Sunday, May 30, 2010

The Final (so far) Step in Language Evolution

ResearchBlogging.org A recently published paper[1] has impelled me to discuss a favorite theory of mine, involving the actual stages in which our species language skills evolved.  Or rather, the final stage.  The paper itself, Dissociating neural subsystems for grammar by contrasting word order and inflection (by Aaron J. Newmana, Ted Supallab, Peter Hauserc, Elissa L. Newportb, and Daphne Bavelier), reports the investigation of differential brain region usage in interpreting two different types of language:  inflected and positional.

In inflected language, a word's relationship to the rest of the sentence is determined (communicated) through case markers such as inflections (I'm including agglutinating languages in this class).  For instance, the subject of a sentence or clause is identified by different inflection (such as case endings) from the object of a verb or preposition.  By contrast, in a positional language (or construction within a language) this information is communicated by its position in the sentence, clause, or phrase.

What Newmana et al. have shown is that different regions of the brain are activated when a hearer (seer, actually, in this case since they studied American Sign Language) encounters phrases or clauses determined by inflection vs. word order:

To summarize, we exploited a property of American Sign Language, unique among languages thus far studied with neuroimaging, to directly compare the neural systems involved in sentence processing when grammatical information was conveyed through word order as opposed to inflectional morphology.  Critically, this comparison was made within subjects, while tightly controlling syntactic complexity and semantic content.  Reliance on word order (serial position) cues for resolving grammatical dependencies activated a network of areas related to serial working memory.  In contrast, the presence of inflectional morphology increased activation in a broadly distributed bilateral network featuring the inferior frontal gyri, the anterior lateral temporal lobes, and the basal ganglia, which have been implicated in building and analyzing grammatical structure.  These dissociations are in accord with models of language organization in the brain that attribute specific grammatical functions to distinct neural subregions, but are most consistent with those models that attribute these mapping specificities to the particular cognitive resources required to process various types of linguistic cues.[1]
The Proposed Theory

Now my theory is that the final stage in the evolution of our language skills involve these inflected (and agglutinating) constructions.  Several facts point to this sequence.... (read the rest in the full post)

The simplest form of language is the pidgin, a "simplified language that develops as a means of communication between two or more groups that do not have a language in common".  In its simplest form a pidgin might consist of 2-3 word sentences containing a noun and a verb, with perhaps another noun representing the object of the verb.

A common theory is that many such pidgins develop naturally into creoles, which are usually typified by

  • a lack of inflectional morphology (other than at most two or three inflectional affixes),
  • a lack of tone on monosyllabic words, and
  • a lack of semantically opaque word formation.
(Some exceptions to this standard have been noted.)  This process might take as little as one generation, leading to the widely accepted hypothesis that children come with "hard-wired" expectations of certain features in a language, and, in essence, they look for them hard enough to find them in their parents' pidgin even though it wasn't really there.[5] [A1] 

What I would suggest is that entirely positional languages of this type represented the previous stage of language evolution, with the development of case marking, agreement, and flexible word order as the final evolutionary step.

For any proposed evolutionary step, a selective value must be identified.  In this case, I would propose that, very simply, flexible word order enables substantially improved epic poetry,[2] [3] which in turn enables multi-generational transmission of essential myths, which in their turn can encode selectively valuable historical information.[4] [7]

I'm going to defer discussion of the adaptive value of myths, and even epic, for a moment while we take a look at how languages evolve in the presence of writing.  Although the earlier forms of a language or protolanguage can be inferred from a study of its current structure, most of our information regarding linguistic development comes from written records of some sort, which means that the culture involved had some contact with writing.

