Thursday, 27 February 2014

Moten Part XII: Irregularities and Exceptions

Okay, so I know this series contains about as many apologies as it contains conlang info, but please humour me again. At the end of last time's post, I explained that the next post would be about syntax (and more importantly, I promised I would have an actual text in Moten to present as example). However, although I've been working for months on that post, it's still nowhere near ready (syntax is hard!). So rather than leaving you hanging for even longer, I decided to write another post I was planning for later, one which should hopefully be somewhat shorter, yet still entertaining. This post is about those wonderful pearls of natural languages which are the bane of so many language students: irregularities and exceptions!


Before we start, let's precisely define the words we're using: what do we mean when we speak of irregularities or exceptions in a language?

As anyone who's spent a modicum of time at school knows (and many who haven't know as well), languages obey rules (what we basically call their grammar). Well, personally I think it would be more accurate to say that languages are described by rules, but the distinction is irrelevant for this post. Suffice to say, you can always describe the behaviour of a language in terms of a certain amount of rules. They may be quite complex, but there just doesn't seem to be any language that cannot be described with grammatical rules of some kind.

Nevertheless, however precise you make those rules, however much you analyse the language and describe it in details, there always seems to be some kind of irreducible kernel of forms or structures that keep on defying those rules. It's true of all natural languages that we know of (that is to say, all languages we have a description of): however detailed the rules used to describe them are, there are always words or constructions that don't follow these. They are what we call irregularities, or exceptions.

Now, not all languages are equal when it comes to exceptions. Some have plenty of exceptional forms in various areas of their grammars (I'm looking at you, English, with you irregular verbs, irregular plurals, unpredictable use of prepositions with verbs, etc. My own native tongue French is also guilty of this, especially in its verb conjugations!), while others only have a few exceptions and the rest of their grammars is otherwise perfectly regular (I've read for instance that Turkish only has a few irregular verbs, and even then those all exhibit the same kind of irregularity. And I know by experience that Japanese has a very regular grammar, with few exceptional forms). But the one rule that truly seems to hold for all natural languages, without a single exception, is that natural languages have exceptions!

Given that, since Moten is supposed to be a naturalistic conlang (i.e. one that could pass for a natural language), it follows that it should have exceptions to its rules. Yet in all the posts I've written before on the language, I haven't talked about them much, if at all. Is Moten unnaturally regular? Well, as it happens, Moten is a bit like Turkish or Japanese. It does have irregularities, but those are few and far between, which is why we haven't encountered many of them yet. But it does have irregularities, in all areas of its grammar, phonology, morphology and syntax. And now I'm going to focus on describing those irregularities in the remainder of this post.

Exceptions in Phonology

Phonology being the system of rules that describes which sounds are used in the language, and which forms syllables can take, exceptions in phonology should be words that don't follow those rules, i.e. words that are considered good Moten words, but make use of sounds that are not considered part of the phonetic inventory of the language, and/or combinations of sounds that are not allowed according to its phonotactics. Does Moten have those?

Why, yes of course it does! And we've met those words already, in this post, in the section called Interjections. As I mentioned then, while interjections are considered fully fledged words of the language, they sometimes contain sounds or combinations of sounds that are not normally considered acceptable in Moten. Examples of exceptions to phonology are:

  • long vowels: zutuun [zuˈtuːn]: "the sound of complete silence";
  • long or double consonants: ssii [sːiː]: "er, um", ikkee [ikˈkeː]: "ouch";
  • sounds not considered part of the standard Moten phonology: pelg [pɛɫx]: "ugh" (notice how this one also features a normally forbidden two-consonant coda).

By the way, interjections are not necessarily the only words with exceptional phonology. Some recent borrowings also have irregular phonological features, like Doj|slan(t): "Germany", which has an affricate in a disallowed position according to Moten phonotactics. Given that native words still simplify affricates when inflectional, derivational or compounding phenomena would put them in such a position, one can truly speak of an exception.

Exceptions in Morphology

Morphology refers to the forms taken by words, especially how they change due to inflections, derivations or compounding. So exceptions in morphology are simply word forms that inflectional, derivational and compounding rules fail to predict.

Let's start with inflections. This refers to noun inflections (cases and number, definiteness, and the functional prefixes), since Moten has very few verbal conjugations (and the two verbs that can be conjugated are both conjugated in exactly the same way).

