Monday 15 April 2013

Moten Part XI: Derivation and Compounding


After all that heavy lifting we've done in the last three posts, let's have a break and look at an area of Moten grammar which is relatively straightforward. This post will focus on derivation and compounding, i.e. the twin arts of creating new words on the basis of existing words. We'll first look at derivation, which is not very productive in Moten, and for this reason shouldn't take too long to discuss, before turning to compounding, which is much more commonly used in the language. We'll finish with a discussion about reduplication, which is quite productive in Moten.

Derivation

Derivation is the process of creating new words on the basis of existing ones, usually through the addition of affixes that are not themselves independent words (for instance, from the English adjective "happy", one can form the adjective "unhappy" with the prefix "un-", and the noun "happiness" with the suffix "-ness". One can also combine both to form the noun "unhappiness"). It's different from inflection, which uses affixes to form grammatical variants of the same word (declensions or conjugations), although the ability to nominalise phrases by surdéclinaison blurs the limit between the two. However, since surdéclinaison has been treated already in the previous posts, I will focus here only on word derivation done by methods other than over-inflection.

Unlike English, Moten is very poor in derivational affixes, and we've actually met a few of them already. We've seen for instance the suffix -ano, used to form ordinal numbers. There are also the degree of comparison affixes (described in the same link as above), although those straddle the border between derivation and inflection. And then there are the infinitive circumfix i-...-i and the participle suffix -z, which can be seen as derivational affixes as well (as I described in the past, the infinitive can be used to derive an action noun or an instrument noun from a verbal stem, while the participle forms a noun describing the typical object of a verb).

But there are a few more of these affixes in Moten, and I will now focus on them.

Actor/Agent Suffixes

In English, there is a generic suffix "-er" used to form agent nouns from verbs. For instance, from "to write", one forms "writer", while from "to read" one gets "reader". Agent nouns don't necessarily refer to people (see for instance "printer" from "to print"), nor do they all use the suffix "-er" (see for instance "guard" and "to guard"), but that suffix is nevertheless quite common.

In Moten, there are not one but three different suffixes that more or less correspond to "-er". They are the suffixes -sif, -non and -vu|z, and they all three form nouns from nominal or verbal stems. However, their uses are quite distinct.

The suffix -sif is the most generic of those three. It is mostly added to verbal stems to form actor nouns (not necessarily agent nouns, as will become clear in a minute), i.e. nouns meaning "someone who 'verb's" or "something that 'verb's". Many of those nouns are quite transparent in meaning, like nanagzif: "writer", from |nanagi (stem nanag): "to write", or eze|sif: "listener", from jezeti (stem ezet): "to hear, to listen". Others may not be as transparent in translation, but their meaning still is, like plisif: "lord, lady", from iplisi: "to reign" (basically: "someone who reigns"), or elojmastulsif: "memory", from jelojmastu|l: "to be remembered" (literally: "something that is remembered". Notice how this is not an agent noun). And finally, some have experienced a bit of a semantic drift, and although their derivation is still clear, their meaning cannot be directly inferred from it, like teolsif: "friend", from iteo|l: "to please, to be liked".

The suffix -sif features also a second, totally unrelated meaning: when used with nouns referring to a fruit, vegetable or flower, it forms the name of the plant that bears them. For instance, from the noun dloa: "pear", one can form dloasif: "pear tree". And from the noun sentalu: "rose", you can form sentalusif: "rosebush, rose shrub".

The suffixes -non and -vu|z are somewhat different. Firstly, they can be used with both nominal and verbal stems with the same meaning. And secondly, they form strict agent nouns that must always have an animate referent (usually a person). The difference between the two is in the shade of meaning they give to the agent nouns, and this difference is not easy to explain clearly. Grosso modo, nouns formed with -non refer to artists, while nouns formed with -vu|z refer to workers. For instance, starting again with the verb |nanagi: "to write", while the word nanagzif: "writer" simply means "someone who writes" (without any special connotation), the word nanagnon means "writer" in the common sense of "author, writer of fiction, who does it for a living, or at least tries to", and the word nanagvu|z more likely refers to a "technical writer", i.e. someone who is employed in a company to write documentations and other practical items.

But things are actually more complicated than that, and the correct understanding and use of -non and -vu|z can be a challenge. I know I myself have difficulties understanding the exact difference between the two! Basically, many human activities (although not all of them) can be classified under two umbrella terms: bel and go. I usually translate bel as "art" or "craft", while go is usually translated as "job" or "profession". When you know under which term a specific activity falls, it's easy to know which suffix to use to name the practitioner: when an activity is a bel, the practitioner will be described by a word ending in -non, while when the activity is a go, the practitioner will be described with an agent noun ending in -vu|z. The problem is that knowing under which term a specific activity falls is not straightforward. The translations I gave are approximate, and do not describe bel and go as well as I would like. For instance, while bel includes indeed most activities we consider artistic (like painting, sculpting, writing fiction, etc.), it also includes things we usually describe as crafts instead, as well as activities we usually consider simply as professions (for instance, "teaching" in Moten is a bel, as are many scientific endeavours). In the same way, go refers to many utilitarian activities we think of as professions, but includes also some activities that we as a society don't usually consider as such (for instance, "parenthood" in Moten is a go). And some activities (like "writing" as in the example above) can be considered a bel or a go depending on context, the fine details of the activity definition, and sometimes even the speaker's opinion.

I'll come back to the bel vs. go issue later in this post, so for now just remember how -non and -vu|z map compared to those two terms. One last thing I do want to add though is that for activities that can be treated as a bel or a go, using one or the other is in no way a judgement of value. There is nothing inherently "better" or "worse" in a nanagnon compared to a nanagvu|z. Those are just names for people having two kinds of related, but ultimately different, activities. Using a noun ending in -sif, when a noun ending in -non or -vu|z would be more appropriate, can be interpreted as pejorative though. That's because nouns in -sif can be used for inanimate things and concepts as well as animate beings, and can have a sense of happenstance (a nanagzif is someone who happens to be writing, not necessarily someone who writes as a job). So while correctly using -non and -vu|z can be difficult, one cannot simply replace them both with the more generic -sif, as this could be understood as an insult.

To finish with this section, I need to mention that like any other affixes, -sif, -non and -vu|z cause various morphophonemic changes when added to stems. For -sif, the changes are similar to those caused by the nominative plural infix -s-:

  • -sif becomes -if after s, z, |s or |z.
  • The sequence t + -sif becomes |sif, the sequence d + -sif becomes |zif.
  • The sequence |l + -sif becomes lsif, the sequence |n + -sif becomes nsif.
  • -sif becomes -zif after a phonemically voiced consonant (as a reminder, the nasals, laterals and approximant are phonemically voice-neutral).

For -non, the changes are similar to those caused by the accusative suffix -n:

  • -non becomes -on after n or |n.
  • The sequence j + -non becomes |non.

For -vu|z, the changes are similar to those caused by the genitive singular infix -v-:

  • f and v disappear before -vu|z.
  • The sequences s + -vu|z, z + -vu|z, |s + -vu|z and |z + -vu|z become zvu|z.
  • The sequence |l + -vu|z becomes lvu|z, the sequence |n + -vu|z becomes nvu|z.
  • A phonemically voiceless consonant becomes voiced before -vu|z.

These morphophonemic changes become second nature once you've internalised Moten's phonotactics.

Diminutives

Diminutives are very common in Moten, as in many other languages, and unlike in English. Formation of diminutives is very productive, and nearly any word (well, nominals and verbs in any case) can have at least one derived diminutive. Those are usually used to convey:

  • A diminished degree of the root meaning;
  • A smallness of object size or quality;
  • A single unit or quantity of some material;
  • Intimacy or endearment (especially with kinship terms);
  • Contempt or distaste (basically an ironic or sarcastic use of the previous meaning);
  • A separate but related meaning of the original root (usually a simple diminutive that changed meaning through semantic drift).

As we will see in a future post, diminutives are also commonly used to form nicknames.