Now if writing can carry information across long time periods, it can to some extent replace epic poetry and myth, reducing the need for full flexibility of word order.  This being the case, we should expect the languages of cultures using writing (or in close contact with other cultures that use writing) to show a trend of replacing fully flexible word order constructions with more word-order dependent constructions.  And so it is, especially in that most-studied of language groups, Indo-European ("IE"): 

It has long been noticed that there are trends in language change, such that certain types of development occur often and in unrelated languages.  For instance, English is one of many languages that have formed future markers from a verb of motion.  The development of Indo-European adverbial particles to adpositions, apparently independently in its daughter languages, results from reanalysis of underlying structures and is a very early development of configurational syntax in the language family.[2]
Thus, we can see a general trend from case-marked (inflected) constructions to systems dependent on word order.  In English this process has been carried almost to completion.  We must note here that the only languages we know about in this group are those of cultures that had, or were in close contact with somebody else who had, writing.  I would predict that any IE languages that remained completely isolated from writing likely retained the fully flexible constructions of the parent language, although until we find them in contact with somebody capable of writing down reasonably large amounts of them we wouldn't know about them, and at that time we would find them already evolving towards more positional constructions.

Since the work of Lord and Parry, it's been recognized (with some debate) that the epic in its original entirely oral form was composed "on the fly" for a specific performance.[A2]  The actual basic story might stay the same, but each performance was a unique, one-time, composition made up of "formulas", "a formula being 'an expression which is regularly used, under the same metrical conditions, to express a particular essential idea'[3]". 

The language of epic was thus not an actual spoken dialect but a conventionalised form that developed in a manner typical of orally transmitted poetry and later became a prestigious literary variety.[2]
We can see, here, that the complete flexibility of word order was a tremendous help in creating formulas to fit any necessary metrical/rhyme pattern.  By the time the Arcado/Cypriot dialect of Proto-Greek was being written down the loss of this flexibility was already under weigh, although the epic traditions in old Ionia (what later came to be called Achaea) may have retained the older form. 

The older IE [Indo-European] system used case to indicate spatial relations, and any accompanying P-word [prepositional and preverbal particles] added meaning without taking over the case function; but both Mycenaean and classical Greek increasingly used prepositional phrases with the preposition governing the noun phrase and determining its case, even though (from a diachronic perspective) it had been the case functions which originally determined what cases a preposition would govern.  In Homer both systems (independent case and configurational syntax) are still present, and [...] a [...] strictly synchronic view of the data is possible if the variation is viewed as the normal outcome of grammaticalisation processes, which typically generate changes which may coexist with the original constructions.  Greek speakers must have been able to recognise and produce both free and syntactically restricted uses of P-words for some time while the reanalysis of the constructions was taking place, just as English speakers can use ‘going to’ in two different syntactic constructions.[2]
Now, it's important to realize that this shift is exactly the one studied by Newmana et al.  It involves the switch from a system that "increased activation in a broadly distributed bilateral network featuring the inferior frontal gyri, the anterior lateral temporal lobes, and the basal ganglia," to one that "activated a network of areas related to serial working memory."  (Since word-order is an essential part of prepositional (and postpositional) phrases.) 

I would argue that as the presence of writing reduced the dependence on epic for essential long-term cultural memory, languages relaxed into the easier, because older and more thoroughly evolved, positional constructions.  This relaxation, added to the fact that pidgins and (most, if not all) creoles tend to be positional, point to it being the older, original mode.  Indeed, the fact that a well developed temporally serial working memory would have been essential for complex movement in the arboreal environment means that all the building blocks would have been there for many tens of millions of years; the evolution of language simply had to tie to these pre-existing (and mostly pre-adapted) features. 

By contrast, the brain structures and circuits needed for inflected language constructions may well have been brand new, or at least substantially adapted from some pre-existing system for handling complex transformations. 

At this point we need to consider what selective advantage the capacity for epic poetry, and the myth it transmitted, offered our evolving ancestors. 

The Adaptive Value of Myth

Any modern study of myth, IMO, should start with When They Severed Earth from Sky:  How the Human Mind Shapes Myth by Elizabeth Wayland Barber & Paul T. Barber.  Quoting from the first chapter:

[I]f people were so smart--just like us--100,000 years ago, why do the myths they passed down often seem so preposterous to us? And not just to us.  Even ancients like the Greek poet Pindar, who made his living telling such stories ca. 500 B.C., sometimes felt constrained to a disclaimer: "Don't blame me for this tale!" The narrators present these myths as "histories".  Yet how can we seriously believe that Perseus turned people to stone by showing them the snaky-locked head of a monster, or that a man named Herakles (or Hercules) held up the sky for a while, slew a nine-headed water monster, moved rivers around, and carried a three-headed dog up from the land of the dead? Or that a man named Methuselah lived for almost a millennium? That an eagle pecked for years at the liver of a god tied to a mountain, or that mortal men--Beowulf, St. George, Siegfried, and Perseus included--actually fought dragons? And how can one view people like the Greeks or the Egyptians, who each believed simultaneously in three or four sun gods, as having intelligence? Didn't they notice a contradiction there? Why did people in so many cultures spend so much time and attention on these collections of quaint stories that we know of as "myths"?

The problem lies not in differing intelligence but in differing resources for the storage and transmission of data.  Quite simply, before writing, myths had to serve as transmission systems for information deemed important; but because we--now that we have writing--have forgotten how nonliterate people stored and transmitted information and why it was done that way, we have lost track of how to decode the information often densely compressed into these stories, and they appear to us as mostly gibberish. And so we often dismiss them as silly or try to reinterpret them with psychobabble.  As folklorist Adrienne Mayor points out, classicists in particular "tend to read myth as fictional literature, not as natural history" [Mayor 2000b, 192]--not least because humanists typically don't study sciences like geology, palaeontology, and astronomy, and so don't recognize the data.  [my emphasis]
One of the most striking examples they give of long-term preservation of information involves the collapse of Mount Mazama to create Crater Lake in Oregon.  To briefly summarize, a major underground spirit fell in love with a local girl and when his demand for marriage was rejected had a temper tantrum (and battle with the great spirit in the sky) that blew the top off the mountain, resulting in the present caldera lake

Geological analysis confirms that there was once a mountain on that spot, and that it erupted violently, spewing around 50 km3 of magma, ash, and lava-bombs until the emptying of its magma chamber caused the caldera walls to collapse inward, forming a pit some 4000 feet deep that later filled with water ([refs]), just as the myth says.  Since the eruption happened almost 7700 years ago ([ref]), this myth must have been carried down for nearly eight millennia.

Our own (typical) assumption, as we read something like the Klamath myth, is that since we do not agree with the Klamath explanation for this fiery occurrence, there is nothing worth looking at scientifically in the story.  But one of our problems as modern observers of myth (or even observers of events such as car accidents) is that people tend to present their observations and their assumed explanations all tangled up together.  On the other hand, if we strip away the explanations proffered but keep and investigate the observations, we can see that the observations in myths are fairly accurate (as far as they go), and at the very least they alert us to something of geological interest that happened in a particular place.  Furthermore, if we take for the moment the Kalamath step of assuming that the Curse of Fire was caused by a wilful being (more of this below), then we can see that the quite logical strategy is to placate that being—with a gift, bribe, or sacrifice—which is exactly what they did in their attempt to prevent or delay future destructive eruptions.  That is, the myth unrolls logically from its own premises—it is not haphazard.  In fact, there are many myths concerning geological events in the Pacific Northwest ([ref]), where until the nineteenth century the population remained stable, that is unreplaced by cultures that had not witnessed the events and therefore did not know what was referred to.[7]
Note that these mythic memories don't need to be geological: 

An example that can serve to illustrate the historical reality lying behind the mythological narration is provided by the famous combat between Heracles and the Hydra of Lerna.  The analysis of this famous story deserves some attention because it can provide useful insight regarding the origin and factual basis of a myth, as well as other mechanisms of myth-making ([ref]).

The slaying of the Hydra has been one of the myths most widely considered, since antiquity, to rest on natural processes.  The always regenerating many heads of the Hydra have been interpreted as a symbol of the many water-sources feeding the large swamps near Lerna, and the struggle between Hercules and the monster therefore an image of the draining effort.  After finally chopping her main ‘head’, said to be immortal, the hero buried it forever, putting a huge and heavy rock on it.  Kirk (1974), following an interpretation first proposed by Palaephatus, maintains instead that this myth more likely records ancient political events.  In a manner similar to the killing of the Minotaur in the Palace of Knossos, the killing of the Hydra at Lerna, as well as the related myth about the killing of the Nemean lion (the first two labours of Heracles, the Mycenaean hero), seems to contain memories of ancient political events in addition to references about fertility rites.