Now, if you approached Moten for the first time, and saw these incomplete paradigms:

linan: bird

indefinite nominative accusative genitive
singular linan lindan linva|n
plural linsan linzan linfa|n

bazlo: town, city

indefinite nominative accusative genitive
singular bazlo bazludon bazluvoj
plural bazluso bazlu|zon bazlufoj

ge|sem: father

indefinite nominative accusative genitive
singular ge|sem gezdemun gezvemi
plural ge|sem ge|zemun gesfemi

ku|lu: language

indefinite nominative accusative genitive
singular ku|lu kuldun kulvuj
plural kulsu kulzun kulfuj

you'd be excused if you thought that they were a mess of irregular forms. While they do have a few things in common, those paradigms are all different in non-trivial ways. Yet as I explained in this old post, noun declensions in Moten can actually be described using completely regular rules, based on a series of infixes and suffixes common to all nouns, along with complex but regular morphophonological rules that explain how those affixes interact with the stems they are added to. I've seen this phenomenon (apparently irregular superficial forms underlain by perfectly regular, if complex, formation rules) described by some as "regular irregularity". And don't think that no natural language is like that! I've read descriptions of the mess that is Basque synthetic conjugations that successfully argue that most of it can actually be described by perfectly regular (if complex) morphophonological rules, along with common affixes used by all verbs (there are still some irreducible forms, but they are few and far between, and can often be explained by generic phonological considerations).

So all nouns follow basically the same inflections rules, and all use the same affixes. However, this does not mean that we cannot find true irregularities in noun inflections. Take for instance the noun with the uninflected form tales, which is also its nominative singular indefinite (that noun means "fruit, vegetable"). Given that form, and the regular inflectional rules used on all nouns, one would expect its accusative singular indefinite to be *taldesun. However, that's unattested (read: wrong!), and the attested form is taldeskun! What's happening here is that tales is one of those nouns I mentioned way back in the first Moten post! Such nouns have stems that end with two consonants. Given that that's a big phonotactics no-no in Moten, the second consonant (which I call "fragile") disappears when it would be word-final, and only resurfaces when a suffix is added. When you only know the uninflected form of a noun, it's thus impossible to predict whether that noun has a fragile consonant or not, and even if you're told it has one, it's impossible to predict which one. Since there is no predictive rule available, those nouns can truly be considered exceptional, and the only way to handle them is to learn them as they come (here, as well as in the Moten dictionary, I always mention them with the fragile consonant in parentheses: tales(k). It's a good and visual way to mark both the presence of a fragile consonant and its value).

Another case of inflectional irregularity, although a rarer one, is represented by words whose stem is a single consonant. No, you read that right: in Moten, some words have a stem that is actually a single consonant, without a vowel to accompany it. An example is the stem g, which is used in the verb igi, meaning "to live", as well as a noun meaning "life". In this case, the main problem is that Moten phonotactics don't allow single-consonant words: the shortest possible word must at least have one vowel. When such stems are used as verbs, it's not so much of an issue: the infinitive igi is disyllabic, and the participle guz has a vowel as well (notice that both are perfectly regular, by the way), so they can be declined without a problem. The imperative ("live!") isn't commonly used, but is an issue (since it's normally identical to the stem). It's easily solved, however: such imperatives are formed by adding -i to the single-consonant stem: gi: "live!".

No, the main issue is when such stems are used as nouns. Due to the absence of a stem vowel, all kinds of weird things happen. Luckily, single-consonant stems generally all behave in the same way, so I can simply describe how g: "life" handles it, and you'll know how to handle others.

The first thing to remember is that in single-consonant stems, that single consonant behaves like an onset. This means that infixes will follow it. It's important to know, as otherwise the position of infixes could have been ambiguous.

I'll start with the definite declension of g, as it is easiest to understand. Since definiteness is marked by the infix -e-, it provides a vowel to add to the single consonant and form a pronounceable syllable. So the nominative singular definite is simply ge: "the life". Since the definite infix appears before the other infixes, the nominative plural definite becomes ges: "the lives". The functional prefixes are effectively added to the nominative forms of the noun, so those are easy as well: koge: "by the life", koges: "by the lives". The accusative and genitive cases are relatively simple as well, with a small caveat: despite what I just wrote above, in those cases, the definite infix somehow gets treated as a stem vowel, and the case infixes end up preceding rather than following it! I don't know why it happens there and not with the nominative plural infix: it's just something you'll have to remember! (that's exceptions for you!) So the accusative singular definite of g is gden (rather than the expected *gedun), and its plural is gzen (rather than *ge|zun). The genitive singular definite is gvej, while the plural is kfej.