In Moten, diminutives are formed using one of three possible suffixes: -sin, -mas and -zes. They are similar in meaning, although their uses vary a bit:

  • -sin is a generic diminutive suffix. It can be used with nearly any nominal or verbal stem to form diminutive nominals, but also diminutive verbs. As for diminutive nominals in -sin, they can refer to animate beings, but also to inanimate objects or abstract concepts.
  • -mas is much more restricted than -sin in meaning. It can also be used with most nominal or verbal stems, but can only form diminutive nouns, which must always have a male animate referent. In other words, they can only be used to refer to a male person or animal. This means, in particular, that nouns with naturally female referents, like di|la: "mother" and e|lon: "woman", cannot normally have diminutives in -mas.
  • -zes is the female equivalent of -mas. In other words, it is used to form diminutive nouns that must always have a female animate referent. For this reason, nouns like ge|sem: "father" or ka|se: "man" with a naturally male referent do not normally have diminutives in -zes.

In terms of morphology, the diminutives suffixes are unique among Moten's affixes: they are not added to the word stems themselves, but to a special version of the stems, usually shorter. That special diminutive stem is formed through a process called clipping, which is very strictly defined in Moten. This process goes as such:

  1. Take the original stem, and keep only its first syllable. Remove every syllable of the stem except its first one.
  2. However, a diminutive stem must be a single closed syllable (i.e. a syllable ending in a consonant). So if the first syllable of the original stem is already closed, it becomes the diminutive stem (for example, the diminutive stem of bazlo: "town" is baz-, and the diminutive stem of sponda: "small animal" is spon-). But if the first syllable of the original stem is open, to form the diminutive stem, one must take it, and add to it the onset consonant of the following syllable in order to close it (so the diminutive stem of linan: "bird" is lin-, while the diminutive stem of kfezi: "grandchild" is kfez-).
  3. However, there are exceptions to the rule that the diminutive stem must be a single closed syllable:
    • If the stem itself is a single open syllable, it stays unchanged (so the diminutive stem of |la is simply |la-).
    • If adding the onset consonant of the following syllable would result in a forbidden coda (remember, while |l, |n, |s and |z are allowed at the absolute end of a word, they are not allowed at an internal coda), the diminutive stem is simply the open first syllable of the word (so the diminutive stem of ge|sem is ge-, and the diminutive stem of di|la is di-).
    • If the first syllable is open but the second one doesn't start with a consonant (i.e. the first two vowels of the word are in hiatus), then the diminutive stem is the first syllable plus the vowel of the second syllable (but only the vowel, even if that second syllable is closed), i.e. the diminutive stem is in this case two syllables, both open (so the diminutive stem of dloa: "pear" is simply dloa-, while the diminutive stem of teol, the root of iteo|l: "to please", is teo-).

Once the diminutive stem is formed, the actual diminutive itself is simply created by adding one of the diminutive suffixes to it. Naturally, such an addition can cause morphophonemic changes, although those are limited since diminutive stems are quite restricted in form:

  • The suffix -sin causes similar changes to the suffix -sif:
    • -sin becomes -in after s or z.
    • The sequence t + -sin becomes |sin, the sequence d + -sin becomes |zin.
    • -sin becomes -zin after a phonemically voiced consonant.
  • The suffix -mas causes hardly any morphophonemic changes:
    • -mas becomes -as after m.
  • The suffix -zes causes similar changes to the suffix -sin:
    • -zes becomes -es after s or z.
    • The sequence t + -zes becomes |ses, the sequence d + -zes becomes |zes.
    • -zes becomes -ses after a phonemically voiceless consonant.

Now that we know how to form diminutives, here are a few actual examples to illustrate their various uses, based on the list of possible meanings I gave earlier:

  • A diminished degree of the root meaning: leksin: "darkish, darkishness" (from leksod: "dark, darkness"), jeksi|n: "to touch lightly, to brush past" (from jeksaj: "to touch, to hit");
  • A smallness of object size or quality: ba|zin: "small dog" (from badi: "dog"), linsin: "small bird" (from linan: "bird"), oksin: "short story, anecdote" (from okne: "story, tale"), bazin: "small town, village" (from bazlo: "town");
  • A single unit or quantity of some material: vonsin: "drop of water" (from vone: "cold water"), a|sin: "flame" (from at: "fire", a fire is seen as a collection of flames);
  • Intimacy or endearment: ba|zin: "doggie", gemas: "dad" (from ge|sem: "father"), dizes: "mum" (from di|la: "mother"), teosin, teomas, teozes: "dear, darling, sweetie" (all three from teolsif: "friend");
  • Contempt or distaste: kamas: "little man, mister" (a disparaging term of address based on ka|se: "man"), ezes: "little woman, missus" (a disparaging term of address based on e|lon: "woman"), dloamas: "fatty man", dloazes: "fatty woman" (insulting terms referring to overweight people based on dloa: "pear");
  • A separate but related meaning of the original root: leksin: "twilight", imsin: "mobile phone" (from imikostu|l: "telephone"), dloasin: "light bulb" (from dloa: "pear", probably due to the shape), kasin: "male" (from ka|se: "man"), esin: "female" (from e|lon: "woman").

As you can see, diminutives have a pretty wide range of uses, and it usually cannot be predicted which diminutive will convey which meaning simply based on the stem and suffix used. However, in general the indication of smallness will be the basic meaning, with the other meanings depending mostly on context (except where semantic drift has acted). The example of ba|zin, which can convey either neutral small size or endearment (or both), is typical here. Still, it takes a lot of experience to understand exactly which diminutive is acceptable in which context.

Zero Derivation

As you may have noticed, Moten words often have more than one meaning. This can range from slight variations in sense (like ipiz, which can mean both "heavy" or "strong") to much broader ones (like jagi, which means "to go" and "to leave", but also "to be worth"). But in any case even with different meanings the words in question don't change category (i.e. a noun stays a noun, and a verb stays a verb).

However, in some cases the various meanings of a stem can be associated with different parts of speech. I am not talking about the infinitive or participle of a verb used as a noun (they are nouns), but about the stem itself being used as a nominal or a verbal stem. Probably the most common example of this is zunla, used as a noun to mean "location, position, place", but also as the verb izunlaj, meaning "to be at, to live at, to be located".

Naturally, to an English speaker, verbing nouns and nouning verbs isn't that unusual. But Moten is much stricter about its grammatical categories, and this pattern is relatively exceptional. But since it exists, I need to mention it. Since derivation can change the part of speech of a word (like verbs suffixed by -sif becoming nouns), I consider this a form of derivation, but one with a zero affix, hence the name zero derivation. Whether that's what actually happens in the language is not actually relevant, just remember that it simply means that a stem is used identically for words of different parts of speech with related meanings.

Despite this pattern being somewhat uncommon, some relatively common words exhibit it. An example is dabolna: "moment, instance, time", which appears also as the verb idabolnaj: "to happen". Another is jelej: "to sleep", whose stem can be used as a noun: elej: "sleep" (it refers here to the abstract concept of sleep, which contrasts with both the participle elejuz, which refers to the state of being asleep, and the infinitive jelej, which refers to sleeping as an activity. The differences in meaning are small though, and it's not uncommon for a Moten speaker to use elej and jelej as synonyms). A last example is the verb imeti: "to greet", whose stem can also be used as a noun: met: "greetings, regards".

Zero derivation isn't very productive, and I haven't been able to come up with a reliable way of predicting which stems will allow zero derivation or not. At best I have noticed that nouns susceptible to zero derivation usually refer to very abstract concepts.

Notice that I've been talking about nouns and verbs only, but zero derivation can involve particles as well. For instance, the nouns |zaj: "beginning, start" and kun: "ending, end" are also used as particles in the formation of ordinal numbers. An extreme example is the root vepe|ne, which can be used as:

  • A noun vepe|ne, meaning "apology";
  • A verb ivepe|nej, meaning "to apologise";
  • An interjection (i.e. particle) vepe|ne, meaning "sorry" or "thank you" depending on the context.

