Answers to Those Pesky Midterm Review Problems

The following are answers to the midterm review problems that were handed out at the review sessions this past Friday and Monday. If you didn't get a copy, you can download them by clicking on these words (but not these).

Okie doke, before I run through the solutions, I have to make an admission. Here goes: This webpage uses unicode characters. I'm sorry to have to say it, but I feel I should tell the truth. If you scroll through this page and you see a bunch of question marks where you think you should be seeing IPA symbols, then you probably don't have a unicode-compatible font the computer you're using. In order to obtain one, just go to, or go to Dr. Berlin's Foreign Font Archive, and scroll down to the link that reads "Unicode Multilingual". If none of these options work for you, then e-mail me, and I'll decode the strange characters for you.

All right, now that that's over with, onto the problems.

All of these languages are not "real" languages, in that they're not naturally occurring languages. They do function as languages, though, and if you want more info on them, you go here. The point of these exercises is to give your mind something to gnaw on leading up to the midterm. All the strategies represented herein will be useful to you on the midterm (as with all the homeworks you've had so far in class, which can be downloaded here), though these problems may, in fact, be harder than anything you'll encounter on the midterm (may).

Okay, down to business. To follow will be what you might have discovered about the data set related to Zhyler passives:

(1) The first thing you might notice about this data set is that neither the infinitive nor the passive is the stem. You can figure this out by seeing that not everything that appears in the infinitive appears in the passive, or vice versa. In other words, for the first form, /ʃomal/ in the infinitive, if you wanted to say that the infinitive was the stem, you'd have to say that the final /-al/ was stripped off, and that then a passive suffix was added. If you wanted to say that the passive was the stem, you'd have to say that the final /-nos/ was stripped off, and that the infinitve suffix was added. If you say, however, that /ʃom/ is the stem, then you need not strip anything off: All you have to do is add a suffix where appropriate.

Okay, now you've got an idea about what the stem is. So what're the suffixes? This is a little tougher. Looking at the infinitve, you should see plenty of /-al/'s and /-el/'s (there are a couple /-an/'s, but we'll deal with those later). So you might say that the suffix is something like the vowel you've seen in the Turkish homeworks plus /l/. But then there's form 13: /mijɯl/ (don't be alarmed by that upside-down "m": It's the IPA equivalent of what we've been referring to as /ɨ/ in the Turkish homeworks). Don't give up on the intuition, though. Take a look at the passive form associated with /mijɯ/. You'll notice it looks different from the passive suffix used in 1 through 10. This should tip you off that something else is going on.

There was a generalization that we made in class awhile back about suffixes. Namely, it seems to be felicitous that a suffix beginning with a consonant should come after a stem ending with a vowel, and that a suffix beginning with a vowel should come after a stem ending in a consonant. (This was Swedish Nouns 1). Remembering this, look again at forms 11 through 16. One thing you might notice is that if you analyze all the stems as ending with a vowel, then that explains why you get the /ɯ/ in /mijɯl/: The infinitive suffix is /-el/ or /-al/ after a consonant, but just /-l/ after a vowel. Further, if you look at 11 through 16, they all have the same passive suffix. So you might further generalize that the passive suffix is /-s/ after a vowel-final stem, and then either /-nes/, /-nos/, /-nis/ or /-nus/ after a consonant-final stem.

