undercuts verified: full

Aha's accession 3111 to 3045 BC by Bayesian radiocarbon, independent of the king lists. Read by the sibling article egypt-and-the-flood-date, which owns the argument.

Integrated Tree-Ring-Radiocarbon High-Resolution Timeframe

Manning, Sturt W., et al. PLoS ONE 11(7):e0157144, 2016. doi:10.1371/journal.pone.0157144.

key: manning-2016 camp: mainstream method: measurement direction: undercuts verified: full

Fixes the early second millennium BC to the Middle or Low-Middle Chronology and excludes the lower ones. Read by the sibling article mesopotamian-absolute-chronology, which owns the argument.

Considered, not used

How many languages are there in the world?

Eberhard, Simons and Robinson (eds.), Ethnologue, 29th edition, insight page.

key: eberhard-2026b used: no verified: full

Gives 7,170 living languages. Not cited because the article counts languages and families from Glottolog's tables, and Ethnologue's unit differs.

Phylogenetic evidence for Sino-Tibetan origin in northern China (abstract row)

key: zhang-2019 used: no verified: abstract

The library row holding only the PubMed abstract of Zhang and colleagues 2019. Superseded by the Supplementary Information, which was read and is cited under its own key.

Response to comment on Atkinson (reply to Jaeger and colleagues)

Atkinson, Quentin D. Science 335(6072):1042-b, 2012. doi:10.1126/science.1215788.

key: atkinson-2012b used: no verified: full

Atkinson's answer to the shuffling test, that 0.2% to 0.5% of shuffled sets match his effect. Not cited; the article reports Jaeger's own figure and Atkinson's general reply.