But stems that can be used in all three parts of speech like that are very rare.

Compounding

With the diminutives and the actor/agent suffixes (and the zero derivation phenomenon), I've basically covered the entirety of Moten's productive derivational morphology. That's right: I wasn't kidding when I said Moten is very poor in derivational affixes! So now let's move on to the much more productive subject of compounding.

Compounding is the process of creating new words on the basis of existing ones, by combining or putting them together. Unlike derivation, the elements of a compound are all independent words that can be used by themselves (examples from English are words like "doghouse", a house for a dog, or "starfish", which is neither a fish nor a star).

Languages differ in their abilities to form compounds. French, for instance, prefers set phrases to proper compounds (although it does have some), while German and Dutch are very compound-friendly. English itself is similar to its Germanic cousins, although this can sometimes be hidden by orthography (while things like "city council member" and "cellar door" are written with separate words, they are arguably compounds, and used as such in speech). As for Moten, it is about as compound-friendly as Germanic languages, although its speaker doesn't seem to create "on-the-fly" ad-hoc compounds as readily as English speakers do. Also, in Moten very long compounds are rather rare. Its compounds are usually formed of two elements only, and even compounds that seem to be formed of three elements or more can always be analysed, semantically and morphologically, as recursive compounds, i.e. as compounds of two elements, one of which (or sometimes both of which) is a compound itself. For this reason, for the remainder of this article I will only discuss two-word compounds.

Before I discuss the various types of compounds used in Moten, I first need to mention some things about the morphology of compounds, as compounding in Moten is not always as simple as just attaching words together.

Compounding Morphology

In Moten, compounding is a phenomenon that applies to stems only. In other words, compounding forms new stems by attaching two stems together. In particular, when a verb is involved in compounding, it's its stem that is attached to another stem, not its infinitive or participle form. The same is true of nominals. For those, the nominative singular indefinite form is usually identical to the stem, but that's not always so (a noun like tales(k): "fruit, vegetable" has the nominative singular indefinite tales, but its stem for compounding is actually talesk, including the fragile coda).

Also, while compounding often involves full stems (as in emelog: "summer", from eme: "sun" + log: "season"), it's not uncommon for the first element of a compound (more rarely also the second one) to be shortened. This shortened compound stem is formed by a clipping phenomenon not unlike the one that forms diminutive stems. In fact, it's pretty much identical to it, except that |l, |n, |s and |z are allowed as coda consonants of shortened compound stems, while they are not allowed in diminutive stems (for instance, while the diminutive stem of di|la is di-, its shortened compound stem is di|l-). An example of a compound using a shortened compound stem is ge|samo: "paternal aunt", from ge|s- (shortened form of ge|sem: "father") and amo: "different-sex sibling".

Whether the stems used in a compound are full or shortened, the phonotactics of Moten imply that putting two stems together will often result in morphophonemic changes at the contact point between them. Unfortunately, unlike with derivational and inflectional affixes, those changes are not always predictable. Still, they do tend to follow a short list of principles I can give you:

  • Identical consonants will merge into one. An example is ge|suko: "paternal uncle", from ge|s- and |suko: "same-sex sibling";
  • Identical vowels will merge into one. An example is izunla|leki: "to find, to locate", from zunla: "place, location" and a|lek (stem of ja|leki: "to find out, to discover");
  • Voice-incompatible consonants will change to agree in voicing, although whether that change will be progressive (the first consonant gives its voicing to the next one) or regressive (the first consonant receives its voicing from the next one) cannot always be predicted. An example of progressive agreement is egzbonda: "two small animals", from eg: "two" and sponda: "small animal", while an example of regressive agreement is akfe|su: "sorry for leaving early", from ag (the stem of jagi: "to go") and fe|su: "sorry to bother you" (notice that we have here a compound particle, formed by compounding a verbal stem with another particle);
  • Consonants like |l, |n, |s and |z will often be simplified (respectively to l, n, s and z). An example is dilku|lu: "mother tongue", from di|l- and ku|lu: "language". An example of this change happening although it is phonotactically unnecesary is o|nigzaj: "September", from o|nig: "autumn" and |zaj: "start". The form *o|nig|zaj is allowed, phonotactically speaking, and yet it's simply not used;
  • Fragile coda consonants will often disappear in compounds if they are followed by a consonant, but not without causing some changes first (usually in the voicing of that consonant). A typical example is velbele: "five minutes", from vel(d): "five" and pele: "minute";
  • Typical phonotactic changes will happen when two vowels come in contact, with usually i becoming j, or a j being added to prevent another vowel from coming in contact with u. An example is japujada: "one hundred years", from japu: "100" and ada: "year";
  • A coda t or d will merge with a following s or z (or |s or |z), resulting in |s or |z. Usually, the result will keep the voicing of the coda. An example is seno|ziza: "Friday", from senod: "earth, ground" and siza: "calendar day". Another example is zoba|saj: "March", from zobat: "spring" and |zaj;
  • The consonants l and j will combine into |l, and n and j into |n, whatever their order. An example is to|los: "four parts", from tol: "four" and jos: "part";
  • Other, unpredictable changes can happen, usually in the form of disappearing consonants even when phonotactic constraints do not apply. Examples abound among the counter compounds, like uzavokez: "eight people" (from uzab: "eight" and fokez: "person". Notice how the b of uzab still voices the first consonant of fokez before disappearing) and uzabos: "eight parts" (from uzab and jos);
  • Finally, in some cases, what we are seeing is actually a blend, where bits and pieces of the two stems are added together to form a new word, sometimes seemingly haphazardly. Luckily those are quite rare. An example is the somewhat familiar term slebe, from slim: "ugliness, inappropriateness, bad" and tlebe: "mediocrity, mediocre", which indicates a quality that can be best described as "trollishness", referring to the behaviour of Internet trolls.

Naturally, more than one of those principles may be active at the same time, and more unpredictable changes may happen when one compounds two stems together. That may be one of the reasons why Moten speakers don't create ad-hoc compounds as readily as speakers of Germanic languages do. Nevertheless, a majority of compounds in Moten are quite regularly formed, so those issues aren't as pervasive as one might think after reading this section.

Compound Types

Now that we've seen how compounds are formed in Moten, it's time to look at what kinds of compounds exist in the language.

There are various ways to classify compounds, but the method I will use here could be called "etymological". It classifies compounds as to which of the two stems is primary: the second one (head-last compounds), the first one (head-first compounds) or neither/both (dvandva compounds). Let me explain what I mean by the "primary stem" (which I will usually call the head).

I've used the term "head" before, in the context of nominal phrases and verbal clauses. In nominal phrases, the head is the central nominal the phrase is about. It is semantically and syntactically primary, and any other word or phrase in the nominal phrase completes or qualifies it. It's basically "what the nominal phrase is about". For example, in the phrase gvaj ko |lavospineas: "for my beautiful daughter" (with ko: "daughter" and vospinas: "beauty, beautiful, good-looking"), the head is the element in italics, i.e. "daughter", as the phrase is about a daughter (who happens to be beautiful, and to be mine), not about me nor about beauty. Notice that unlike in many other languages, in Moten the head is often not the nominal that carries the phrase's inflections. So one cannot identify the head of a phrase simply by looking at where the inflections fall.

In clauses, the head is quite simply the verb. It's the only part of a clause that is truly mandatory (although it can sometimes be omitted, especially in speech, that can only be done when context makes it clear which verb is meant), and while noun phrases in the clause are marked for grammatical function, the actual meaning of that function (what I believe linguists usually call the thematic relation) is usually assigned by the verb itself. For instance, take the phrase nanagduzun: "a book" (in the accusative case). This single grammatical function actually covers various roles depending on the verb. For instance, with the verb |nanagi: "to write", the phrase becomes a patient, i.e. it undergoes the action and is modified by it (in the simple sentence Ga nanagduzun |nanagdin ito: "I'm currently writing a book", the object described by the accusative phrase is actually being created by the action described). With the verb joknestu|l: "to read", however, it takes on a different role, which has been called theme, source or simply experiencee, i.e. it is a source of sensory input (in the sentence Ga nanagduzun joknezdu|lun ito: "I'm currently reading a book", the object described by the accusative phrase isn't modified by the action. Rather, it's the subject of the verb which is modified, through the reception of information stored in the object). Because of this vital role, it makes sense to call the verb the head of the clause.

