Njaama Tone Sandhi
YEEEEEEEEEEEEEEEEEEHAAAAAAAAAAAAA! Door's open, boys! Anything goes now!
By that I, of course, mean that Njaama no longer is a perfect (i.e., unrealistic) tone language. I wanted to make it so that contour tones were disallowed, as were tone melodies of greater than two tones, so I just changed every offending tone, but alas! That can be no more. For in this great wide world of ours there's a thing called upstep, and poor Njaama has fallen victim to the beast. Yes, if Esperanto were a tone language, it might've acted like the old Njaama, but (thankfully?) it is not. Now, in this world with a brand new Njaama, tone has become slightly more complex (just slightly). Here's how it goes:
Autosegmental Phonology
What frightening words those are... Anyway, with the advent of autosegmental phonology, a whole new world was opened up with respect to tone, vowel harmony, gemination, Arabic, and many other linguistic phenomena. With respect to tone, what autosegmental phonology does is it allows us to talk about tones as something separate from the actual phonological form of a word (i.e., the letters). Before, tones were thought of as features of vowels, and this caused all sorts of problems, 'cause, man, tones are their own thing, and they float along of their own free will, not listenin' to what no consonant or no vowel does or don't do, dig?
Tonal Word Types
So here's what the upshot is. Take a word like mula, "to release". Both its vowels carry a low tone. There are two ways of saying this: (a) Each vowel has a feature "low tone"; or (b) the whole word has a low melody, and the result of this is that each vowel has a low tone associated with it. According to things like the OCP and the Twin Sisters Convention, it's better to go with hypothesis (b), but, unfortunately, Njaama can't show you why... A shame, that. Moving on...
Njaama words are a conglomeration of two things: Accents and pitch (for the purposes of this language, "pitch" will be defined as "high tone"; "no pitch" will be defined as "low tone"). Some have both, some have just an accent, some have just a pitch, and some have neither. So, for example, a word with all low vowels, like mula, has neither. As a result, a default tone is inserted, and this default tone is low, so all the vowels in the word have a low tone. This type of a word (a word with all low tones) is called a Type 1 word.
So, we've already seen what happens to words with neither an accent nor a pitch. Let's see what happens to the others. First, let's take pálá, "tower". Pálá has a pitch, but no accent (i.e., the word is inherently high in pitch, but there is no accent, or stress). So a pitch word with no accent will have all high tone vowels. This type of word is called a Type 2 word.
Moving on, let's look at a word without pitch, but with an accent: mbotú, "guy". Mbotú has an accent, or stress, on the second syllable. In order to differentiate it from the other syllable, the tone is high, since the stressed syllable has to stand out. Since Njaama allows tonal melodies that are maximally HL or LH, though, what would happen with a trisyllabic word whose accent was in the middle? Here's an example: kayáásyá, "sky". Here the accent is on the second syllable. If it were on the third, the word would be *kayaasyá. Since it's on the second, the high tone spreads to the next vowel, or, in other words, prevents any tones that follow it to be low. This is because Njaama doesn't allow a tonal melody of LHL. Whenever such a melody results, it's "fixed" (more details on that later). This type of word, that has low tones all the way up until the accented syllable, and then has nothing but high tones after it, is called a Type 3 word.
Finally, we'll look at the fourth type of word: A word with a pitch and an accent, such as háme, "one". Since the word starts out with high tones, the only way to distinguish the accent is to give the accent a low tone, and so that's what happens. Think of the accent, then, in this case, as something that causes the following tone to differ from the previous, though it does this in different ways. At any rate, in this case, the accent is on the second syllable. So what about a trisyllabic word? Here's an example: tímula, "to scatter". Here the accent can be said to be on the second syllable, and so that's where the pitch drop is. (This particular word is actually bimorphemic, however, so it's a little different, but for our purposes, it works.) Anyway, so this is a Type 4 word.
That's how the tone system works for stems.
Affixes
When affixes are added to a word, different things happen, since Njaama allows only four tonal patterns: H, L, HL, and LH. Affixes are of two types: Pitched and pitchless (by definition, affixes in Njaama can't have an accent). I'll now show you how each type of word deals with having affixes added to it.