Strong connections are known to have existed between Lerna until the Early Bronze Age (Lerna III), and the Cretan civilization.  The end of Lerna III was in part evidently due to the invasion of the Indo-European Greeks in c. 2200 BC.  These patriarchal Indo-European-speaking invaders, from whom later the Mycenaeans would originate, marked the end of the Early Bronze Age in many areas of the East Mediterranean.  According to typical Minoan settlement patterns, the political and religious centre and the ‘head’ of the local community, would have been the Palace of Lerna (‘House of Tiles’).  The destruction of the Lernean Palace (2300–2200 BC) is marked by the peculiar singularity, seemingly unique in the whole of Greece, that the Palace was buried by the conquerors under an enormous funerary tumulus ([ref]), considered nevertheless an enigma by archaeologists because it contains no tombs.

This unusual tumulus, deliberately positioned above the ‘head’ of the defeated society, strictly corresponds to the huge mythological rock placed by Heracles above the head of the beast ([ref]).  As such, the facts described by tradition largely coincide with what can be observed on the site.  Even the position of the buried Palace, corresponds to the location of the head of the Hydra, buried in the myth on the side of the road to Elaeus.  The mythological account can therefore be regarded as quasi-historical, recalling an Early Bronze Age phase of the Mycenean conquest of the Greek mainland against the Lernean Minoan related settlement.  The seeming truth behind the myth, and the relevance of the tumulus itself, apparently was already forgotten by the end of the Middle Helladic period (c. eighteenth–seventeenth century BC), as indicated by the fact that the tumulus was then reoccupied by the village after being left untouched for nearly 500 years.  We can thus consider this date as the moment when the local historical memory transmitted by oral tradition became a new myth as transmitted by Hesiod, Ovid, Apollodorus and other ancient writers, because the politico-religious factual story lying behind the myth had been forgotten.[7]
This long-term cultural memory isn't limited to historic events, either.  For example, in the Iliad, the process of sacrifice and sacrificial meal is twice described in almost identical language (1.458-469 and 2.421-432).  The entire description of the process may have been a single formula. 

While we today may see little value in the long-term retention of sacrificial processes, the same technique could have been used for the hunting of animals that are temporarily unavailable, the killing of "monsters" (rare large predators that are making nuisances of themselves), even the manufacture of weapons and other tools requiring materials temporarily unavailable.  Even if the precise process has been distorted in the centuries since its last use, it would have given a creative inventor a major "leg up" in recreating it.

Overall, I think we can regard the long-term cultural memory process of myths as offering a major benefit to the population that possesses it.

Epic as a Carrier of Myth

The final link in the chain is either the easiest or the hardest:  epic poetry as the primary and most effective carrier of myth.  It's generally accepted that poetical systems such as meter, alliteration, assonance, and rhyme, aid in memorizing both complete poems and the poetic formulas used in longer epics.[3]  There is also evidence that these techniques add to the acceptance of the message by the audience.[9]  The use of the formulas in epic, in turn, eases the "on-the-fly" composition during performance.  If the story-teller wants to say "next morning", but instead recites "when rosy-fingered dawn the early one appeared", there's much more time for composing the next portion of the performance.  The latter formula would fit metric patterns the former would not.  Because it's a formula, it doesn't have to actually be composed on the fly, rather the story-teller pulls it from his (or perhaps her) mental toolbox based on its "fit" within the metrical pattern and recites it "on autopilot" while thinking about the following formulas.

We have, then, the following logical sequence: 

  • Poetic tools (meter, etc.) aid in memorization and audience acceptance,
  • composing epic "on the fly" requires a large toolbox of "formulas" with tight requirements to fit into metrical (and other) slots,
  • fully flexible word order allows a much larger collection of "formulas",
  • the use of inflection allows flexible word order,
  • therefore the capacity to use and understand inflected constructions provides the most effective transmission of any story (mythic or otherwise).
Of course, this sequence is hardly "proven", and testing it probably involves some serious challenges given the widespread influence of written material on just about everybody.

