Monday, 13 May 2013

'An Introduction to Historical Linguistics' - Terry Crowley and Claire Bowern. Part 1: Introduction

This is the first part in a series on historical linguistics, using Crowley and Bowern's book as a base.  Updates will be posted until the book is finished.


One of the best books I read as a graduate student was a book I read on my own initiative with no compulsion from any source.  It appeared on none of the reading lists for any of my courses and I didn't read it to improve my exam scores.  (Actually, this applies to almost all of the reading I did at Oxford!)  This book was An Introduction to Historical Linguistics by the late Terry Crowley, one of the doyens of Pacific linguistics, and Claire Bowern, a linguist at Yale who was primarily responsible for editing and updating the work in the wake of Crowley's untimely death.  I read it because I thought it would be a good introduction to the discipline, which is in turn important for understanding humanity and prehistory.  Having now read a couple of other works on the same subject, I have to say that it's still the most readable book for the novice.

Historical linguistics is a fantastic discipline and one that is exceptionally under-rated.  It is the victim of many amateurish studies and its findings are seldom taken all that seriously by non-linguists - and those that are are often the fringe elements.  As these findings can have an enormous impact on our understanding of prehistory in almost every part of the planet, it is vital to have a bit of a background in the subject if you want to speak knowledgeably about the really ancient past.

The reason historical linguistics is so important is because it is a well-established population science, meaning that we can infer historical relationships and activities from linguistic data.  If two groups of people speak languages that are clearly related, we can infer that they share some kind of history.  That's a very useful thing.  In fact, linguistics is such a reliable indicator of shared history that some archaeologists and other non-linguists use language families (more about these later) as hooks on which to hang their theories - 'Austronesian' migration into southeast Asia with rice farming, etc.  They're generally pretty reliable hooks.

I really want anthropologists, archaeologists, and geneticists to be conversant with this branch of the human sciences.  I also believe that learning is the purpose of life, and that a new tool for learning new things is a gift that keeps on giving.  Do all you can to encourage learning and knowledge, that's my view.  So, what I'm going to do here is to re-read Crowley and Bowern's book and write about it for you.  I'm using the fourth edition, published by OUP in 2010, for this series.  I'm going to go through it chapter-by-chapter, noting the important ideas and key terms in each section.  As this is an introductory text, and a summary of an introductory text at that, you shouldn't expect it to be in-depth, but it should be comprehensive, in that all relevant topics will be covered.  It should serve as a mini-introduction to an introduction.  An amuse-bouche for historical linguistics.

Let's begin.

In the first chapter, the introduction, Crowley discusses what historical linguistics is, how it works, and what historical linguists do.  This is all fairly easy stuff: historical linguistics is the study of language change and of how languages relate to one another.  We can then use the information from this to see how different populations relate to one another.  What we are looking for is systematic similarities between languages, not isolated words or sounds, and we can use these to establish degrees of relationship between separate languages.

Crowley discusses the insights of famed early-twentieth century linguist Ferdinand de Saussure, who established a distinction between synchronic linguistics (the study of language as a system as spoken at a particular point in time) and diachronic linguistics (the study of language change over time).  Synchronic linguists are your typical linguists, working on the structure of language and the nature of utterances in a particular language.  They look at noun phrases and verb phrases and particles and other signifying bits and logical bobs.  Diachronic linguists are those people who look into the history of language and determine the relationships between languages, which can also be analysed synchronically as discrete, functioning systems.  Diachronic linguistics just is historical linguistics - the study of language change.

This insight of Saussure's was based on the notion that the grammar of a language could be studied independently of its history and that grammatical explanation could effectively stand alone.  Grammars are systematic and logical and do not depend all that much on historical circumstance, at least not in the same way that lexicons (vocabularies) do.  The history of language is therefore a separate subject from the study of grammar, or it can be considered separately for academic purposes.  So we can separate historical linguistics from the kind of thing Chomsky et al do, even if only for the purposes of the division of labour.

Saussure is also relevant to historical linguistics because he noted the lack of a necessary connection between the sound of a word and its meaning - in other words, that the relationship between a dog and the word 'dog' is arbitrary.  There is no connection between sound, or any other kind of sign, and the meaning it is ascribed.  There is no natural meaning inherent in the sounds we utter, even when it comes to onomatopoeia.  Strictly speaking, this wasn't Saussure's observation, and can be found in Plato (Cratylus, I think).  But Saussure introduced the concept of the arbitrary nature of signs to modern linguists.  The concept is vital to historical linguistics because it shows that history, rather than a presupposed 'natural' meaning, is behind the development of words.

