Center Embedding as
(not contact-induced innovation)
In an inexplicable lapse of judgment, the editors of Speculative Grammarian recently published an analysis by Küçük (Küçük 2008), claiming that several languages which were previously without passive constructions have created them through a type of center embedding; the motivation given for this emergent construction is said to be contact with English. Küçük’s article, in spite of the unassailable nature of the data presented, is without the slightest grain of reliability or trustworthiness in its conclusions. In the present article, I will show where Küçük went astray in her so-
I hasten to say that I have no doubt whatsoever that Küçük is a truly excellent scholar, in her area of expertise, but when it comes to the motivations proposed in the article mentioned above, I find her proposal worth less than a red herring.
In fact, center-
Küçük is correct in her claim that the languages in her sample represent an exhaustive list of human languages which allow center embedding for any purpose (my entire research center has devoted years to the pursuit of center embedding, and can authoritatively vouch for the veracity of this claim). This list, in alphabetical order, is:
E (Tai-Kadai; China; 30,000 speakers) Ere (Austronesian; Papua New Guinea; 1,030 speakers) Erre (Australian; Australia; 1 speaker) Malayalam (Dravidian; India; 35,000,000 speakers) Mam (Mayan; Guatemala; 500,000 speakers) Manam (Austronesian; Papua New Guinea; 7000 speakers) Mum (Trans-New Guinea; Papua New Guinea; 3,286 speakers) Mutum (Trans-New Guinea; Papua New Guinea; 400 speakers)
Now, anyone who’s had even minimal linguistic training will see immediately that a certain principle underlies the very names of these languages
Consider the tree diagram given in Figure 1:
Nearly all of the historical relationships illuminated by Figure 1 can be explained by a single phonological principle. Namely, changes a, d, and e each simply involve vowel copying and insertion of a consonant to break up an impermissible VV string. In all cases, this process operates on a pivot vowel (one which is in the exact center of the word). It is this key factor which gives rise to the phenomenon of center embedding.
Furthermore, I have arranged the tree so as to show clearly that this change happened at the same time in all three daughter languages. Rather than independent innovations, this was a single process, which had its roots in a single ancestor language, operating simultaneously on three branches of the family tree. This pivot elaboration process is clearly rooted in the essential nature of the parent language, and thus naturally inherited into each daughter.
There are a few changes in Figure 1 which are not straightforwardly predicted by this simple rule, and here I treat them in turn.
Changes f and g are essentially the same as the basic rule. Change f shows that where the language name has a pivot consonant instead of a pivot vowel, the consonant is copied and a VCV string is inserted. Similarly, change g shows that where the pivot consonant is /r/, insertion fails and /rr/ results. These differences from the basic rule are trivial.
The only interesting issue arising from Figure 1 is in change c, which superficially seems somewhat different from what I have seen with all the other changes. However, detailed research into the history of Mam and E has shown me that this change is due to language-
Critical readers (who aren’t too lazy) may take issue with the notion that three daughter languages would undergo the crucial pivot transformation at the same or nearly the same time. Even the fact that the three languages in question are all found in Papua New Guinea may be insufficient to convince the ardent pivot-
For complex cultural reasons, speakers of Mam and its daughter languages track such reversals using an ingenious apparatus (which I can’t discuss at length here, but I will note that it is made of chewing gum, baling wire, a battery-
One more important point can be made, based on some information which Küçük provided, and which I had never noticed before. In each of the successive changes given in Figure 1, the population of speakers of the daughter language is significantly fewer than that of the parent language. (I won’t do the math for you, but you can see it for yourself, if you aren’t too lazy.) This clearly indicates that each additional layer of center embedding increases significantly the difficulty in learning and maintaining the resulting language. Thus, I may predict that if Malayalam, Mutum or Erre undergoes another round of pivot elaboration, the resulting language will disappear entirely.
Observant readers (at least those who aren’t too lazy to do a little math) will have noticed that Malayalam represents a rather striking departure from the tendency I have just noted. This I cannot at present explain. I rather suspect that the number of Malayalam speakers given by Küçük is somewhat inflated. It is likely that there are actually less than 1,000 speakers of this language.
To the Editors of Speculative Grammarian, I say: let’s have no more of these silly and indefensible explanations. Please call on scholarly reviewers who are actually relevant experts. Then perhaps your devoted following of professional linguists will be spared the necessity of reading refutations like this, regarding papers like Küçük’s.