Nevertheless, it provides a coherent model for the development of inflection and the associated tools of flexible word order as an evolutionary adaptation, allowing us to perhaps glimpse the evolutionary precursor to our modern language skills.

Appendices:

A1  "... widely accepted hypothesis that children come with "hard-wired" expectations of certain features in a language":  A brief Google search or perusal of the Wiki articles will find many mentions of opposition to this idea.  I'm not going to get into this much, IMO this opposition represents the last-ditch defense of the discredited[6] "blank slate"[8] concept of human cognition, and thus falls into the same class with the last-ditch opposition to plate tectonics in the '70's, religious opposition to evolution, and the more specious of attempts to "falsify" the "Greenhouse effect".

(Yes, I know some people will feel their sacred oxen have been gored here, but science is about finding the facts, and IMO twisting the facts to support an outdated Marxist ideology of how human cognition works is just as bad as twisting them to support Biblical literalism.)

A2  "... the epic in its original entirely oral form was composed "on the fly" for a specific performance":  I'm referring here to conditions before any contact with writing had taken place.  Even writing in another culture in close contact with the one under discussion.  Obviously, we can only project what this would have been like, since writing of some sort was necessary to preserve the literature itself, at least until Milman Parry began his studies recording oral performances,[3]  and of course writing was already present in the culture he studied.  The introduction of writing not only (potentially) replaces some of the essential purpose (value) of epic and other poetry, it permits the permanent preservation of a specific poetic composition, making it available for rote memorization.

It seems likely to me that the appearance of the written Iliad, as well as any other recorded (in writing) performances contemporaneous with it, would have seriously distorted the entire formulaic tradition, potentially distorting our model of what the original, truly pre-literate, epic tradition was like.  Barry Powell has suggested that the Greek alphabet was specifically created to record the work of Homer,[10] especially the Iliad, and while that may be plausible, I find it more likely that the alphabet was created to record the more typical shorter performances found in the Little Iliad and the other poems of the "Epic Cycle".

These performances would have lasted about 2 hours at full speed; if they were dictated at 1/3 speed an entire dictation session would have required about 6 hours, a good day's work for an aoidos and his scribe.  Based on Parry's research, only the more skilled of aoidoi would have been able to do this.  Alternatively, the inventor of the Greek alphabet might have developed some sort of shorthand allowing him to write at full performance speed.

In the beginning, I suspect that recitals of these dictated poems would have been much less popular than true oral performance:  the most likely scenario is a ship captain or other traveler who carried his written texts with him and performed them in locales where (and when) no true aoidos (or rhapsode) was available.  Or perhaps he was carrying stories popular in the eastern Aegean Sea (i.e. Chios) to the Ionian settlements in the West

Once the Greek alphabet was created, and in use (if perhaps only by one man), it seems entirely plausible that a particularly skilled aoidos ("Homer") undertook to create a special masterpiece good for a 20-hour performance.  Thus, the Iliad.  If the same aoidos also dictated the Odyssey, it seems likely that it was many years later.

We know even less about the process by which the recording of Mesopotamian and Babylonian mythologies interacted with the presence of writing.


Newman, A., Supalla, T., Hauser, P., Newport, E., & Bavelier, D. (2010). Dissociating neural subsystems for grammar by contrasting word order and inflection Proceedings of the National Academy of Sciences, 107 (16), 7539-7544 DOI: 10.1073/pnas.1003174107

Links:

1.  Dissociating neural subsystems for grammar by contrasting word order and inflection

2.  Prepositions and preverbs in Hellenistic Greek

3.  The Singer of Tales by Albert Lord

4.  When They Severed Earth from Sky:  How the Human Mind Shapes Myth by Elizabeth Wayland Barber & Paul T. Barber

5.  The language instinct: how the mind creates language (page 21) by Steven Pinker

6.  The fateful hoaxing of Margaret Mead: a historical analysis of her Samoan Research by Derek Freeman

7.  Exploring the nature of myth and its role in science

8.  The Blank Slate by Steven Pinker

9.  Birds of a feather flock conjointly (?): Rhyme as reason in aphorisms

10.  Homer and the origin of the Greek alphabet By Barry B. Powell


8 comments:

  1. Hi,

    I v. much like your post :

    Tuesday, July 14, 2009
    Concepts, Cognition, and Anthropomorphism

    wh. deals with some extremely important stuff. I too think - if i understand you - that concepts are represented in some kind of outline/ideogram mode in the brain.

