I suggest avoiding Ruhlen like the plague since he is considered a pseudoscientist by most serious historical linguists. The same goes for "edenics" which is the linguistic equivalent of creation science - i.e. not science at all.
You need Lyle Campbell and William Poser's book "Language Classification: History and Method", and Campbell's "Historical Linguistics". Specifically about semantic reconstruction here are a couple of references:
Magnus: I appreciate your comments. Over a decade ago I was looking at early Amerind historical linguistics, read a bit, and wrote a couple of paragraphs. Do you mind if I run them by you, with a request for criticism? This is taken from an unpublished revised version of my dissertation of 2005, *Los otomíes: cultura, lengua y escritura*:
Joseph Greenberg elaboró una de las reconstrucciones más influyentes –y polémicas– de la prehistoria de América. Según este lingüista, los primeros pobladores de América hablaban un idioma relacionado con el macrogrupo euroasiático. Greenberg emplea un método de análisis multilateral, comparando listas cortas de palabras “conservadoras” en centenares de lenguas. Llega a la conclusión de que todas las lenguas nativas de América pertenecen a tres grandes grupos, relacionados con sendas olas migratorias desde Asia. El grupo esquimal-aleuto corresponde a la región ártica. El na-dene se encuentra en el oriente de Alaska, el noroeste de Canadá y algunas regiones del Suroeste estadounidense. Ni el esquimal-aleuto ni el na-dene son grupos novedosos; ambos fueron propuestas por Edward Sapir hacia principios del siglo XX.[1] La innovación de Greenberg fue el grupo amerindio, que incluye todas las demás lenguas indígenas de América, desde la bahía de Hudson hasta Tierra del Fuego. Greenberg propuso alrededor de 300 etimologías para apoyar la creación de este macrogrupo hipotético; su discípulo Merritt Ruhlen ha agregado otras 200. Según Greenberg y Ruhlen, la primera migración, anterior al año 10,000 a. C., fue la de los hablantes del proto-amerindio. La segunda migración fue la de los proto-na-denes, probablemente entre 8000 y 3000 a. C. La tercera migración fue la de los proto-esquimal-aleutos, hacia el tercer milenio a. C.[2]
El trabajo de Greenberg y Ruhlen ha sido objeto de críticas fuertes. Según Johanna Nichols, el alto grado de diversidad estructural y genética de las lenguas americanas hace poco aplicable el método multilateral de Greenberg al estudio de este continente.[3] Lyle Campbell rechaza el método multilateral; prefiere la comparación intensiva de dos lenguas a la vez, construyendo los grupos desde las relaciones más cercanas hasta las más lejanas. Este método, más conservador –y mucho más laborioso–, tiende a negar la existencia de las relaciones genéticas remotas.[4] Pero Greenberg y Ruhlen presentan argumentos estadísticos de la probabilidad infinitesimal de que sus etimologías sean meras coincidencias.[5] Este debate no da muestras de una pronta resolución.
[4] Campbell (Lyle), 1997 (las críticas del trabajo de Greenberg llenan buena parte de este libro; véase la entrada correspondiente en el índice de autores, p. 506).
Greenberg does not make any actual reconstructions at all - he makes a classification. This is the main critique against his 1987 work. He does not use the comparative method which is the way linguistic reconstructions are made, and genetic relations between languages are demonstrated. Instead he relies entirely on comparing surface forms visually across multiple languages and making subjective judgments of relatedness.
His and Ruhlens subsequent proposed etymologies and statistical claims are meaningless, since even much of the data that Greenberg's claims rely on has been demonstrated to be invalid to begin with (erroneous and spurious word forms, known loan words, even entire non-existing languages) - which makes statistical arguments based on the flawed data irrelevant.
From the point of view of mainstream historical linguistics the debate is fully resolved, and Greenbergs work (and Ruhlen's) work is considered junk science. Ruhlen has certain connections to a school of Russian historical linguists who work on reconstructing deep relations between languages all the way back to a supposed Nostratic or Proto-World language. This body of work is mostly disregarded by mainstream historical linguistics.
I also dont like the way you describe the well established method of mainstream historical linguistics as some kind of quaint preference of Lyle Campbell. The method Campbell espouses is the only one that is generally accepted in historical linguistics, and the only one that has produced the evidence that has stood the test of time. Some find it too tedious and prefer to make hypotheses of long distance deep historical relations by scanning vocabulary lists at a glance and leave the reconstruction to others, but there are very good epistemological reasons this kind of research should not be accepted.
The main problem is that there are very few linguists working on the historical linguistics of Native American languages, meaning that scholars who specialize in languages of other areas largely have to accept classifications of American languages based on their faith in the methods and data of a given classification. Whenever linguists specializing in individual American language families have reviewed Greenbergs data for their area they have found it to be riddled with basic errors - and his methodology does not allow making basic checks on his reasoning or conclusions. I do not know any historical linguists working on American families who consider Greenbergs classification to merit consideration.
So in short I think your summary misrepresents the relative standing of the claims, and leaves out the mainstream classification of Native American languages - which considers Greenberg's Amerind to be entirely baseless. I would not personally spend two paragraphs on Greenberg and Ruhlen's work in a work on the linguistic history of Otomi or Nahuatl.
Thank you, Magnus, for dedicating a considerable amount of time and thought to my query. Would you mind if I cite your contributions to this RG thread in a footnote? I don't know if I'll ever do anything with this revised version of my dissertation, but it's fun struggling to keep it current and continuing to polish the prose, now without the pressure of a deadline.
Not at all, but there are a lot of other works you could cite instead for the same criticisms of Greenberg and Ruhlen's work.
Bolnick, D. A. W., Shook, B. A. S., Campbell, L., & Goddard, I. (2004). Problematic use of Greenberg’s linguistic classification of the Americas in studies of Native American genetic variation. American journal of human genetics, 75(3), 519.
Berman, H. (1992). A Comment on the Yurok and Kalapuya Data in Greenberg's Language in the Americas. International Journal of American Linguistics, 230-233.
Poser, William J. "The Salinan and Yurumanguí data in Language in the Americas." International Journal of American Linguistics (1992): 202-229.
Goddard, I., & Campbell, L. (1994). The history and classification of American Indian languages: What are the implications for the peopling of the Americas. Method and theory for investigating the peopling of the Americas. Center for the Study of the First Americans, Oregon State University, Corvallis, 189-207.
Ringe Jr, Donald A. "The mathematics of'Amerind'." Diachronica 13, no. 1 (1996): 135-154.
Kimball, Geoffrey. "A Critique of Muskogean," Gulf," and Yukian Material in" Language in the Americas"." International Journal of American Linguistics (1992): 447-501.
Curnow, Timothy Jowan. "Why Paez is not a Barbacoan language: The nonexistence of" Moguex" and the use of early sources." International journal of American linguistics (1998): 338-351.
Thomason, Sarah G. "Hypothesis generation vs. hypothesis testing: a comparison between Greenberg's classifications in Africa and in the Americas." Unpublished manuscript: University of Pittsburgh (1994).
Ramer, A.M., 1996. Tonkawa and Zuni: two test cases for the Greenberg classification. International journal of American linguistics, pp.264-288.
1. The language organ- proposing the biological genesis of language.
2. Proto-world & the monogenesis of current languages.
Again, as above, the nature-nurture dichotomy seems to be the predominant strands of thought. One can situate arguments quite reasonably along one of these veins. Citations have been given by some in this forum.