Evolution and Grammaticalization

[jess tauber]

Things aren't as random as all that- each new generation takes its cue from the prior one, however imperfectly. This means there is already a 'memory' of prior selection. The historical dimension must be taken into account for both linguistics and genetics.

An example of genetic order recently announced on the science feeds- apparently genes that share common activation are spaced, on DNA, in a periodic fashion over VERY long stretches of nucleotides, with intervening genes that don't share in that particular pattern. This would imply that all one has to do is shift the activation up or downstream a notch to change the 'phase'. Many genes in higher eukaryotes exist in chains of differentiated forms, for instance hemoglobin, which has embryonic, fetal, and other forms for later parts of the human life cycle. My guess would be that many of the 'early' forms of expressed genes for young organisms would pattern together. So one might regulate the entire system temporally just by shifting the packaging (and thus inactivation/activation) on histones and other proteins.

It was already known that the genes which control the layout and development of the metazoan body plan were laid out in linear fashion parallel to that of the body segments.

It has also been recently discussed that chromosomal rearrangements are also not as random as previously believed. Obviously real errors do occur, but when things work smoothly such rearrangements, in the long term, may be more like the shuffling of decks of cards. Given that the linear orders which govern temporal and spatial developments (as laid out above) must needs be maintained for the organism to be viable, one wonders then whether the entire system, as it operates inter- and intra-generationally, in the short and long term, and over short and long space (cells, organs, bodies, families, populations) has been optimized for just such a task.

As mentioned in my first post, genes can be split, appositional, or overlapping. The former have the greatest flexibility, due to higher level editing processes and regulation. You can do a lot to such genes and still have enough play to make a usable product in the end. The overlapping viral genes are 'trapped' by their own efficiency of use of space on nucleic acid chains- one little change in sequence and its all over.

If we look at the linguistic parallels, one sees that the split gene analog, analytic/isolating languages, can vary wildly in the particulars of their lexicons (just take a look for instance at Tibeto-Burman cognate roots, compounds, etc.). 
Such languages are very context-dependent (as are genes in the cells of eukaryotes).

My point is that it may not be one-size-fits-all with regards to 'random' 'mutation'- the multilevel, cosmopolitan system is very robust, with many redundancies constituting 'plans b, c...', etc. If you ignore the tweaking that goes on at higher level, it might look like chaos. But its organized chaos. The simple systems on the other hand are more likely to show nice clear form/function mappings at what in the bigger systems would consitute the lowest hierarchical levels, with little or no tweaking necessary. Tab A fits slot B. End of story. The most flexible systems can create and modify slots and tabs as necessary. But this requires higher level system memory (such as eukaryotic cells have). Some of this isn't even genetic at all (for instance flagellar orientation in the pellicle of various Paramecia, shapes of those pellicles, etc.- these change with direct damage to the pellicle, and are heritable.).

As you may know, my decades long obsession with form/meaning mapping nonarbitrariness (whether sound symbolism, syntactic iconicity, etc.) tends to make me somewhat wary of blanket application of terms like 'random'. When it comes to order, absence of evidence does not necessarily constitute evidence of absence. 

But perhaps that's just me.

Jess Tauber