Speech Errors as Evidence for Historical Generative Phonology—Joseph Paul Stemberger Son of Lingua Pranca Contents

Reconstructing the Proto-Indo-Europeans

Joseph Paul Stemberger

In the recent years, much attention has been given to the vocabulary of Proto-Indo-European (PIE). What words the PIE-ans did and did not have can tell us much. Benveniste (1973) discusses PIE kinship systems, economy and ways of life. In these studies, however, there is one very important area of the lexicon that has been neglected. No one has yet analyzed the terms for body parts to determine what the PIE-ans actually looked like. Most have probably assumed that they looked much like their descendants do today, but there is no empirical evidence supporting this assumption, and we really must attempt to verify or discredit it.

At first glance, the lists in Buck (1949) make the PIE-ans seem pretty normal. They had bodies (*krep-), bones (*ost-), heads (*kap-ut-) and most parts thereof, throats (*gwer-), shoulders (*osmo-s), arms (*armo-s), bellies (*udero-), navels (*ombh-), penises (*pes-), testicles (*orghi-), rumps (*ers-), knees (*genu-), feet (*ped-), fingernails (*nogh-), and hair (*pi-lo-). However, as everyone knows, it is often the lexical gaps that are most revealing, and PIE has several. There is no word for women’s breasts (* ) or genitalia (* ), no word for hand (* ), leg (* ), eyebrow (* ), or moustache (* ), nor are different types of hair distinguished.

Lexical gaps are usually assumed to show that the PIE-ans knew nothing of the concept in question, but this is obviously absurd here. We know from comparative anatomy that the PIE-ans must have had most of these. Why, then, did they have no words for them? The answer may be found in Rosch’s (1978) concept of basic level category. Basically, this states that there is a natural level of specificity in a lexical field that will be used in ordinary speech, and people will use more specific or less specific terms only in special instances. Rosch has emphasized that the basic level category is determined by perceptual salience alone, and never by function. We may thus conclude that lexical gaps represent characteristics which, while present and perhaps functionally important, are in one way or another not salient, or not perceptually different enough from other parts of the body. The highly salient male genitalia had a lexical item for each structure, while the equally functional but nonsalient female genitalia had no lexical item. Eyebrows and moustaches must not have been distinctively separated from other head hair, and all hair on the body must have been of more or less the same type. The female breast must not have been much different from a male breast perceptually. Legs and arms, and hands and feet, must have been more similar structurally than they are today. The word *ped- must have meant ‘foot, hand’, and *armso-s ‘arm, leg’. This explains also the general identity of words for finger and toe (e.g. Lat. digitus), and for ankle and wrist (e.g. OE wriste ‘ankle, wrist’) in IE languages. As these structures differentiated historically, new words were coined to distinguish them. We thus find that the earlier assumption that the PIE-ans looked like us cannot be correct.

The PIE-ans’ appearance (a body covered with hair, hands and feet structurally identical, arms and legs more similar that they are today, men’s and women’s breasts very close in appearance) does not describe a hitherto unknown race of man. It corresponds to a high degree with that posited for the common ancestor of man and the great apes, which lived 12-20 million years ago. This can hardly be accident, and we must conclude that the PIE-ans were this ancient people. Of course, we must now also radically revise our best estimate of the date when PIE was spoken, from 6000 years ago as previously thought, to over 12 million years ago.

This conclusion, that the PIE-ans were pre-hominids, is supported by evidence from three other areas: lexical gaps in the vocabulary of behavior, comparative neuroanatomy, and studies of the phonetic capabilities of the primate vocal tract.

The reconstruction of behavioral words in PIE also shows a systematic gap. All such words that can be reconstructed refer to common mammalian characteristics: *an- ‘breathe’, *ghī- ‘yawn’, *kwās- ‘cough’, *pster- ‘sneeze’, *sweid- ‘sweat’, *spyēu- ‘spit’, *wem- ‘vomit’, *leigh- ‘lick’, *swep- ‘sleep’, *perd- ‘fart’, *meigh ‘piss’, *kakka- ‘shit’, *yebh- ‘fuck’, *nogw- ‘naked’, etc. Completely lacking are terms referring to characteristically human and higher primate behavior: ‘kiss’, ‘caress’, ‘walk’ (cf. *dhregh- ‘run’). This gap is explainable only if these characteristics had not yet evolved, e.g. 15,000,000 years ago. Another word lacking is ‘clothes’; we would expect pre-hominids to have no knowledge of this, but a group of humans living in Europe 6000 years ago could not have lived through the winter without them.