    However, as I understand you, you seem to think that these outlines are relatively fixed:

    "it's a good guess that those neurons occupy a specific location in some n-dimensional "concept space" that corresponds to the specific act involved."

    I however believe that those concept outlines in the brain must be extremely rough and **fluid**. Bear in mind for a start that the concepts often are derived from moving objects.

    They are there IMO to be fleshed in - in extremely flexible fashion, according to the situation - and particular application of the concept.

    I am v. grateful to you for linking concepts to direct physical action, as you do. This is clearly an essential way of approaching concepts - as opposed to beginning with v. airy abstract concepts, (such as "ideas","politics").

    And physical actions demonstrate the need for extremely fluid concept outlines. We think both proactively and reactively, for example in terms of "hold the cup", "eat/ chew your food", "catch the book", "step on the ball" etc etc.

    The concept outline for "hold" must be capable of being instantiated in a virtually infinite diversity of hand forms according to the object held. Ditto physical actions like "eating"/"chewing"/ "catching" will have to take vastly diverse physical, muscular forms according to their relevant objects. Stepping on a ball will be radically different to stepping on an ordinary stone step.

    P.S. Who are you? I can't find any bio details.

    ReplyDelete
  2. Thanks for your comment. (Sorry about the time it took to show up, my available time and access to on-line activity are spotty.)

    First I should mention that I strongly suspect that there are many different modes of representing "concepts" in the brain, some of which are probably very stable, changing typically over a time-frame of years (in humans). Others probably change more quickly.

    The basic (slowly changing) concept representations probably are used in more abstract thinking, while the more fluid one(s) lie closer to the instantiation and interpretation end of the spectrum.

    For me, the best representation of a "primitive concept" is that of picking up something off the ground in (e.g. Rhesus) monkeys. It's known that there are several functions within the monkey's brain that act very similarly both when the monkey takes such an action, and when it sees another monkey take the action.

    Picking something up off the ground would be a fairly abstract concept, with more concrete representations involving specific items, specific locations, and specific social consequences, the latter being partly determined by which other monkeys are present to see the action. (Note that monkeys in general don't seem to have an awareness of other monkeys' points of view (how much they can see of the action), but great apes have been (probably) shown to have such abilities.)

    Another good example would be grasping a branch during brachiating travel. This would be closer to your discussion of variability, but again at some level the "concept" would have to be very stable, so as to encode valuable knowledge about how branches in general behave when grasped by a moving animal. This fits well with the high level of connectivity between the pre-frontal lobe and the cerebellum.

    In general, I suspect the fluidity you mention is more a matter of neural firing patterns, while the stability I'm talking about is in the behavior that produces any immediate firing pattern.

    As for bio, I prefer what I say to stand on its own merits.

    ReplyDelete
  3. This is a very good blog!!
    Computer-based technologies for Neurology Transcription offer a number of practical advantages. Therefore medical transcription companies keep specially trained transcribers to provide transcription services to a number of specialties.

    ReplyDelete
  4. Good post, AK! I'm also of the opinion that 'Ameslan' has an important role to play.

    ReplyDelete
  5. I find the development of languages fascinating. There are things we cannot explain like how there are smaller groups who are gepgraphically remote from each other and yet share common features in the in languages. There is the theory that some of out langiage ability is somehow hard wired. Its an absorbing subject.

    ReplyDelete
  6. Thanks for your sharing, i am looking forward for more knowledge and hope next time i will get more ideas!

    ReplyDelete
  7. I will use few of your information on my site as it is very interesting and good contents.I hope you will allow me to use it.

    ReplyDelete
  8. Nice site. I have added it to my bookmarks and will be back soon. Keep the articles coming. Mesothelioma Cancer Blog

    ReplyDelete