Wednesday, 22 May 2013

Sound Addition, Metathesis, and a bit of Fusion - 'Introduction to Historical Linguistics', Part IV

This is the fourth post in a series on historical linguistics, using Introduction to Historical Linguistics by Terry Crowley and Claire Bowern as the source.  Part I is here.  Part II, here.  Part III, here.

Sound addition can be a very common phenomenon.  We can see it in a lot of loanwords, in particular; think of borrowings into Chinese and Japanese, which typically add vowels to break up consonant clusters (Las Vegas becomes Lasi Weijiasi in Chinese).  In fact, breaking consonant clusters into CV (consonant-vowel) structures is a common tendency in human languages, and so this kind of sound addition isn't at all surprising.  Crowley gives the example of English loanwords in Maori, which, like most Oceanic languages (indeed, most Austronesian languages), strongly tends towards CV structures:

korofa   'golf'
mapi    'map'
kuki    'cook'

The trouble is, when reconstructing past languages, it can be difficult to distinguish between apocope (see the last post) and sound addition (this is one of many reasons why it is necessary to compare as many related languages as possible).  Crowley gives this advice:
One way to tell [...] is to look at the types of vowels that occur at the end of the word.  Can you predict what vowel it will be?  Is it always /a/?  Is it always the same as the second-last vowel?  If so, that is good evidence that there has been an addition of vowels, not a subtraction.  (p. 30).
This is not watertight, of course.  Again, we need to look at as many languages as possible to make an accurate assessment.

Other types of sound addition include:

This is where a consonant is added to create a consonant cluster.  The aim is to make it easier to say, even though it goes against the tendency to generate CV phonology.  For example, Spanish hombre, from Latin hominem, has an excrescent [b].  This occurred also in the English word thimble, which came from *θymle.

Anaptyxis, or epenthesis, is the name for the insertion of a vowel between consonants to break up a consonant cluster.  The Irish pronunciation of film as 'fɪləm is an example of anaptyxis.  Crowley also gives the example of Ukrainian, which has an epenthetic (or anaptyxic) vowel after liquids in consonant clusters (e.g. *dervo  >  dérevo   'tree').

Prothesis is the addition of a sound at the beginning of a word.  Crowley's examples are from Motu, which consistently applies a prothetic [l] before [a], e.g.,

*au  >  lau    'I'     (cf. Malay aku, Seeqiq yaku, Fijian au, etc).

This isn't too common, but it can clearly happen.  This is why historical linguistics is historical and not futurological - retrodictive, not predictive.  And again we see a good reason why extreme time-depth linguistic reconstruction (e.g., 'Eurasiatic', 'Nostratic', 'Amerind', etc) will always be highly speculative.  We just can't predict exactly what language will do or what bottlenecks have occurred, obscuring the presence or absence of certain phones in protolanguages.  We can make accurate retrodictions based on regularities in sound change, but we need accurate data from several closely-related languages.

Incidentally, epenthesis can be used either as a technical name for sound addition as a whole, or more specifically for anaptyxis.

That's pretty much all of sound addition.  I'm not aware of any other kinds.  So let's move on to metathesis, which is a curious and generally sporadic sound change involving the switching around of either vowels or consonants.  Spontaneous metathesis is relatively common - Crowley gives the example of pronouncing relevant as 'revelant', and I can't think of a better example right now - but it isn't such a common sound change in terms of having a wider effect on the language.  Crowley also notes that the word bird is a product of metathesis.  Its earlier pronunciation was [brɪd].

It can occur as a grammatical feature, however, and its use in this way is one of the curious features of the Meto language of West Timor, Indonesia.  For instance, the word ume, 'house' (from proto-Austronesian *Rumaq), is commonly seen as uem.  This is a grammatical feature, not a universal sound change.  The word for 'house' is still ume.  It's just that as an object, it usually appears as uem.

Crowley's historical example also comes from the Austronesian language family.  In the history of Ilokano, one of the many languages of the Philippines, there has been a consistent switching between word-final [s] and initial [t].  Crowley demonstrates the change by comparison with Tagalog:

Tagalog:  taŋis     'cry'
Ilokano:  sa:ŋit

Tagalog:  tubus     'redeem'
Ilokano:  subut

...and so on.  This is comparatively rare - metathesis is generally a sporadic sound change, not a consistent one - but it's evidently possible and therefore something you need to be aware of when reconstructing past languages.