Corroborating evidence is not limited to the lexicon, but is also forthcoming from more biological aspects of linguistics. Geschwind and Levitsky (1968) discovered a neuroanatomical asymmetry between the planum temporale of the right and left hemispheres of the human brain, with that of the left being larger in most individuals. LeMay (1976) notes that the same is true of the nearby areas of the parietal lobe as well. This general area corresponds to Wernicke’s Area, a part of the left hemisphere that is thought to underlie language function. Geschwind posits that this anatomical asymmetry, larger on the left where language is located and smaller on the right where language is not located, originally evolved as a response to language, to increase the efficiency with which man could use language. LeMay (1976) also notes, however, that all the higher primates show this asymmetry, not just man, even though they cannot speak; if speech is the cause of the asymmetry, there is a problem here. The anatomical asymmetry developed at a time when language was being used, and has been maintained in the other higher primates even though language has subsequently been lost in these species. The asymmetry is explainable only in this fashion.

Lastly, Lieberman’s (1975) studies of the phonetic capabilities of the primate vocal tract greatly support our conclusions. He notes that all higher primates are capable of producing a wide range of consonants, including labials, dentals, and (possibly) velars, stops, fricatives, nasals, liquids, and glides. The larynx of most primates is more complex than man’s and able to produce a wider range of sounds. They are not able, however, to produce the cardinal vowels [i, u, a], but only a shrunken vowel triangle with the endpoints [ɪ, ʊ, ʌ]. This description should also be true of the PIE-ans as I have posited. It presents no problems. As Lehmann (1955) has pointed out, PIE had only one vowel, /e/; this is within the capability of any pre-hominid! Further, the more agile larynx of our ancestor would be capable of producing all the laryngeals posited for PIE, whereas our own are not. The phonetics of PIE seems uniquely designed for a pre-hominid vocal tract, and this strongly supports the conclusions above. It also allows us to calmly accept the typological impossibility of the PIE phonetic inventory, since the PIE-ans were not human and hence cannot be expected to follow human universal patterns.

With all this evidence, it can hardly be disputed that the PIE-ans were our pre-hominid ancestors. The history of the language must have been as follows. PIE was the original language of mankind, from which all modern languages are descended, and was spoken 12-20,000,000 years ago. This is why so many universals can be discovered by studying the Indo-European languages. The daughter languages broke apart 12,000,000 years ago, and have been maintained as separate, despite probably interbreeding. We know that culturally isolated groups show a slower rate of language change than those with rapidly changing cultures. Thus, those languages that can no longer be easily derived from PIE must have had a great deal of cultural change much more rapidly and much earlier than that of the Indo-Europeans. We can conclude from this that the IE-ans were the real dummies of the human world, constantly being shoved to the periphery and remaining unchanged while brighter peoples developed high cultures. The IE-ans were indeed very late bloomers, since the languages spoken only 2500 years ago show incredible resemblances to PIE itself. Obviously, we must conclude that languages can change much more slowly than we had believed possible earlier.

The implications of this discovery are manifold and still remain to be worked out in detail. It suggests, for example, that more success may be had teaching chimps language if PIE is used rather than English or ASL as in previous attempts. The most startling implications, however, are theological and relate to God’s appearance, but that is far beyond the scope of this paper. (For those interested, the relevant Biblical line is ‘Man was created in the image and likeness of God’. As an aside, we may note that the chimpanzee’s genus name is Gk. Pan ‘the All-Thing’.) Suffice it to say that our picture not only of the evolution of PIE, but also of the evolution of language and of man himself will have to be radically revised.


Benveniste, Emile. 1973. Indo-European Language and Society. London: Faber and Faber.

Buck, Carl Darling. 1949. A Dictionary of Selected Synonyms in the Principal Indo-European Languages. Chicago: University of Chicago Press.

Geschwind, Norman and W. Levitsky. 1968. Human brain: Left-right asymmetries in the temporal speech region. Science 161.186-7.

Lehmann, Winfred. 1955. Proto-Indo-European Phonology. Austin: University of Texas Press.

LeMay, Marjorie. 1976. Morphological cerebral asymmetries of modern man, fossil man, and nonhuman primate. In S. Harnad, H. Steklis, and J. Lancaster (eds.), Origins and Evolution of Language and Speech, pp. 349-66. New York: New York Academy of Sciences.

Lieberman, Philip. 1975. On the Origins of Language. New York: MacMillan.

Rosch, Eleanor. 1978. Principles of categorization. In E. Rosch and B. Lloyd (eds.), Cognition and Categorization, pp. 27-48. Hillsdale: Lawrence Erlbaum.

Speech Errors as Evidence for Historical Generative Phonology—Joseph Paul Stemberger
Son of Lingua Pranca Contents