On Dec 31, 2006, at 8:25 PM, Amanda Babcock Furrow wrote:

> I got The Unfolding of Language (Guy Deutscher, 2005, Metropolitan  
> Books, NY)
> for Christmas, and in reading it I thought it would be neat to  
> review it
> from a conlanging viewpoint.

I read it just before Christmas. It was recommended to me first by a  
member of the "linguists" community on Livejournal, when I asked  
there if anyone had information or hypotheses on how Semitic root/ 
template/pattern morphology got started. Subsequently, Donald Boozer  
and Chris Peters on CONLANG also recommended it to me.

> Briefly, this book is not aimed at the average conlang subscriber as I
> conceive of hir, as it assumes no familiarity whatsoever with  
> linguistics.
> This is certainly a good thing for most readers, but a seasoned  
> conlanger
> may feel a thrill of dread when reading a footnote to the word "case"
> reading: "All linguistic terms used in this book are explained in the
> glossary".  Fortunately, the rest of the book is not quite as low- 
> level
> as that footnote had me thinking, or else I adjusted quickly.  The  
> book
> would actually be very much on target for beginning conlangers, who  
> could
> get quite a bit of inspiration from both the examples and the overall
> thrust of the book.

I was a little concerned that it would be too basic, before I started  
reading, but once I got into it there really was a lot of information  
which I didn't know, but probably should have. After checking it out  
from a library, I'm considering buying it to reread and consult later.

> I would note, however, that it felt like fully 50%
> of the text could be excised as it served no purpose other than to  
> chivvy
> the reader along and hammer home whatever point he had just made.

There was a little of that, but luckily for me it didn't seem to slow  
down the reading.

> That said, the book was a goldmine for me of arcane examples, which I
> felt to be its best point and what kept me reading.  If you are  
> already
> familiar with the Akkadian languages, proto-Germanic and PIE, the
> development of Old French from Vulgar Latin, and how the Semitic verb
> system arose, then there remain only random drive-by example sentences
> from several African and the occasional Asian or North American  
> language
> to whet your appetite.

The material on the Semitic verb system (which I believe was Ch. 6  
and one or two of the appendices) was really an eye-opener for me. It  
basically hypothesizes that the cause of nonconcatenative (root/ 
template/pattern) morphology consisted of a few sound changes and a  
lot of analogy and back-formation. It's fairly speculative, but with  
a good dose of (I think) sound reasoning. Its speculativity shouldn't  
turn conlangers off anyway, since, even if the real Semitic languages  
*didn't* develop that way, it can still be used as a good example of  
how to achieve a similar sort of nonconcatenative morphology in a  
conlang, starting from a purely concatenative basis.

This was the most valuable part of the book for me. I have tried  
working out this sort of morphology diachronically starting from a  
purely concatenative system before, and never had much luck. I think  
my lack of success has been due to my concentration on arriving there  
through sound change alone (which Deutscher actually states cannot  
have produced it in the real world). Although I have known about  
analogy for a while (and even heard that it was involved in the  
development of the Semitic system), it has been hard to actually  
*remember* to use it in my conlanging. I think the examples of it  
Deutscher gives will help me to make more use of it.

(I just wish he'd written a whole book about the development of  
Semitic morphology. The few patterns whose origin he postulates are a  
small fraction of the complexity of the mature Semitic system, but as  
he says, just a few patterns were probably enough to get the ball  
rolling, with patterns accumulating via analogy over hundreds or  
thousands of years subsequently. Luckily he lists a lot of sources  
for further reading.)

> The second most valuable feature of this book, to me, was what was  
> intended
> to be its main point, which is the ubiquity of erosion and abstraction
> in creating new affixes or inflections.  I knew that there was a cycle
> of words eroding to affixes, eroding to inflections, and eroding away,
> but this author goes to great lengths to back up his assertion that
> every single morpheme in language came from somewhere: either by  
> eroding
> from an originally more concrete word, or by analogy to what appeared
> to be systematic operations elsewhere in the language.  It gave me a
> feeling that I should be doing a great deal more to provide my  
> conlangs
> with affixes that have a history - even though I do not usually  
> conlang
> diachronically.

I just wish he had given more information on the tendency of more  
frequent words to "wear down" more quickly. The model of sound change  
I've internalized is the Neogrammarian one, which (as I understand  
it) holds that sound changes affect *equally* all words at a given  
point. Although I know that that is no longer believed to be true, it  
is hard for me to break out of that line of thinking -- if I am  
designing a diachronic conlang and decide a given sound change only  
happens in a few words, I feel like I am in some sense "cheating," if  
I don't come up with damn good reasons for the unequal treatment. And  
I believe Neogrammarian sound change is still considered useful in  
comparative and historical linguistics, perhaps the same way that  
Newtonian physics is still useful.

But I digress. I would just like Deutscher to explain about both  
conceptualizations of sound change, and explain what the current  
thinking in linguistics is, and how they relate to each other, and  
perhaps to give some evidence and/or reasoning for *why* certain  
words change more quickly than others. (If anyone out there on the  
list has a good grounding in this, I'd love to hear your thoughts.)

> My least favorite part of the book is the last chapter.  The author
> attempts to show, in a broad sketch, how language as we know it would
> have developed naturally and inevitably from what he calls the "me  
> Tarzan"
> stage to the fully subordinating structure that we have today.  He
> illustrates the progress of his thought experiment using a story about
> a father spearing a mammoth to save his daughter.
> A mammoth?  "Me Tarzan"?  I found this section painful to read; I was
> embarrassed for the author.  I'd rather get more details on the  
> Semitic
> verb system instead.

It's a little silly, especially the label "Me Tarzan" (which, as  
Deutscher notes, was never actually uttered by Tarzan in the book or  
movies). But then I don't know what I would call it, especially if I  
were trying to avoid "scary" technical words.

Overall for me the chapter works pretty well. I hope to use it as a  
blueprint for developing a conlang from a "Me Tarzan" stage some day.  
I wasn't quite convinced about the emergence of embedded clauses with  
finite verbs. I think more information in that area would be good,  
but then perhaps it'd be too technical.

> Overall, I enjoyed the book.  Along with the urge to write a more  
> diachronic
> conlang, it has left me with the feeling that I really need to  
> learn the
> Akkadian languages and Old English, plus probably proto-Germanic  
> and Old
> French.  Too bad I have a baby.  Five appendices allow the author  
> to go
> into more detail than his mainstream audience wants to know about,  
> and the
> endnotes (which I haven't read yet) look like they have more  
> examples and
> details in them.
> tylakehlpe'fo,
> Amanda