Adding Affixes to Type 1 Words
Going back to our Type 1 word, mula, let's first see what happens when you add a pitchless affix. Our pitchless affix will be the passive prefix, wa-. Three things happen: (1) First, a low tone is inserted by default into the world mula, giving the whole word low vowels; (2) the pitchless prefix wa- is attached; and then (3) since all vowels must have a pitch, the nearest pitch spreads, that pitch being the low tone on the u vowel in mula. The result is wamula, "to be released", which, in turn, is a Type 1 word. If we wanted to show this process graphically, we could do it like this:
If you wanted to add a pitched affix to a Type1 word, like, say, tí-, the iteratrive prefix, then what you would in fact be doing is creating a Type 4 word. Since the prefix is not pitchless, it requires no tone spread, and since it produces an allowable tonal melody, HL, there is no problem to fix. Thus, you get the word tímula, "to scatter".
Adding Affixes to Type 2 Words
Type 2 words are a bit strange, in that, no matter what kind of an affix you try to attach to them, the result is the same. Here's what I mean by that.
Take our word for "tower", pálá, and add the dual suffix -tá to it: Nothing happens. (Well, actually, the Twin Sisters Convention causes the added H tone to delete, leaving the suffix toneless, and then the tone which caused the deletion spreads, and, that tone being a H tone, the suffix that formerly had a H tone now has...a H tone.) The result is the word pálátá, "the two towers" (do you hear that sound? That dial-up modem sound? Somewhere on Earth a geek just got its wings...!).
So what happens if you add a pitchless affix? Well, I'm afraid you don't get a Type 3 word, no no. Using our word for "tower", let's add the augmentative suffix, -ngo. When we add dispela suffix, the same thing happens as happened above with wamula, only with the tones reversed. Since the suffix pitchless, it needs a tone, and the nearest tone is a high tone, so the result is pálángó, "big ol' tower".
Consequently, the simple rule is: If you add an affix to a Type 2 word, you get another Type 2 word.
Adding Affixes to Type 3 Words
Moving onto the slightly more complex types, let's take our Type 3 word, mbotú. Taking a step back, we can see that this word has a LH tone melody. What this means is that it (the tonal melody) must change the tone of a pitchless suffix, or a pitched prefix. So if we added the diminutive prefix kí-, the L tone from the first syllable, mbo, delinks the kí-'s H tone and makes it a low tone. (What happens to that H tone? For an almost full explanation, go on to Adding Affixes to Type 4 Words.) Thus, the result is kimbotú, "little guy" ("hobbit", maybe? Ooh, look! Another wingèd geek...!)
For completion's sake, we'll add the human plural suffix -lá to mbotú and see what happens. Surprisingly, the suffix retains its H tone (via the magic process discussed above). If the suffix had originally been pitchless, it would've gained a H tone, but since it wasn't it, all it did was gain a H tone, and the result is mbotúlá, "guys".
That said, this is basically how Type 3 words are handled. Now we'll proceed to Type 4 words.
Adding Affixes to Type 4 Words
Finally, we'll take a look at Type 4 words. Our Type 4 word of the day is háme, "one" (the numeral). Given that, it doesn't really matter what the pitch quality of the affix is so much as what happens when a suffix or prefix is added. I'll give an example of each.
If we add the abstract prefix tyá- to our word, then the prefix gets to keep its high tone, as illustrated above, and the result is tyá'áme, "unity".
If we add the plural suffix -kí to our word, though, something funny happens. We know what the result has to be: hámeki, "ones (as in more than one number one—the character, for example)". What happens to the H tone, though? Well, the short, incomplete answer is that it gets delinked by the preceding low tone. What happens then? Maybe this picture will help to explain:
Look ye upon that green-circled H! It doesn't affect the tone of the word in question, but I have a feeling that it may still have a role to play in this story... (I do almighty Gandalf's work this day.)
That, in a nutshell, is what happens when you add affixes to, not only Type 4 words, but to all types of words in this beautiful language we call Hausa—I mean, Njaama. If you dare to venture further, I'll explain the dreaded phenomenon known as upstep.