Now, as it happens, the vast majority of Moten compounds can be seen as the simplification of a nominal phrase or verbal clause. In fact, I'd be hard pressed to find a single counterexample to that statement. Basically, take a simple noun phrase with its head and one dependent, remove all inflections, and concatenate the stems together in the order they appeared in the original phrase, and you've got a valid compound! (in the following sections, I will discuss this statement in more details)

Once you know this, defining the head of a compound is easy: it's simply the stem of the word that was the head of the original phrase. If you are still unsure what this means, don't worry: the next sections should clarify everything.

Let's now discuss each compound type in turn.

Head-Last Compounds

The majority of compounds in Moten are head-last, i.e. the head of the compound is its second stem. Examples in English are words like "doghouse" and "smalltalk". Such compounds result from the simplification of various types of phrases:

Noun + dependent noun phrase:
Many compounds can be seen as the simplification of a noun and the genitive phrase completing it. For instance, the noun ge|suko: "paternal uncle" is clearly derived from the phrase gezvemi |suko: "father's brother". Sometimes semantic drift or the metaphorical use of the head can obscure the relationship, but it's still there. For instance, the noun jespoma: "wrist" comes from the phrase juvezi poma, literally "hand's neck". Another example is knamafin: "flower", from knamvaj fin: "grass's top". And of course the genitive phrase may be a result of surdéclinaison, as is the case of bnameno: "shoe, boot", from tebnevami meno: "glove for the foot".
Noun + relative subclause:
Sometimes the dependent stem is verbal instead of nominal. Such compounds can be seen as the simplification of a noun with a relative subclause. Since many functions can be relativised in Moten, this can lead to various subtypes of compounds. The relativised function may be the subject, as in sonsizea: "yesterday, the day before" (an example of a very productive construction, as we'll see below), from iso|n itos sizea: "the day that precedes (today)", with iso|n: "to precede, to go before". It can naturally also be the object. But many other functions can be relativised as well. For instance, location can be relativised, as in penlatina: "waiting room", from ipenlaj itos tina: "room where (one) waits". And in neg|nalam: "easiness, easy (to do)", from |negi itos |nalam: "lightness by which (someone) does (something)", it's the manner that is relativised.
Verb + argument or adverbial phrase:
This is the opposite of the previous case. Here the head is verbal, and the dependent stem is nominal. Such compounds are usually verbs (but see below) formed by simplification of a verb and a noun phrase added to it. That noun phrase is often the verb's object, as in izunla|leki: "to find, to locate", from zunledan ja|leki: "to discover the location (of something or someone)", or joknesej: "to recount, to recite", from okneden isej: "to tell the story (of)" (notice that despite being formed from a transitive verb and its object, both compound verbs are still transitive). But any argument or even adverbial phrase can become the dependent stem of such compounds. An example is imikostu|l: "to call by phone", that probably comes from the phrase momikevoj istu|l: "to call from afar" (literally: "to summon from the remoteness", with istu|l meaning "to summon, to call", and miko meaning "remoteness, remote, far"). As this example shows, metaphor and semantic drift happen here as well. This is also the case of ibivostu|l: "to count, to measure", from bivdon istu|l: "to summon a quantity" (from bivo: "number, quantity").
Other:
All the compounds shown so far come from phrases that show a well-defined grammatical relationship between head and dependent stem. That is however not always the case, and sometimes compounds look like they come from words that just happened to be next to each other, without a specific relationship between them. Probably the most common example of such a compound is the expression akfe|su: "sorry for leaving (early)" (we'll see in a future post how this expression is used). It's a compound of the verb jagi: "to go, to leave" (stem ag) and fe|su, an interjection meaning "sorry to bother you". Since fe|su is an interjection, it cannot, by definition, have a grammatical relationship with a verb. Instead, the interjection akfe|su seems to come from the simplification of a sentence that may have looked like this: et jagvi ito, fe|su: "(I) have to go now, sorry to bother you". In such a case, it may seem illogical to talk of a head stem and a dependent stem. Nevertheless, akfe|su is still treated as a head-last compound, as it has the same part of speech and a similar meaning as its last component.

So far, all the examples I've shown are compounds of the same part of speech as their heads, and (barring metaphor and semantic shift) represent something that is at least similar to those heads. For instance, ge|suko is a noun referring to a family member, like |suko. In the same way, sonsiza refers to a specific siza, i.e. "calendar day". And imikostu|l is a verb referring to a specific way of calling someone, compared to the generic istu|l. Linguists call those types of compounds endocentric.

But those are hardly the only kind of compounds. Take in English the word "barefoot", for example. Not only does it not refer to a kind of foot, but it's not even a noun! Such compounds are called exocentric, and are common in many languages. And that's the case in Moten too. For instance, we have the noun zoba|saj, literally "spring start". It doesn't actually refer to the beginning of spring, but is a name for the month of March, which is mostly in the winter. Another case is the nominal sugi|no: "simplicity, easiness, easy (to understand)". It's a compound of the numberal su: "one" and the verb igi|noj: "to understand", showing that just as in English Moten compounds are not always of the type as their heads.

Notice, by the way, that this last case cannot be explained by zero derivation, as there is no verb *isugi|noj. But that doesn't mean zero derivation never happens with compounds, as shown by the nominal vospinas: "physical beauty" (from vo|sa: "appropriateness" and ipinasi: "to feel, to look"). In this case, there is a verb ivospinasi: "to look good", so the compound is actually endocentric.

In any case, whether a compound is endocentric or exocentric doesn't change anything in terms of grammar. Just remember that the category and meaning of the head do not always allow one to guess the meaning and use of a compound.

Head-First Compounds

Moten is predominantly head-last, which is why when Moten phrases are simplified into compounds those are head-last as well. However, there is one area where Moten is head-first, and it's in noun+adjective phrases. Those can be combined into compounds, which are then themselves head-first.

Head-first compounds in Moten are always of the noun-adjective kind, so there is much less variety among them than among head-last compounds. They are also less common, although they are often used in people's names (as we will see in a future post). Their meanings are often only partially related to the meanings of their constituents (as in ibipiz: "storm, stormy weather", from ibo ipiz: "strong wind"), when they are not downright exocentric (like mosezgo: "cheetah", from mosu sezgo: "quick paw"). Notice that in both cases the head of the compound takes its shorter, compound stem form. This is very common in head-first compounds. Having the second, dependent stem in its shorter form can happen as well, but is somewhat less common. An example is dloalum: "apple", from dloa luma: "fake pear".

There is no way to distinguish a head-first compound from a noun-noun head-last compound, so the existence of both kinds may lead to ambiguities. In practice the danger isn't as common as one might think, as the semantics of their constituents usually make clear which type of compound we are dealing with.

Dvandva Compounds

So far we've only seen compounds with clear head and dependent constituents: head-last compounds formed from head-last phrases, and head-first compounds formed from noun+adjective phrases. But there's another type of construction where two elements can be associated together without one being dependent on the other: coordinative constructions, where the two elements are associated using a conjunction like "and". Such constructions can be combined into compounds as well, headless compounds that are commonly referred to as copulative or dvandva compounds. A typical example in English is the adjective "bittersweet", literally "bitter and sweet". Another example is the verb "to sleepwalk", i.e. "to sleep and walk at the same time".