The indefinite declension of g is slightly more complicated to describe, but not by much. In this case I'll go backwards, compared to what I've done with the definite declension. It's just easier to explain that way. So let's start with the genitive case. Luckily, this case provides us with a vowel thanks to the -i suffix. Remembering that the single consonant g is treated as an onset (infixes follow it), the genitive singular indefinite is thus the regular gvi, while the plural is regularly kfi. The accusative case doesn't give us an obvious vowel, until you remember that its suffix -n becomes -un after a consonant. With that in mind, the accusative singular indefinite of g is simply gdun, and its plural is gzun. The nominative plural works in the same way, you just need to remember that its infix -s- becomes -us- when it would otherwise create an inadmissible cluster. So the nominative plural indefinite is gus (and functional prefixes can be added to it).

And that leaves us with the nominative singular (as well as the uninflected form used when followed by an adjective, which for all intents and purposes behaves like the nominative singular indefinite), which I saved for last on purpose. That's because its shape varies depending on its use and what surrounds it. Let's first get the functional prefixes out of the way: since all functional prefixes end in a vowel, one can simply add g to them and the result in pronounceable. And indeed that's how you do it: kog: "with life". Also, when g is followed by an adjective starting with a vowel (or with a single consonant that can form an onset cluster with g, according to Moten phonotactics), it will stay as is, and will be pronounced as part of the onset of the adjective. So for instance g olem: "the same life" (with olim: "same, similar") will be pronounced as if it was *golem. But what do you do when you actually need a naked nominative singular indefinite? (or an uninflected form followed by an adjective starting with an incompatible consonant or a consonant cluster) Well, you will need a vowel to be able to pronounce it, so you'll just have to add one. And that's where things get really confusing, because, at least in the case of g, you can choose between two vowels! Indeed, the naked nominative singular indefinite of g can be either gu or gi! Basically, gu seems to be the more common alternative (probably because of the use of u as an epenthetic vowel throughout the inflection paradigm of nouns), but gi appears regularly as well (even within the speech of the same speaker), especially when the word is surrounded with high or mid front vowels. So while one will tend to use gu on its own, when saying "another life" that same person will be likely to say gi tel (and the written language follows pronunciation here).

To sum it all up, here's a table with the different inflections of g: "life":

indefinite nominative accusative genitive
singular g, gu or gi gdun gvi
plural gus gzun kfi
definite nominative accusative genitive
singular ge gden gvej
plural ges gzen kfej

Now that inflections are out of the way, let's look at derivation and compounding. Derivation is actually very easy. As you may remember, there's very little productive derivation in Moten, and it's all very regular. So far, I haven't been able to discover a single exception to the morphophonological rules I described in last post. The rules may be somewhat complex (and reminiscent of the inflection rules), especially when it comes to diminutives, but they are completely regular.

Compounding, however, is a different beast entirely. As I mentioned before, the changes undergone by compounded stems are not always predictable, so irregularities abound. They run the whole gamut of possibilities, from unpredictable use of the short compound form (sezbon: "velocity" is formed by using the short compound form of both its components sezgo: "high speed" and bontu: "low speed", while similar compounds usually use the short compound form of their first component only, as in ukodun: "age", from ukol: "old age" and odun: "youth"), to unpredictable voicing or devoicing of clusters (eg+sponda -> egzbonda: "two small animals" features progressive voicing, while ag+fe|su -> akfe|su: "sorry for leaving early" shows regressive devoicing. Progressive devoicing and regressive voicing are also possible), to unexpected elisions (uflebe: "quality" uses the short compound form of ufan: "greatness", but also removes the first consonant of its second component tlebe: "mediocrity"), all the way to portmanteau formations (like slebe: "trollishness", from slim: "inappropriateness" and tlebe). Even addition of sounds is attested (as in oskana|not: "ceremony", from oskan: "event, show" and |not: "source, cornerstone, main"), although it is uncommon.

Exceptions in Syntax

While morphology is about the forms words take, syntax is about how those forms are put together to form understandable utterances. Exceptions in syntax are then unexpected word orders, and/or unexpected forms for a specific function in the sentence. I'll keep this section short, as we haven't had a complete overview of Moten syntax yet (next post, I promise!). Still, I can mention at least two exceptional syntactic behaviours.

As you already know, one of the strictest rules of syntax in Moten is that adjectives follow nouns. It's so strict because it's basically what makes them adjectives in the first place: the adjectival function of nouns is marked by position, and by position only. That said, even such a strict rule has exceptions, and we've met them already. As I mentioned then in the post about numbers, short cardinal numbers used attributively tend to precede rather than follow the noun they complete. So one will usually say su ka|se rather than ka|se su for "one man" (either is correct, but the first one is just far more common). I don't know why this happens, but it does. It's a bona fide syntactic irregularity.