The next figure to be mentioned is William Jones, a British judge in colonial India.  Jones is mentioned in every single text on historical linguistics, largely because his main insight was profoundly important in the development of the discipline even if it wasn't strictly original.  The insight is this: languages diverge from common sources through changes in grammar, lexicon, and pronunciation.  Before English and German there was a common language ancestral to both.  We now call the common sources of groups of languages protolanguages.  The languages that share a protolanguage are said to be genetically related members of the same language family. Jones didn't use this terminology, of course.

Jones worked in Calcutta and regularly dealt with Indian legal cases concerning Indian litigants.  These required an understanding of the basis of legal argumentation in Sanskrit, which Jones proceeded to learn with the assistance of a local pandit.  Given that he was classically educated, Jones saw the obvious similarities between Sanskrit, Greek, and Latin and proclaimed that they all derived from a single common source.  Earlier linguistic commentators had attempted to derive modern and ancient languages from another existing language, like Hebrew, but Jones a) saw that related languages must have a shared source instead of deriving simplistically one from the other and b) he used morphological evidence to demonstrate the connection (primarily noun declensions and verb conjugations - affixes and grammatical particles instead of words).  These were relatively new ideas, although some had been presaged.

It is also worth mentioning, as the Indian archaeologist Dilip Chakrabarti does in India: an archaeological history (OUP 1999), that in Jones' third lecture on the second of February 1786, in which he outlined his view that Latin, Greek, and Sanskrit all came from a common source (correct), he also claimed that Japanese and some Peruvian languages came from the same source (incorrect).  Suffice it to say that linguists no longer believe that Japanese and Aymara shared a protolanguage, either with each other or with Sanskrit and Greek.  So perhaps we shouldn't see him as far above the average amateurish linguist despite the apparent originality of his observation.

Anyway, this sets us up for an important point: the fact that languages diverge from protolanguages in systematic ways that can be discerned through grammatical or lexical analysis.  Protolanguages therefore don't die, but morph into new forms.  The change is so gradual that no one notices it.  This has the implication that Latin never became extinct 'in the wild', as it were; it merely became Romanian, Portuguese, French, and so on, which are its genetically related descendants (Latin is a named protolanguage!).  Just as a baby doesn't die in order to become a child, so Latin did not die before becoming French.  After some time, it simply happened that French speakers noticed that what they were saying sounded different to the Latin words on the page.  Therefore, linguistic change is usually  gradual, although it can sometimes be perceptible, especially if we have recording devices of some kind, like paper and ink.

Next we come to theories of language change.  The first to be mentioned are those that are clearly spurious: cultural temperament causing languages to sound 'softer' or 'harsher', whatever those words mean; tough climatic conditions making language sound rough and aggressive; etc.  No need to argue with those.  The good ones are these:

Substratum Theory
This is the view that languages change due to contact with other languages.  As a language picks up more speakers who previously spoke other languages, the prior language of each speaker affects how they speak the language they have just learnt, an effect which is then passed on to the next generation.  Learn English as a German speaker and you'll have a tendency not only to pronounce English in a German way, but also to use certain lexical items and grammatical constructions, due to their feeling simpler and more natural to you.  Pass this tendency on to your children and bam!, you've created a substrate effect with German as the substrate (and English as the superstratum).

There are some demonstrable substrate effects in plenty of languages, and this is a reasonable view.  However, it fails to explain everything.  We'll come back to substrates later in the series (English shows the effect of several substrates and superstrates, by the way).

Local Identification
This theory says that languages change due to the desire to differentiate an in-group from an out-group.  Teenagers use slang their parents don't in order to differentiate themselves; some speakers of British English use certain words and structures to differentiate themselves from Americans; working class men speak with a certain accent to differentiate themselves from the effeminate middle classes.

This view is historical linguistics as sociolinguistics.  This mechanism clearly plays a role, but it seems to be only one of many determining factors in language change.  We therefore need a few more theories.

Functional Need
This is a less well-established theory that posits that languages change due to the need for new ideas and structures in response to new circumstances: we need a word for televisions and quantum computing because we have those things now.  This is a reasonable point, but actually vocabulary can be quite flexible, and Crowley provides a number of examples from Sye, a language of Vanuatu, to demonstrate the fact that there's no more need for Sye speakers to have these words than for English speakers to have them.  For instance:

elantvi    'complain unjustifiably that something is insufficient or not good enough'

orvalei    'touch something unpleasantly soft or mushy'

So we can dispense with this theory as the primary determinant of language change.

Simplification
This is the theory I back, to a point.  Crowley doesn't.  The idea is that people optimise their utterances - they make them simpler in order to communicate more easily.