In Moten, dvandva compounds are very common. In fact, they are nearly as common as head-last compounds, and much more common than head-first ones. Due to their semantics, there can be only two types of dvandva compounds: nominal+nominal or verb+verb. I will look at each type in turn:

Noun + noun:
The simplest kind of noun-noun dvandva compound just combines two elements and refers to the unit formed by coordinating those two. A typical example takes the phrase ka|se opa e|lon: "husband and wife" and forms ka|se|lon: "(married) couple" (interestingly, I'm told this word can refer to same-sex romantically-involved couples, despite the obvious derivation). Another is gomdod: "day, 24-hour period", from gom opa dod: "day and night". Such compounds refer to the pair of concepts as a unit, and as such are just normal nominals that can be used in the singular and the plural: eg ka|selson: "two couples".
Naturally, semantic drift can happen here as well, usually generalising the sense of those compounds. For instance, from |suko opa amo: "brother and sister", one gets |sukamo: "generation" (both in the narrow sense of "family generation" and in the more general sense of "all of the people born and living at about the same time, regarded collectively"). Another example is emekel: "skies", from eme opa kel: "sun and moon" (Notice that emekel is singular, unlike the translation I've given. I've translated it as "skies" to indicate that it's a poetic/literary word, not the generic word for "sky" —which is bo—).
Of course, abstract nominals can form dvandva compounds as well, as in odamla: "brand new", from odun opa amla: "newly made and newly acquired". Similar is the portmanteau slebe, a somewhat familiar word that can be best translated as "trollishness", i.e. the "quality" exhibited by Internet trolls. It's a combination of slim: "inappropriateness, unfit for purpose, bad" and tlebe: "mediocrity, mediocre".
But with abstract nominals, there is also a separate type of dvandva compounds with quite different semantics. Those are formed by combining two opposites, resulting in a nominal that refers to the general concept represented by these opposite words, without any indication of degree. For instance, while sezgo means "quickness, high speed" and bontu means "slowness, low speed", their compound sezbon means "velocity, speed", i.e. it refers to the concept of speed in general rather than to a specific value of speed. Similarly, from odun: "young age" and ukol: "old age", one forms ukodun: "age".
Notice that not all such generic concept nouns need to be formed from combining two opposites. For instance, while one can take fin: "summit, tall, high" and piv: "bottom, short, low" to form finpiv: "height", it's actually more common to use the unrelated be|s for that meaning.
Finally, there is a last subtype of dvandva compound I need to talk about. So far, all the compounds I've shown have two elements with separate referents (i.e. they are the simplification of a phrase using opa). But one can also form dvandva compounds by simplifying a phrase using de, i.e. the Moten equivalent of appositions. In this case, the compound is formed of two elements that have the same referent. In English, examples of such compounds are words like "actor-director", "singer-songwriter" and "maidservant". A Moten example is the noun subumpi: "shophouse", from subo de umpi: "shop as well as house" (no idea why this word even exists in Moten. Maybe it's a common form of housing wherever it is that language is spoken).
Verb + verb:
Verb-verb compounds are very numerous in Moten, and a common way to form new vocabulary out of existing stock. They are at least as common if not more common than noun-verb compounds. There are basically two subtypes of verb-verb compounds: compounds where the two elements correspond to actions happening after each other (usually repeatedly), like jaguba|si: "to come and go, to move back and forth", from jagi: "to go" and juba|si: "to come", and compounds where the two elements correspond to simultaneous actions, like ipenlabutaj: "to sit and wait", from ipenlaj: "to wait" and ibutaj: "to sit".
However, most often those compounds have undergone semantic drift or have always been used metaphorically. For instance, jaguba|si is often used in the sense of "to be nervous" (probably from the metaphorical picture of someone pacing back and forth), while ipenlabutaj is most often used in the sense of "to be patient". Other examples are ipe|lastu|l: "to show, to reveal" (from ipe|laj: "to see, to watch" and istu|l: "to summon, to call"), ipenlastu|l: "to invite" (from ipenlaj: "to wait" and istu|l again) and isestu|l: "to read aloud, to recite" (from isej: "to say" and istu|l once more).

With this I've thoroughly discussed the three compound types that are present in Moten. All compounds you will come across in the language fall under one of the types and subtypes presented above. But that doesn't mean there is nothing left to learn about compounds. So the next two sections will focus on some nifty things Moten does with them.

Pseudo-Suffixes

One of the main differences between derivation and compounding (besides the fact that the former uses affixes while the latter uses independent stems) is that derivation is predictable, while compounding is more free-form. Let me explain what I mean with an English example:

When you add the suffix "-able" to a verb, you know the result will always be an adjective referring to the capacity to accomplish the action described by the verb. This is automatic and always valid. The only uncertain thing is that some verbs do not accept the suffix "-able", but if they do then what I wrote above will always hold.

On the other hand, compounding is much less predictable. Having the same element in the same position in two compounds is no guarantee that the two compounds will be similar in meaning. For instance, despite have the same second element, the compounds "homework" and "firework" have very little in common.

Sometimes, though, compounding patterns can be predictable, i.e. using a certain stem in a certain position will always result in compounds of related meanings. An example in English is the noun "berry" which, when used as the head of a compound, always forms nouns that refer to some kind of fruit ("strawberry", "blueberry", "blackberry", etc.). If I were to suddenly start talking with conviction about a "catberry", people would assume it's some kind of fruit they'd never heard of before (maybe a berry that our feline companions find delicious). In a way, "berry" here nearly behaves like a suffix, used to regularly form names of fruit.

In Moten, there are quite a few stems, both nominal and verbal, that are commonly used in regular compounding patterns. Such stems form compounds with predictable meanings, as if they were derivational affixes. Since they can still be used as independent words, I call them pseudo-affixes. And since the vast majority of those stems are used as the second element of compounds, I usually call them simply pseudo-suffixes.

Compounds formed with nominal pseudo-suffixes are always nominals themselves. Here is a list of the most common nominal pseudo-suffixes used in Moten:

  • tina: "room". The noun tina is used as a pseudo-suffix basically like its English equivalent "room", to form names for types of room. It can be added to nominal stems, as in zubatina: "bathroom, shower-room" (literally: "warm-waterroom", with zuba meaning "warm water", i.e. water typically used for a bath or shower) or oknetina: "reading room" (literally: "storyroom"), but it's most commonly added to verbal stems. Examples are penlatina: "waiting room" (literally: "waitroom"), elejtina: "bedroom" (literally: "sleeproom"), o|zemetina: "dining room" (literally: "eat-and/or-drinkroom", as jo|zemej refers both to eating and drinking) and vajagdina: "classroom" (literally: "studyroom").
  • subo: "shop, restaurant". Like tina, this noun's use as a pseudo-suffix is quite straightforward. It is used to form nouns referring to two types of establishments: shops, and restaurants or other food service facilities. Basically, when subo is added to a stem that refers to an item or group (or category) of items, the result describes an establishment where those items are sold. For instance, with tales(k): "fruit, vegetable", you get talesubo: "greengrocer's". Another example is nanaguzubo: "bookstore" (from nanaguz: "book"). This is a very productive pattern (which can be used with very recent borrowings, like oto: "car", resulting in otosubo: "car dealership"). However, if the item is a type of food or drink, adding subo will form the name of an establishment where those are consumed rather than simply bought, i.e. a restaurant, bar, pub, etc. For instance, with the noun volaj: "poultry, fowl", one can form volajsubo: "poultry restaurant". Notice that the name only refers to the most topical item sold and consumed in that establishment. Other items may still be available for consumption there. For instance, a kafesubo: "café" (from kafe: "coffee") will serve more than only coffee, and a |sesubo: "tea house" (from |se: "tea") will usually serve other things besides tea (the same is true when subo is used to mean "shop"). Also, another pattern is to add subo not to a noun representing a type of food, but to the name of a country or a people. The result is a restaurant that serves food prepared according to the specific cooking tradition of that country or people. For instance, we have filansisubo: "French restaurant" and ita|lasubo: "Italian restaurant". Notice the specific distinction: while both filansisubo and Filanzvi subo translate as "French restaurant", the first one is a restaurant that serves food prepared according to the French cuisine tradition (that may be anywhere in the world and run by people of any nationality), while the second one is an establishment that happens to stand on French soil, and is probably run by French people. If it's a restaurant, it doesn't necessarily serve French cuisine.
  • ku|lu: "language". In English, names of languages are usually simply a nationality adjective used by itself. For instance, France has French people speaking French, Spain has Spanish people speaking Spanish, and Japan has Japanese people speaking Japanese. Of course, things are always more complicated than that, and we have American people speaking English, Brazilian people speaking Portuguese, not to mention Indian people speaking Hindi (which is only the name of the language: there is no adjective "Hindi". There is a related adjective "Hindu", but it refers to a religion, and not all Hindu people speak Hindi, and vice versa). In Moten, the same difficulties appear when one talks about language names, but one thing is constant: Moten language names always use ku|lu as a suffix. It's added to various stems depending on the situation. In simple cases, when one talks about a nation-state with its single official language, the name of the language is simply formed by adding ku|lu to the name of the country. So the language spoken in Filansi is filansiku|leju: "French", while the language spoken in Ita|la is ita|laku|leju: "Italian" (notice that language names in Moten are not capitalised and they take the definite article: they are common rather than proper nouns). But things get complicated quite quickly, as countries may have no associated national language, or more than one of them. For instance, while Doj|slan: "Germany" has doj|slanku|leju: "German" as official language, so does Estalaj: "Austria". One could talk about *estalajku|leju, but that would sound like someone in English was referring to "the Austrian language" (it may work in some contexts, but usually one would simply say doj|slanku|leju). And what about Balat: "India", with its many constitutionally recognised languages? (of which both Hindi and English are considered official at country-level) Which of its languages, if even any of them, should be called *balatku|leju: "the Indian language"? (as it happens, none) So while it's common for language names to be formed by adding ku|lu to the name of a country, it's by no means the only way to create them. The pseudo-suffix ku|lu is always involved, but it can be added to various things: names of regions, islands, etc. (e.g. from Kolsika: "Corsica", one gets kolsikaku|leju: "Corsican"), names of people (e.g. ejuskalku|leju: "Basque", from Ejuskal: "Basque people"), and more rarely various other nominals. What about Moten's name for itself? As it happens, it's motenku|leju. Does it mean that "Moten" is actually the name of a country, region or people? That is a question I unfortunately have no answer to. It's one of those things C.G.'s amnesia makes impossible to know. So in the absence of better information, I will keep using the word "Moten" for the name of the language in English, as an abbreviation of its correct, unwieldy name.
  • fokez: "person". In English, nationality names for people are formed in various ways. For instance, a man from France is "a Frenchman", a person from Spain is "a Spaniard", and a person from Italy is "an Italian". And women often have different names from men. In Moten, things are much simpler: all those names are formed by using the pseudo-suffix fokez, which can be used for both men and women. Otherwise, fokez is used much like the pseudo-suffix ku|lu, and the same caveats apply. So fokez can be used with country names (e.g. filansifokez: "French person"), geographical names (e.g. kolsikafokez: "Corsican person"), names of people (e.g. ejuskalfokez: "Basque person") and sometimes other nominals.
  • go: "job, profession" and bel: "art, craft". In the section about agent/actor suffixes, I mentioned I would come back to the classification of human activities under bel and go. That time has now come. Basically, as I wrote before, many human activities (though by no means all of them) can be classified as a bel or a go. The exact distinction between the two is difficult to pinpoint, and the translations I've given are approximate at best. Typically, bel activities are more artistic, while go activities are more utilitarian. But most crafts are bel, despite often being utilitarian as well. Scientific endeavours are usually bel as well, while the closely related engineering jobs are typically go. Teaching is also always a bel, while learning is neither a bel nor a go. And some activities can be treated as a bel or a go depending on their goal or product, context, the fine details of the activity description, and sometimes even simply the speaker's opinion. Knowing under which term an activity falls is important not only from a theoretical point of view, but also because the name of an activity will usually end with either the pseudo-suffix bel or the pseudo-suffix go depending on its type. Conversely, this means that hearing the name of an activity will usually also make immediately clear under which type that activity falls. Practically speaking, activity names are almost always formed by adding one of bel or go to a nominal or verbal stem that best represents that activity. For instance, we have already seen the agent nouns nanagnon: "author" and nanagvu|z: "technical writer". The activities those two practice are respectively nanagbel: "fiction writing" and nanago: "technical writing". In the same way, from ifotoj: "to look for, to research, to study", one forms fotobel: "(scientific) research". An example that uses a nominal stem is spondakitpel: "zoology", from spondakit: "fauna, animal kingdom" (itself a dvandva compound of sponda: "small animal" and kit: big animal).
  • dati: "way, manner, concept" and ipi: "tool, machine, object". In Moten, there is no single word that covers the same semantic field as the English word "thing". Instead, that semantic field is covered by two words: dati: "abstract thing, concept" and ipi: "concrete thing, object". These two words cover more ground as well, with dati also meaning "way, manner (of doing something)", while ipi also means "tool, machine". In those senses, they are also used as pseudo-suffixes, usually added to verbal stems. Way back then, I explained how, besides being the citation form of a verb, the infinitive could also be used to refer to: either the concept or action described by the verb as an abstract noun, or a person or object that somehow embodies or facilitates the action described by the verb as a concrete noun. An example of the former use is |lezuj: "singing" (from |lezuj: "to sing"), while an example of the latter one is imikostu|l: "telephone" (from imikostu|l: "to call by phone"). But instead of simply using the infinitive, which can be ambiguous in some contexts, one can use the pseudo-suffixes dati and ipi to form nouns with much the same meanings, but much less prone to ambiguity. Using dati forms abstract action nouns, similar to the first meaning of the infinitive I described above. For instance, instead of |lezuj, one can say lezudati: "singing, manner of singing". Using ipi forms names of objects that embody or facilitate the action described by the verb. Those are typically tools or machines. They are similar to the second meaning of the infinitive I described, except that they can only refer to objects, while the infinitive can refer to people as well. For instance, instead of calling a telephone imikostu|l, one can call such an object mikostulipi instead. The difference is that you could use imikostu|l to refer to a person who is on the phone all day long, while you would never use mikostulipi in this sense.
  • va: "colour, hue, tint". I am planning on discussing colour terms and the use of va as a pseudo-suffix in a future post (as there is a lot to say about this topic), so I will only make a short comment here. Basically, Moten has only two basic colour terms: leksod: "dark" and no|se: "bright" (I will describe their full semantic ranges in that future post). But va can be used as a pseudo-suffix to form ad-hoc colour terms when one needs to be more specific. This system, which relies a lot on analogy and context, is too complicated to describe here, so I'll just stop for now. As I wrote, I will devote a future post on this very subject, so stay tuned!

Verbal pseudo-suffixes are normally always added to other verbs, and give them shades of meaning that can often be considered aspectual. Here's a list of the most common, and most productive, verbal pseudo-suffixes in Moten. You'll notice that a few of them are compounds themselves:

  • i|zajstu|l: "to start" and i|za|negi: "to start (with the intention of completing the action)". Moten has two different verbs corresponding to the English "to start, to begin". The first one, i|zajstu|l, is the closest in meaning to its English equivalent, while the second one, i|za|negi, adds to it the idea that the action is started with the explicit intent to reach its natural conclusion (in particular, it implies that the action has a natural conclusion that one can reach). When used as pseudo-suffixes, both form the inchoative aspect, while still keeping the same difference in meaning. In other words, they form new verbs that mean "to begin to...". For instance, with jelej: "to sleep", one can form jelej|zajstu|l and jelej|za|negi, which both mean "to fall asleep". The difference is that jelej|za|negi means "to fall asleep with the intention of waking up fully rested", while jelej|zajstu|l doesn't imply any such natural conclusion. In particular, due to not focussing on eventually waking up, jelej|zajstu|l can also be used as a euphemism for "to die".
  • ikunegi: "to finish". When used as a pseudo-suffix, this verb forms what one can call the cessative-completive aspect, i.e. it forms verbs that show that one has reached the natural conclusion of an action. For instance, see izu|legunegi: "to finish clearing up, to finish getting clean" (often used in the causative voice to mean: "to finish cleaning"), from izu|lebi: "to become clean" (notice the slightly irregular disappearance of the final b of the stem, which still causes voicing of the initial k of the pseudo-suffix). Another example is igunegi: "to die (of old age), to pass away", literally "to finish living", from igi: "to live" (a rare case of a single-consonant stem g).
  • izenki: "to stop" and ikunstu|l: "to give up". Strictly speaking, both verbs means "to stop". But the first one implies a momentary pause only, while the second one implies abandoning whatever one was doing, hence the translation given here. When used as pseudo-suffixes, both verbs form the cessative aspect, i.e. "to stop...". The second one does keep its specific shade of meaning, and can best be translated as "to give up...". For instance, we have jelejzenki which means "to stop sleeping, to wake up", while with ipsenaj: "to swallow, to ingest, to smoke", we can form ipsenakunstu|l: "to quit smoking".
  • iso|n: "to precede". As a pseudo-suffix, this verb indicates preparing oneself to do something, i.e. it can be translated as "to get ready to". For instance, from jagi: "to go, to leave", one forms jagzo|n: "to get ready to leave".
  • izeki: "to follow". As a pseudo-suffix, this verb marks a continuous action. Depending on the context, it can be strong and mean something like "to carry on, to continue", or relatively weak and equivalent to an English progressive form. When it refers to an action that we know was previously stopped, it can be translated as "to resume". For example, with |nanagi: "to write", you can form |nanagzeki: "to keep writing, to resume writing".
  • jagi: "to go" and juba|si: "to come". When used as pseudo-suffixes, these verbs have nothing to do with motion. Instead, they take on a somewhat aspectual meaning, not totally unlike the prospective and perfect aspects (respectively). As a pseudo-suffix, juba|si indicates that an activity has been going on for a while already, i.e. it started sometime in the past, and it is still going on right now. It is similar in meaning to the English perfect continuous ("to have been doing something"), although it even more strongly focusses on the continuity of the activity. The main difference between this form and the Moten perfect aspect is that the perfect aspect focusses on a state resulting from a prior activity (which may have happened at any time in the past and needn't be still happening right now), while this form focusses on the activity itself, which is still currently happening. Here's a short example: imadan nanagubva|si ito: "I've been writing for three years" (literally: "(I) have been writing during three years". Notice that the verb is in the middle voice, which by de-emphasising the object emphasises the action itself. Such a use of the middle voice will be clarified in a future post). As for jagi as a pseudo-suffix, it looks towards the future rather than the past. It indicates that an activity has started, just now or sometime in the past, and that this activity is expected to carry on for some time in the future. It is not unlike an English future continuous ("will be doing something"), except that the activity must have started by the time one utters a statement using this form (it is not unlike saying: "from now on"). The difference between this form and the prospective aspect is that the prospective aspect describes a current situation based on its expected consequences, and the current situation and its expected consequences needn't be the same activity. An example is: motenku|ledun sizevaj ivajagagdin ito: "from now on, I'll study Moten every day" (literally: "(I) will be studying Moten each day", with ivajagi: "to learn, to study". Notice the use of the imperfective aspect, due to the action being repeated. In this case, using the perfective aspect ivajagagi ito would have been correct as well, as the pseudo-suffix jagi already encodes the ongoing, repeated aspect of the action. The complex interactions between verbal forms and pseudo-suffixes will be discussed in a future post).
  • ja|zi|n and joplej: "to give, to take, to receive, etc.". I will not linger much on those two for now. Their use as pseudo-suffixes is strongly tied to the polite speech register, and I will give them a full analysis in a future post about language registers in Moten.
  • ifotoj: "to look for" and ja|leki: "to find". These two verbs are special in that they are not as commonly used as pseudo-suffixes as the ones mentioned above. Basically, when their object is a completive subclause or an action nominal phrase, these verbs take on a different meaning, respectively "to try (to do something)" and "to succeed (in doing something)". Here are simple examples of this usage: ga oknuden |nanagi itos ifodo|n ito: "I'm trying to write a story" (literally: "I'm looking for that (I) write a story") and ludozvu|n ipe|leda|n ja|leki etok: "I managed to meet him" (literally: "(I) found the seeing of him"). In that sense, the subject of the completive subclause or action nominal is usually the same as the subject of the main verb (although this is not mandatory in Moten), and one could think that those sentences could be shortened by turning the main verbs of those sentences into pseudo-suffixes. And indeed, the sentences above can be rendered that way: ga oknuden |nanagvodo|n ito and ludosun ipe|la|leki etok. However, despite those sentences being grammatical and lighter than their variants, C.G. always feels a bit uncomfortable using them. He still prefers the versions with completive subclauses or action nominals, however heavy they may become. I'm not sure what this means for the grammaticality of these constructions in general. Since C.G.'s idiolect is the only known example of native Moten we have, my only possible conclusion is that at least for C.G., using ifotoj and ja|leki as pseudo-suffixes is still on the edge of grammaticality. But there's no way of knowing whether this is just some idiosyncrasy of his, or a general feature of the language.

What about pseudo-prefixes? As I mentioned above, those are very rare in Moten. In fact, I can only think of two stems used productively as the first elements of compounds: the stems of the verbs iso|n and izeki. They are added as prefixes to nouns to mean respectively "last, previous" and "next, following". And while they are productive, they are quite restricted in their use: they can only be added to simple, original Moten stems indicating time periods.

So they can be added to nouns like siza: "calendar day", dod: "evening, night", daj: "hour" and ada: "year". For instance, one can form the expressions sonsizea: "the previous day, yesterday" and zeksizea: "the following day, tomorrow". With mune: "month", you have sonmune: "the previous month, last month" and zekmune: "the following month, next month". You can even add them to funa: "second", although the results are rather idiomatic: zekfunea literally means "the next second", but it's mostly used as an interjection to ask for someone to wait a short moment, so zekfunea! is basically equivalent to "one moment!". In the same way, sonfunea means literally "the last second", but it's used mostly adverbially to mean "a moment ago" or "just now".

On the other hand, these stems cannot be added to compounds like gomdod: "day", over-inflected forms like negesizdan: "week", or recent compounds like ada|zaj: "January". For those, the normal way to render the meanings "next" and "last" is to use the origin-less ordinal numbers kun egano: "second one after some unspecified origin, next" and |zaj egano: "second one before some unspecified origin, previous". So for instance one has negesizdan |zaj eganeo: "the previous week, last week" or zoba|saj kun eganeo: "the following March, next March".

Reduplication

Reduplication is a phenomenon where a word or part of a word is repeated, for various purposes, from expressiveness to inflection. It is often misunderstood by speakers of modern Indo-European languages, like English, because in those languages reduplication is relatively uncommon, and limited to iconic purposes or baby-talk. But in many languages of the world reduplication is a common feature, and there is nothing childish about it. And as it happens, Moten is such a language.

In Moten, reduplication functions in a way that falls under derivation or compounding, and is never used for inflectional purposes, which is why I am discussing it here. It is also not as common as in some languages, but it is more common than in English, and happens in all registers of language. There are two main kinds of reduplication: full reduplication and partial reduplication, and Moten features both.