Before we move on to fusion, which is a more complex topic, consider the repercussions of all of this. Imagine that there were once many more languages in Eurasia.  This isn't hard to imagine: before Indo-European expanded in the Eneolithic, both Europe and South Asia must have contained a great many more languages.  Perhaps we can see the relics in Burushaski and Basque and other isolates.  Before Afroasiatic languages became dominant, as well, the Near East was also home to a lot of different languages: Elamite and Sumerian don't appear to have any relatives, and they aren't even related to one another.

It is also clear that there are mechanisms for the spread of language families in non-agricultural, non-pastoral conditions, as we can see with the example of Pama-Nyungan in Australia.  So way back in Epipaleolithic and Mesolithic Eurasia, it is possible that language families scattered themselves across the land and contracted in response to a lot of different variables.  Imagine - and this is what a lot of fringe linguists want you to imagine - that most of the languages of Eurasia are genetically related and descend from the same protolanguage that spread across Eurasia before agriculture, way back in the Pleistocene.

Now, imagine trying to reconstruct this language family, like Eurasiatic or Nostratic or any of these other super-families, using the evidence we have available from languages and families like Burushaski, Basque, Indo-European, and others.  Let's say you look at Burushaski, a language without any close relatives.  Perhaps its parent language was a daughter language of the same super-family shared by Indo-European.  Perhaps.

Without comparative evidence from the same branch, how on earth would we be able to tell?  Any number of sound changes could have occurred in Burushaski or its parent language, obliterating the connection.  How would we know if there had been any consistent or sporadic examples of metathesis?  How would we be able to tell if an epenthetic [a] had been added to break up a consonant cluster without evidence from a more closely related language?  We can only tell that these things have occurred when examining closely related languages like Tagalog and Ilokano.

This is also why we can be unsure of the affiliation of certain languages even though sound is regular.  Sound changes are retrodictable, but there are a lot of possibilities, and we need close relatives to make accurate analyses.  Without these, any reconstruction is entirely speculative and rests only on assuming that the most common sound changes are the ones that have occurred.  This isn't a very reasonable assumption.  Indo-European is so secure because we have written evidence of earlier languages and lots of surviving members of lots of branches; without this, we'd be a little lost, and if we only had English, German, Dutch, French, etc, to compare, we wouldn't have an Indo-European language family.  So in a non-literate situation, linguistic reconstruction is difficult, and ascribing language family affiliation in the absence of written evidence of earlier forms can be a hard task.  Incidentally, it's also a good argument for language conservation and research on endangered languages - they may provide the piece of evidence that can be used to link languages together into families, allowing us to better understand human history.

Anyway, moving on to fusion, very, very briefly.  Fusion is a complex topic because it involves delving deep into the properties of sounds.  Remember: historical linguistics starts with applied phonetics (and phonology).  So we need to introduce, as Crowley does, the concept of feature.  Sounds are made up of features; they are nasal, labial, consonantal, continuant, etc.  A phone can be analysed wholly in terms of its features.  Crowley gives the examples of [m] and [a]:


[+ consonantal]
[+ voiced]
[+ labial]
[+ nasal]


[- consonantal]
[+ voiced]
[+ low]

All sounds can be broken up into features like this.  Fusion is the loss of features in the context of other sounds.  If we had no concept of feature, understanding a lot of different sound changes would be impossible, and many would look completely ridiculous.  So feature is pretty important (and not just for understanding fusion).

French originally had no nasal vowels - or, rather, the ancestor of French had no nasal vowels.  It had none of those characteristic sounds we associate with French, in words like cent, sont, blanc, and so on.  Over time, however, a rule developed:

*vowel + nasal  >  nasalised vowel

For instance:  *yn  >  *ỹ  >  œ̃     'one'  (from Latin ūnus)

(My apologies if this isn't showing up correctly in your browser.)

Instead of being two sounds, [y]  ([- consonantal] and [- nasal]) and [n] ([+ consonantal] and [+ nasal]), we have a new sound - a nasal vowel.  The consonantal feature of the [n] has disappeared, but the nasal feature hasn't.  The non-consonantal feature of the [y] has been retained, but the nasal feature of the [n] has been added.  This is what fusion is.  (There has also been a change in the height of the vowel, which seems to be common with nasal vowels - see also Yoruba, which is also full of nasal vowels.)

This is a relatively trivial case, but there are many permutations.  Distinctive features are not only important in understanding fusion, but also in understanding related phenomena, like fission, assimilation, and dissimilation.  Those, alongside a fuller explanation of fusion, will constitute the next post in the series.

In any case: here's a BBC piece on American/British borrowings and the reasons for change.  Lynne Murphy, a linguist at the University of Sussex, briefly mentions why she thinks sound changes are occurring differently in America and Britain at the moment.

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.