A real classic in which Greenberg tries to examine the couple Marked-Unmarked in languages, hence in linguistics, a book we cannot ignore. This couple of concepts are fundamental and basic in and for phonology and they were actually invented for phonology by Jakobson among others. After analyzing them in that context, Greenberg tries to transfer them to grammar and lexicon, and the attempt is interesting but probably not the best thing to do. It would have been a lot more constructive to analyze the real universals of both grammar and lexicon. The fundamental rule is that the unmarked category is more frequent, and he reverses it into one criterion to know an unmarked category is to look for the more frequent one. That seems to work very well for phonology though in his fourth chapter he pays lip service to the major difference between phonology on one side and grammar-lexicon on the other side : "the sound meaning relation [...] is absent in the former and present in the latter." (p. 63)
It is surprising that does not derail his attempt because it is a fundamental parameter in genetic linguistics, and Greenberg remains one master in that field: any resemblance or similitude has no genetic value if it does not associate both the form and the meaning. This goes back to de Saussure and was vastly amplified by Meillet. It is also in full contradiction with what he himself states page 68: "such highly infrequent formation must follow analogically other parts of the system while only a fairly frequent form can preserve irregularities." The system, so dear to Meillet for example, in grammar and lexicon is seen by Greenberg himself as a regulating instance that eliminates highly infrequent forms when opposed to highly frequent ones.
I am now going to make a few remarks on particular points. His frequency hierarchy Singular-Plural/Plural-Dual is bizarre because it does not take into account the phylogeny of number that started from a collective singular containing plurality, then extracted smaller and smaller meaningful groups, the dual being the most famous one because of the two hands, feet and other organs of the same type and the extension of this physiological dual to shoes, gloves, and even oxen. "Children" is not a dual but is a typical extracted collective plural that survives today. Then when the unit is reached the plural is invented as an exterior dimension by multiplication. His hierarchy then does not correspond to a logic of any sort and cannot explain why that dual can survive even in our languages.
When he deals with personal pronouns he does not take into account the second person because in English it is the same word in the singular and the plural. But how can he compare the first person to all the third persons, singular or plural, when it is obvious that first and second persons are linked together in the very communicative act: the speaker and the listener, the emitter and the receiver, etc? His statistics that give a vast dominance to the third person singular or plural are unrealistic.
Dealing with adjective couples like "good" and "bad", he says: "there is as far as is known to me, no language which lacks a separate term for "good" and expresses it normally by "not-bad"." The first language to which I think is Pali where for example there are four negative nouns: "vajja" (wrong belief), "mobha" (attachment), "dosa" (ill will, hatred), and "moha" (delusion) with the four positive terms built with the use of the privative prefix /a-/: "avajja" (right belief), "alobha" (generosity), "adosa" (goodwill, loving kindness), "amoha" (wisdom). And we could think of American that invented the negative adjective for a natural, hence positive state: "un-circumcised".
What he says about the masculine agreement of adjectives attributed to both masculine and feminine nouns is maybe right in many languages (like Spanish that he quotes) but is wrong in French. The French could say: "j'ai acheté une robe et un manteau blancs" [literal: I have bought a (feminine) dress (feminine) and a (masculine) coat (masculine) white (masculine plural)] but the French would say: "j'ai acheté un manteau et une robe blanches" where the adjective is plural and feminine because "robe" is just before it. I am pretty sure there are many other languages that would consider agreement as more complex than what Greenberg says.
He finally comes to a remark that is the admission his whole parallel between phonology and grammar-lexicon is questionable: "Thus in phonology diachronic process explains frequency, while in grammar, frequency explains diachronic process." I am far from sure this generalization is warranted.
His last chapter on kinship terms based only on frequency forgets simply that this set of words directly reflect the social organization of the family in the clan, the tribe and the society. In fact it is a set of words that reflect the level of development of the whole procreative and reproductive survival instinct of the species, the clan and the individual. That cannot be reduced to a few frequency statistics and if in Bantu languages there is a special term for the maternal uncle, it is because in the Bantu family the authority over the children is not the father but the mother's brother, at least in the traditional family. Nelson Mandela says a lot about this particular family structure. An important classic but that has to be taken with a good sense of criticism.
Dr Jacques COULARDEAU, University Paris 1 Pantheon Sorbonne, University Paris 8 Saint Denis, University Paris 12 Créteil, CEGID