Upstep
Downstep is not uncommon in the world's tone languages. Upstep is. Nevertheless, it does exist. There is, for example, the language Masa, whose tonal system looks so similar to mine that I think they stole it! Grrrrr...! Hostility aside, though, the definition of upstep is, shall we say, the phenomenon which occurs when two L tones which are associated with two separate tone bearing units (e.g., vowels) are separated, on the tonal tier, by a H tone which is associated with nothing (i.e., it's floating). The phenomenon that results is that the second L tone will be higher then the first L tone. It won't be a high tone itself, though. In fact, if any more high tones were added to the word, the rest of those tones would be of the same pitch (i.e., slightly higher than a normal tone). Sound strange? It isn't. Not in the least. Uncommon, but not strange. If you think about it. (I hope.)
So let's see how this actually works in Njaama. First off, let's grab one of our words from above that had a floating H tone: hámeki. We've seen already that this word has a floating H tone at the end of it. So, let's see what would happen if we stuck a word after it that begins with a low tone:
The particular arrangement of words depicted above could be part of a sentence, but wouldn't really constitute an utterance. The most probable way to construe the words above would be that "the ones" is the direct object (without its necessary prefix), and that sa is the third person subject marker on the verb. But you get the point.
Oh, a point of clarification. See that little thing next to the a in sa above? Looks like a small i? What it actually is is an upside down exclamation point. Why is it there? To indicate that upstep has taken place. And why an upside down exclamation point? Because the convention most commonly used to indicate downstep is to write a right-side up exclamation point. So, there you have it.
Attack of the Floating Tones!
One thing that happened historically is that Njaama allowed two vowels to come next to each other. One result of this was long vowels. Another result was the formation of the glide codas w and y (these derived from o and u, and e and i, respectively). One thing that happened when the codas were formed, though, is that what previously was a tone bearing unit (i.e., a vowel), now could no longer bear tones, and so the tone previously associated with the coda vowel was lost: Left to float on for all eternity... Sad.
That lost little tone, however, can resurface in the form of upstep, though! So be happy! YAAAAY!!!
Given the vague description above, the astute reader might wonder, "Hey... Why is there no downstep?" Good question. Logically, if you were able to have floating high tones, why not floating low tones? The answer is simple enough. Since Njaama has a kind of off/on relationship with tone, the floating low tone (the default tone) wasn't considered a tone, but rather the absence of one. For that reason, the floating low tones formerly associated it with coda vowels were reanalyzed out of existence! So, while you do get Type 1 words with floating H's at the end, you never get Type 2 words with floating L's at the end.
Lost? If so, I haven't done my job (which I'm well paid to do, let me assure you). Here's an example derivation that shows how the floating tones arose using the question word kaj:
As a result, that floating H tone does one of three things: (1) Disappears, due to the OCP (two identical tones next to each other collapse into one tone on the tonal tier); (2) causes upstep; or (3) attaches (provided there's an empty tone bearing unit to attach to).
As of yet, I haven't shown an example of possibility three. I'll do that now.
Among the many things the phonological form sa does is it's used for emphasis with any given noun phrase. It itself is pitchless, but before it gets a default L, it gets a tone from the end of the noun phrase it modifies. So, if you want to emphasize pálá, "tower", you get pálá sá "the tower" (or something. Makes more sense in a sentence). If you want to emphasize ndómba, "zebra", on the other hand, you get ndómba sa.
The question word kay, by itself, can serve to mean "what" and "who", and is, itself, a noun phrase. Thus, it can also be followed by the emphatic particle sa. Since it has a low tone, one would predict kay sa, but because of that floating H tone, it's actually kay sá. Here's what that looks like graphically:
And that should just about do it for the whole tone issue, save for one last lingering little lidbit (that's Haitian for "tidbit"):
An Orthographic Issue
It remains for me to tell you how I resolved the orthographic/romanizational issue related to floating tones and upstep. First, the natural upstep that occurs when high tone affixes are added to a word which can't take a high tone affix will not be rendered orthographically, or in the romanization. The reason for this is that it's predictable, provided you know whether an affix that's been added was originally pitched or pitchless.
The other manner of floating tone, though—the one that's lexical—that will be handled both by the orthography and the romanization. Since the only time you get a floating high tone on a lexical item is following a low tone, this last low tone will be marked specially: With a grave accent. So, for example, the question word will from here on out be written kày.
For the romanization, each lexical entry with a floating high tone at the end will be marked with the rising tone marker, and will end with the switch-tone marker. So, orthographically, kày will look like this:
That, then, is all there is to know about tone in Njaama. For now.