As Crowley points out, it is difficult to define 'simplicity' in linguistic terms.  For instance, the change from a CVCV structure (consonant-vowel-consonant-vowel) to a CCV structure (consonant-consonant-vowel) could be seen as simplification (ie, there is one less sound in the utterance), but so could change from a CCV structure to a CVCV structure (this might be easier to say, and therefore 'simpler').  The word for Sumatra in Indonesia is, unsurprisingly, Sumatra, but you can also encounter Sumatera, with a soft vowel (a schwa, as in the '-er' of the word 'player') inserted between /t/ and /r/.  Which is simpler?  Neither.  Since we can't define simplicity, we can't say that language tends towards simplicity.

...Except that we do have an absolute standard of simplicity.  It comes from information theory.  Simplicity of communication would be the ability to convey or store the same amount of information - the same bits - in the smallest amount of stuff (ie, the shortest utterance).  In pragmatics ('the study of how people make sense of each other linguistically', according to Pragmatics by George Yule (OUP 1996)), this is known as 'relevance theory', the idea, proposed by Dan Sperber and Deirdre Wilson, that humans maximise the efficiency of their communication in terms of bits of information communicated versus time and effort taken to communicate it (the 'effort/effect ratio').

In general, people do indeed maximise their communication: if someone asked you, 'how many beers have you drunk this evening?', you wouldn't say 'I have drunk fifteen beers', unless you really wanted to be explicit about it.  You'd say, 'fifteen'.  You instinctively maximise the efficiency of your communication without consciously thinking about it (which has significant implications for theories of human action).  Even anaphora, which includes the use of pronouns and other referential devices, and which is found in all human languages (AFAIK!), is an example of this efficiency in communication.

We'd need to look at overall theories of human action in order to see how this relates to the other theories of language change, but in general, caeteris paribus, humans do seek to maximise the efficiency of their utterances in terms of communicating the same information in the easiest sequence of sounds, and this has clear effects on the direction of language change.  We will see this more as we go through the book.  (I actually will look at how this relates to overall theories of human action later on, looking especially at the views of the American philosopher Donald Davidson on human practical reason, but this is a topic for another time.)

The final theory Crowley and Bowern mention is:

Structural Pressure
Languages have all kinds of structures.  They have syntactic structures, phonological structures, etc.  I hope you've seen the neat little trees of noun phrases and verb phrases linguists produce, because they give some idea of what we're talking about here.  Language practically is structure: it's a series of units that relate to one another in systematic ways.  So linguistic structures ought to be able to influence the direction of linguistic change, given their ubiquity.  If the linguistic system becomes unbalanced in some way, people compensate with innovation.

To see what this means, look at an English sentence, like:

I am washing my goat.

This is in the present continuous (or present progressive) tense.  It indicates, among other things, that the action, the goat-washing, is happening now.

Given this structure, and given the fact that English already distinguishes between events happening right now, for which we use the present continuous, and those that happen regularly or habitually, for which we use the present simple tense ('I wash my goat', etc), we might expect that English would have a range of other forms to express different kinds of continuous action.  And indeed it does:

I was washing my goat.

I have been washing my goat.

I will be washing my goat.

These differentiate past continuous, present perfect continuous, and future continuous actions from simple ones.  They are necessary because of the pressure from the present continuous.

The evidence that structural pressure changes language comes from the fact that languages do exhibit some kind of structural equilibrium.  English has a present continuous; it also has a past continuous.  French has no present continuous and there is consequently no structural pressure to generate a past continuous.

It's the same with the phonological system.  A language with a five-vowel system (/a,e,i,o,u/), to use Crowley's example, is phonologically balanced.  If the /e/ changes to become /i/ in every instance, it will become phonologically unbalanced, as the contrast between back and front vowels will remain but there will be only one mid-vowel.

Four-vowel systems (/a,i,o,u/) are rare, while balanced three- and five- vowel systems are not.  This is evidence that structural pressure is significant.  However, there are some languages that show that its influence is not the primary determining factor.  Plenty of them, in fact.  Crowley lays out the evidence from Motu, a language of New Guinea, to show that structural pressure is not always an important force.

And that's all of the theories.  Crowley and Bowern close the chapter with a discussion of attitudes to language change - primarily the fact that most people are uncomfortable with perceptible language change and tend to see earlier forms of language as better in some way.

To sum up:
-->  Saussure inaugurated the division between the study of languages as systems and languages as historical products
--> Earlier linguists, including amateurs like William Jones, noted that languages change slowly from ancestral forms, and that related languages descend from protolanguages
--> A language that bore descendants, like Latin, is not truly dead, but has merely undergone massive changes
--> The most reliable theories of language change are based on the properties of language and communication rather than race, culture, or climate
--> Efficiency, structural pressure, social considerations of various types, and influence from other languages can all have impacts on the direction and pace of language change

Tune in next time for Chapter 2, 'Types of Sound Change'.

1 comment:

You can post anonymously if you really want to, but I would appreciate it if you could provide some means of identifying who you are, if only for the purpose of knowing who has written what.