The second irregularity I want to mention concerns another rather strict rule of Moten, the one about the form of subjects of transitive verbs. As I explained way back then, the subject of a transitive verb only appears in the nominative case when it's actively, willingly participating in the action. Experiencers and other non-volitional subjects will appear in the instrumental instead. This means that in general, inanimate objects and abstract concepts will never appear as subjects of transitive verbs in the nominative case (at least in formal registers and ignoring some forms of personification), since those cannot have a will of their own. There is one group of exceptions to this rule though: weather phenomena. Words referring to weather phenomena break the rule completely: not only can they appear in the nominative case when they are the subject of a transitive verb, but they actually have to appear in the nominative case. Appearing in the instrumental in that role is actually a grammatical error! It's as if weather phenomena were always actively and willingly participating in the action described by the verb. In fact, this exception goes so far that even in mythical stories, fables or tales where weather phenomena are personified, they will appear in the nominative case when subject of a transitive verb, even if they should actually be playing an experiencing role! So nouns like ibo: "air, wind", tlap: "light rain", ibipiz: "storm", la|zi: "high weather temperature", devodi: "west wind" and eme: "sun" (when used to refer to the way it warms up the air on a sunny day) will always appear in the nominative case when subject of a verb, regardless its transitivity or their actual role.

There are other exceptions in Moten syntax, but I want to wait until I've properly discussed the rules of syntax first before I spend more time on them. Moreover, most of those exceptions happen in specific, informal speech registers, so I will need to explain those as well before I can go into more details.

Exceptions in Semantics

Put it simply, semantics is the relationship between words and their meanings. But since the relationship between words and their meanings is essentially arbitrary (well, more or less), where are the rules for exceptions to break?

Well, there's one area where semantics do have some kind of rules: derivation and compounding, and those rules predict the meaning of a derived form or a compound based on their components. In English, those rules are of the kind "the suffix -er forms an agent noun from a verb" or "nouns ending in -berry always refer to some kind of fruit".

That said, as you probably already know, natural languages tend to be very creative when it comes to applying those rules, and exceptions are common. For instance, in English, although it is formed regularly, a choker will usually not be understood as someone who chokes someone else, but as a piece of jewellery that only chokes if it's too small! In the same way, regular compounds needn't always have a meaning that regularly derives from the meanings of their constituents (for instance, a greenhouse isn't necessarily green, and is definitely not a house!).

So what about Moten then? Well, it also tends to play fast and loose with the meaning of derived words and compounds. For instance, in spite of the suffix -sif being used to form simple actor nouns from verbs, the regularly formed agzif (from jagi: "to go") doesn't simply mean "goer", but "traveller". Diminutives are especially guilty of this. Look for instance at dloasin, the diminutive of dloa: "pear". It has nothing to do with small pears, and means instead "light bulb".

Compounds are even more all over the place, and their meanings are often only metaphorically linked to the meanings of their components. For instance, look at no|sezgo, which literally means "quick brightness". Its actual meaning is "lightning bolt". Then there's tolmos (literally "four paws/legs"), which means "dining table, meal, fare, food". And I'm not even talking about all those verbal compounds where the second element, the verb istu|l: "to summon, to call", is almost completely bleached of any meaning, and works more like a generic verbalising suffix (for instance, while ipenlastu|l: "to invite" can reasonably be seen as a combination of istu|l: "to call" and ipenlaj: "to wait for, to expect", in joknestu|l: "to read", istu|l is really just a meaningless suffix used to turn okne: "story, tale" into a verb).

And then there are all those nouns formed by nominalising an inflected form, thanks to surdéclinaison. The connection between the original word and the resulting nominalisation is often pretty tenuous. See for instance devodi: "west wind". It's actually the genitive singular definite of dod: "evening, night" and thus literally means "from the night" (the step from "from the night" to "west wind" is definitely not an obvious one!).

What's Next

So, if you had any doubt that Moten is a naturalistic language with its fair share of irregularities and exceptions, this post should have put those to rest! Moten may not be the paragon of exceptionalness that English is, but it has a healthy amount of irregularities.

So, what's next then? Well, next time I'll finally get this syntax post published that I promised nearly a year ago! And as promised as well, that post will contain a full text example, rather than a few words or disconnected sentences. I am not promising anything in terms of when that post will be published, but hopefully you won't have to wait for 10 months again. In any case, until next time!