Full reduplication is the repetition of an entire word. In Moten, there are two patterns of full reduplication that are used productively. The first one I will go over quickly, as it is very easy to understand and use. It is very similar to the "word word" pattern of English, both in shape and usage. This reduplication pattern is used only with nominals, and consists in repeating a nominal as an adjective to itself. For instance, with badi: "dog", you can form badi badi: "dog dog", and with |suko: "same-sex sibling", you can form |suko |suko: "sibling sibling". The meaning of such forms is the same as in English: they refer to the prototypical meaning of the repeated word, i.e. they indicate that the repeated noun isn't used figuratively, and that its referent is the "real thing", rather than a representative, ersatz or replacement. For instance, the expression badi badi can mean "an actual dog, as opposed to those clothed chihuahuas that hardly behave like dogs anymore". As for |suko |suko, it typically means "an actual sibling or cousin, i.e. a family member, rather than just a very good friend" (since |suko can be used figuratively to mean "very good friend"). This form of reduplication isn't unlike using isis: "truth, reality" as an adjective: "true, real". However, the reduplicated form is often preferred as it is more expressive and more strictly refers to the prototypical meaning of the nominal (whereas |suko |suko strictly refers to a member of one's family, |suko isis can also mean "true friend", i.e. it can still be used with a figurative meaning). Also, unlike in English it isn't reserved to colloquial, spoken language, but can be used also in polite or even formal registers. Notice also that since the reduplication pattern is considered a noun + adjective pattern, inflectional affixes are only added to the second nominal, not to the first (of course, if other adjectives or determiners are added to the noun phrase, they take the inflections with them). Here's an example:

I|zevu|z |laj|zeveju|z gebvezi ige!: I want to talk to the manager, and no one else! (literally: "(I) want to talk to the manager manager!", with i|zevu|z: "manager, director". Notice how only the second instance of i|zevu|z takes the benefactive prefix and the definite infix)

The second pattern of full reduplication operates at stem level, and is usable by both nominal and verbal stems. Unlike the previous pattern which operated at the syntactic level, this one operates at the morphological level and consists in compounding a stem with itself. That's to say, take a stem, repeat it, and merge the two together to form a compound stem. For instance, with ku|lu: "language" one forms ku|luku|lu, while miko: "remoteness" forms mikomiko. Those are true compounds, so morphophonemic rules can change the actual shape of the reduplicated compound. For instance, the reduplicated form of fokez: "person" is fokezvokez, while apa: "star" forms apapa. Also, reduplication is done at stem level, so for verbal stems the infinitive circumfix is added only after reduplication. So the reduplicated form of juba|si: "to come, to arrive" is juba|suba|si. As to the meaning of the reduplicated stems, it can be described as being related to a general idea of completeness, whose specific application depends on whether the stem is nominal or verbal.

For nominal stems, the prototypical meaning of the reduplicated compound of X is "all of the X". The reduplicated compound refers to the set of all possible referents of the original stem, and is normally treated as a mass noun, i.e. it appears only in the singular. For instance, the meaning of fokezvokez is basically "all of the people", although a more accurate translation taking into account its use as a mass noun would be "humankind, humanity". Naturally, like any other compound, reduplicated forms can undergo semantic drift, and as a result move away from their strict meaning "all of the X". For instance, ku|luku|lu, rather than meaning "all languages", refers to "language" as the "capacity to communicate with words", i.e. our conceptual ability to communicate with tongues. In the same way, apapa, rather than simply meaning "all stars", is usually better translated as "night sky". As for mikomiko, it's used in Moten as a nominal equivalent to the English phrase "here be dragons", i.e. it refers to unknown, unexplored and potentially dangerous territories (both literally and figuratively).

With verbs, this pattern of reduplication forms new verbs whose meaning are related to the meaning of the original verb, but with an added shade of absolute completion of the action, or doing the action to its fullest degree. What this exactly means depends on the verb. So for instance the meaning of juba|suba|si is "to settle, to fix one's residence" ("to settle" is seen as "to arrive" taken to its fullest possible degree). Another example is ipolpolti, from ipolti: "to open". It means "to break open, to open in a way that cannot be closed again" (once again, "to open", taken to its fullest degree). Naturally, transitive verbs can be reduplicated as well, like ifotoj: "to look for, to research". Its reduplicated form ifotofotoj means "to fully research, to try to get to the bottom of".

Once again, those forms are not limited to the colloquial language. They are used in all language registers, including the most formal ones.

Partial reduplication in Moten is more limited in that only one such pattern exists, and it can only be used with verbal stems. On the other hand, it's much more productive than the patterns of full reduplication I've described so far, and very commonly employed.

In terms of morphology, partial reduplication is clearly a form a compounding. It consists in taking a verbal stem, and prefixing to it its own shortened compound stem. The result is a new verbal stem. For instance, the stem of juba|si is uba|s, and its shortened compound stem is ub-. So the result of partial reduplication is jububa|si. For a short verb like jagi: "to go", the shortened compound stem is identical to the stem itself: ag-, so the result of partial reduplication is jagagi (identical to the result of full reduplication, but context generally disambiguates). Naturally, morphophonemic changes keep happening, so with ipe|laj: "to see, to watch" one forms ipelpe|laj, from the shortened compound stem pe|l-.

In terms of semantics, partially reduplicated verbs have various meanings, usually of a derivational kind compared to the original verbs, and usually referring to action cycles. Partial reduplication can indicate:

  • Repetition of an action: |nanagi: "to write" -> |nananagi: "to rewrite"; ifi|zo|n: "to ask" -> ifisfi|zo|n: "to ask again"; ipolti: "to open" -> ipolpolti: "to reopen" (notice that this one is identical to ipolpolti meaning "to break open". Context will generally disambiguate);
  • Return to a previous state: juba|si: "to come" -> jububa|si: "to come back"; jagi: "to go" -> jagagi: "to go back"; joplej: "to give, to take, to put" -> jopoplej: "to give back, to take back, to put back";
  • Reciprocation: istu|l: "to call" -> istulstu|l: "to call back"; ipe|laj: "to see, to look at, to watch" -> ipelpe|laj: "to look back at"; iteo|l: "to please, to like" -> iteoteo|l: "to like back".

Naturally, in some cases semantic drift has set in and the meaning of the reduplicated verb is not directly derived from the meaning of the original verb anymore. For instance, while imonuj means "to turn", its reduplicated form imonmonuj has taken on the meanings "to roll" and "to wrap" (originally "to turn again and again").

At the risk of repeating myself, I will mention that these partially reduplicated forms are used in all registers of language.

For the sake of completeness, I will mention a last pattern of reduplication, although it doesn't really belong to this chapter. A very long time ago, I presented the affixes used to form the degrees of comparison. And at the end of that section, I mentioned how the intensifier affixes pen-: "very, too, much, many" and len-: "little, few, too little, too few" could be added to forms already featuring comparison or intensifying affixes. One of the examples I gave then was penpenodun: "very very young, much too young", featuring a repeated pen- prefix. Although it looks like reduplication, one could argue that this is only incidental, and that this penpen- pattern is part of a bigger pattern using pen- and len-. And indeed, one can have lenpenodun: "a little too young" and penlenodun: "really not young enough, hardly young at all". However, things break down when you look at the last possibility: lenlenodun. This form is actually synonymous to penlenodun, and more common than that alternative! This means that what we have here is an actual case of reduplication, used to strengthen the meanings of the prefixes pen- and len-. As for the penlen- and lenpen- forms, C.G. tells me that he thinks those are innovations of his own. He hardly ever uses them, and they "feel" newer to him. The reduplicated forms penpen- and lenlen- are the ones he uses generally, and they feel more like an integral part of the language to him.

What's Next

Okay, it seems that I am apologising at the end of every Moten post, but once again I had no idea this would be such a long article. Luckily, with this post the discussion about Moten morphology is well and truly done (well, except for a discussion of proper nouns, which will have to wait a little, and shouldn't be too difficult anyway). Everything that I still have to talk about belongs to the realm of syntax and semantics, as well as matters of prosody that I will eventually get to.

So far, everything I described was at the level of the word, the phrase, the clause or the sentence. But people speak in utterances and conversations, i.e. groups of sentences and sentence fragments. For my next post, I want to describe the generic syntactic rules that govern Moten utterances, and what they mean for the shape of connected speech and written texts. For the first time in this series, I will use an example that is longer than a single line of text, so please look forward to it!

2 comments:

  1. Hi there,
    I just came across this post (4 years late!) and wanted to say that it's a really excellent piece. I'm studying morphology and appreciated the clear discussion on compounding and also the unique examples from Moten.

    ReplyDelete
    Replies
    1. Thanks a lot! It was a long article to write, and I am glad that you enjoyed it. I am never quite sure how I come across in my writing, so I am always happy to get feedback.

      Delete