(3) Okay, now we're getting somewhere. We've figured out the general shape of the stem and the suffixes for the infinitive and passive. Now we need to try to explain some of the phonological alternations we're seeing. First I'll deal with the easiest. The infinitive ends in an /l/, except for three forms: /ðalan/, /valan/ and /lan/. Notice anything funny about these forms? You might notice that the infinitive usually ends in /l/, and, oddly enough, the only place where it doesn't, is where the preceding consonant is also an /l/. This is a process known as dissimilation: Where one consonant changes when it comes next to a like consonant. This is rather common with liquids (which is why we have words inherited from Latin like "molar" and "moral". Both are, in fact, the same adjectival suffix: The Romans just didn't like two liquids to come next to each other).
(4) Finally, we have to deal with the vowel harmony issue. One little bit of vowel harmony should be very easy to figure out, since you've seen it with the dative in Turkish. Basically, you have five allomorphs of the infinitival morpheme: /-l/, /-al/, /-el/, /-n/, /-an/, and /-en/. We already know why you get an /n/ every so often, and we know why you just get a consonant, so all that needs explaining is the vowels. Like I said, this should be easy. The vowel is an underspecified vowel. It's a [-high], [-round] vowel that is unspecified for its [back] feature. It gains its [back] feature from the previous vowel, so that /ʃom/ plus the infinitive is /ʃomal/, because /o/ is [+back], and thus the suffix should be [+back], as well.
(5) Now comes the Million Dollar Birthday Fries question. If you get it write, you get a special order of Million Dollar Birthday Fries, delivered especially by Moe Szyslak*! The question: How can you predict the vocalic alternation in the passive suffix? (Another question you might have had was, "Is the vocalic alternation in the passive suffix predictable?" The answer is "yes".) Well, look at what each one comes after, and write out the features of the vowels (both in the suffix and in the stem). Especially, look at numbers 4, 5, 6 and 8. What you might notice is that the suffix vowel and the stem vowel will have the features [high] and [round] in common. Its feature for [back] is also predictable, but only based on the inventory of vowels in the suffix. So, you'll notice that the only suffix vowels you get are /e/, /o/, /u/ and /i/. If you divide these by [high], you'll see that there are two [+high] vowels and two [-high] vowels. The rest of the features, though, are difficult, because there are two [-round] vowels and two [+round] vowels, and then two [-back] vowels and two [+back] vowels. So how does the distribution work? Basically, you use this formula: If a stem vowel is [αback], [βround], [γhigh], then the suffix vowel will be [βback], [βround], [γhigh] vowel. This is expressed in alpha notation, where the Greek characters are just variables standing in for either plus or minus. What this is saying is that the suffix vowel will have the same value of the stem vowel for [high] and [round], but that the [back] feature will have the same value as the [round] value of the suffix vowel (or stem vowel). So, if you have a [-round], [-high] vowel, that leaves you with either /a/ or /e/. The vowel you'll get, though, will always be /e/, because /e/ is [-back], and the value for [round] was [-round].

So, that's the solution. Not too tricky, but just tricky enough. This was probably the second-most difficult problem of the set. Now let's move onto the most difficult one: The second problem.

*This is in no way a guarantee of any kind, as Moe Szyslak does not, in fact, actually exist.

This is a good problem--"put hair on your chest", as my greatgrandfather would say. Anyway, I actually made this problem slightly simpler, so you don't have to worry about the phonological variation between the trill and the flap, and you also don't have to worry about stress-placement. Nevertheless, there's still a lot in here, so let's dive right in:

(1) Okie doke. Where to start? Let's start with something easy. Look at forms 7, 12 and 16. Each of these forms ends in a voiceless fricative (/mɔs/, /ʔɛʃ/, and /pɛθ/, respectively) and when the plural suffix is added, these fricatives voice (/moziks/, /ʔeʒiks/ and /peðiks/, respectively). This doesn't happen with stops (see form 3, for example), or with any other segment, so you can just state, as a phonological rule, that a voiceless fricative will voice intervocalically.
(2) Moving on, let's look at the form of the plural suffix itself. It tends to have a /ks/ in it somewhere. Sometimes it has a vowel before it; sometimes it doesn't. Sometimes it just shows up as /s/; sometimes it doesn't show up at all. Each of these things has to be figured out. Let's start out with the /s/ form. The forms where you just get /s/ are forms 3, 8, 10 and 20. What do all these have in common? Well, it's not syllables, it's not vowels, it's not semantics, it's not the consonant that begins the word... Aha! They all end in /k/. And really, this should look familiar. In Zhyler, you had an /l/ changing to an [n] after another /l/. Similarly, in Gweydr, you have a /k/ disappearing after another /k/: Another case of dissimilation. So now that's figured out. You might assume that this is a general phonological rule for the whole language, in which case you'd state that /k/ deletes after another /k/. It should be noted, though, that there's no evidence for this being a general phonological rule, since it only occurs in one environment in this data.
(3) Next, let's look at forms 15 and 19. These exhibit no change in the plural. You could predict this, though, because the singular forms just happen to end in the same phonological sequence of the generic plural morpheme. Adding another /ks/ to these forms would be quite a mouthful, so, basically, there's no overt realization of plurality for forms that end in /ks/. This is just like the past tense forms of some English words, such as "hit", "bid" and "quit" (the normal "quit", not the weird "quit", as in "he quitted his post").
(4) All right, we've figured out the consonantal variation, so let's work on the vowel. One thing should be obvious. Specifically, if a word ends in a vowel, then the suffix will simply be /-ks/ (see forms 4, 6 and 14). That's well and good. There are other forms, though, that also have merely the /-ks/ suffix. These forms are 2, 9, 13, 17 and 18. These forms end in /j/, /w/, /l/, /n/ and /r/, respectively. What can you say about all these? They should look familiar: They're all sonorants. More than that, though, they're non-syllabic sonorants. Look at forms 1 and 5. These end in sonorants, but they're syllabic sonorants, and so they can't get the /-ks/ suffix. Why should this be? There are reasons, but they're beyond the scope of this problem, so don't worry about them. The distinction is what's important, because if you didn't pick up on it, you'd predict incorrect plurals for 1 and 5. (Note: There's also another phonological process happening with form 17. This is what's referred to as nasal assimilation. Basically, a nasal will take on the place feature of the consonant that follows it. So, the alveolar /n/ became a velar /n/, [ŋ], because it occurred right before a velar stop. This is a very common process, and happens in even an exotic language like English.)
(5) Having figured out where you get the /-ks/ suffix, you should, by process of elimination, now know where you get the suffix with the vowel. But what is that vowel? Well, the two choices are either /-iks/ or /-ɪks/ (a subtle distinction, but it shouldn't be beyond an English speaker, since we make this sublte distinction in words like "peaks" and "picks"). Something you might notice is that these vowels have one and only one distinction between them--namely, that [i] is a tense vowel (also referred to as a [+ATR] vowel), and that [ɪ] is a lax vowel (also referred to as a [-ATR] vowel). Keeping that in mind, look at forms 1, 5 and 11, and compare them to forms 3, 8, 12, 16 and 20. Notice anything about the vowels in the plural forms of these words? If not, look at them again, and ask yourself: Are these vowels tense vowels or lax vowels? What you should find is that you get /-iks/ when all the vowels in the word are tense, and that you get /-ɪks/ when all the vowels in the word are lax. In this way, the vowel is predictable. In fact, just like you did for Turkish and Zhyler, you can make an underspecified vowel for the plural suffix. This vowel will be [+high], [-round], [-back], and [αATR], and the specification for [ATR] will come from the previous vowel. This is what's referred to as ATR Harmony, and it's a common feature of a good many African languages, such as Leggbò.

It's time again for the Million Dollar Birthday Fries question, and that question is this: Can you predict when a vowel will be [-ATR] and [+ATR]? A tough question, but (at least partly) doable. To help answer this question, you might look at forms like 7, 10, 12, 16 and 20. In each of these forms, the vowels start out as [-ATR] (or lax), and become [+ATR] (or tense). Any ideas? If not, consider what changed. Take /mɔs/, for example. It starts out as CVC. When you add a suffix, it becomes CVCVCC. Before you ask, no, it has nothing to do with the number of syllables. Look, for example, at /gri/: It's monosyllabic both in the singular and plural. So what's the difference? Ideally, the difference should be the opposite of the difference between the singular and plural of /mɔs/, because the opposite thing happens to its vowels. Well, no more beating around the bush. The difference is this: The first syllable of /gri/ is an open syllable (i.e., it doesn't end in a consonant), whereas the first syllable of /mɔs/ is a closed syllable (i.e., it does end in a consonant). When you pluralize these forms, the status changes: /gri/ now ends in a consonant (two of them, in fact), so the vowel becomes [-ATR], or lax; /mɔs/ has a vowel (plus some other stuff) added onto it, so the /s/ (now a /z/) no longer ends the first syllable, but rather begins the second syllable--thus, the first syllable is now an open syllable, and the lax vowel becomes tense, or [+ATR]. And that's the mystery. Now all you need is a rule that states that the [ATR] feature of the first vowel of a word will spread to all the other vowels in the word, and you're set.

However, that's not the whole story. If you look at the tense/lax correspondences I give you at the bottom of the sheet, you'll notice that /ø/ is a tense vowel, and /œ/ its lax counterpart. Now look at forms 2 and 9. They end in a consonant, but the vowel is /ø/, not /œ/. In fact, even in the plural, where more consonants are added, there's no vowel change. What's the reason? Basically, the vowel is followed by a coda glide, and this coda glide forces the vowel to be [+ATR]. Further, if there were vowels that followed, they would all be [+ATR], as well. Tricky, huh?

All right, that's it: The most difficult problem of the four. If you got it all, 100%, then you should feel justified in referring to yourself as hot stuff, because this one warn't easy.

This problem may look intimidating, because it has clicks and tones, but I assure you, it's the easiest of the four. In fact, I made it even easier than it could have been, because I took out all the nasalized vowels (they would've been a mess to represent). All right, here's the solution to Njaama Verbs:

(1) First, you should recognize what you've got to do. Basically, there are four verbal prefixes whose forms you have to decipher. It should be fairly obvious that the forms in the first column, the habitual forms, can serve as stems (assuming otherwise would complicate the problem needlessly, and would cause problems elsewhere in the language). Now, looking at the examples for the four prefixes, you should notice that they vary in form depending on whether the stem begins with a consonant or a vowel. After noticing this, you could also notice that there are basically two different types of prefixes: (1) Those of the form CV- before consonants and CVC- before vowels; and (2) those of the form CV- before consonants and C- before vowels. Of the prefixes listed, the transitive and passive are type 2 prefixes, and the perfect and reflexive are type 1 prefixes.
(2) With that figured out, let's look at the type 2 prefixes. The forms (ignoring tone) are either /k-/ or /kɛ-/ for the transitive, and /w-/ or /wa-/ for the passive. We've already figured out the distribution: The V-final prefixes occurring with C-initial stems, and the C-final prefixes occurring with V-initial stems. So, we're done with type 2 prefixes.
(3) Type 1 prefixes are slightly different. You get /ni-/ and /nij-/ for the perfect, and /su-/ and /suw-/ for the reflexive. Nevertheless, you already know the distribution: The V-final prefixes occurring with C-initial stems, and the C-final prefixes occurring with V-initial stems. So that much is done already.
(4) Having figured that out, there are some phonological bits left over. Specifically, you see /w/ alternating with /v/, and, though you only have one example, /j/ alternating with /ʒ/. Nevertheless, the answer should be fairly obvious, because it has to do with our old friend dissimilation (it pops up everywhere, don't it?). Basically, a glide will go to a nearby voiced fricative (or another way of saying it would be that the glide goes from [+vocalic] to [-vocalic]) when it occurs before a vowel that shares all its features save the feature [+consonantal]. What this means is that /j/ and /i/ are identical, featurewise, save that /j/ is [+consonantal] and /i/ is [-consonantal]. Same goes for /w/ and /u/. Since these segments are too similar, Njaama causes them to dissimilate in order to make the sequences more pronounceable. This kind of thing happens in lots of languages (though not English).
(5) As a final note, there's the issue of tone. This is kind of like a Million Dollar Birthday Fries question, but, really, it's so simple that it's not good enough to be considered a Million Dollar Birthday Fries question. Maybe a Ten Cent Birthday Fries question. Basically, the answer is this: If a prefix has a vowel in it, that vowels tone will be the same as the tone of the following vowel (either high or low). When a prefix is added, it's toneless, and so it has to get its tone from the next syllable, and so the value of the tone, high or low, spreads from the stem to the prefix.

That does it for Njaama Verbs. Not tough at all. Now onto a problem which might be slighly trickier.

This problem isn't too tough, phonologically, but you're still going to have to think about things, I guarantee you that. So, without further ado, here's how to do Kamakawi Plurals and Diminutives:

(1) The first thing you need to do is decide what the stem's going to be. For each form, you'll have a basic form and a derived form, in either the singular or the plural. But no matter what kind of morphology you get with the singular and plural, the diminutive is a suffix added directly to the stem, so that might be a good way to start. In fact, there's a trick to it: The first letter of the diminutive will be the first letter of the stem, while the last letter of either the singular or plural will be the last letter of the stem. Using that method, you should be able to figure out what every stem is. Now we can move on.
(2) There are three little phonological bits to figure out, so I'll do those first. Take a look at form 4. Using the method outlined above, you should see that the stem is /fale/. If you've got that figured out, then you should be able to figure out that the prefix for the singular for this word is /apeo-/. When you add it, the /f/ becomes a /v/. There's your first rule. It's probably going to be just that specific, because there are no other voiceless fricatives in this data. Well, almost. Look at form 1. In form 1, the stem is /hopoko/. When you add the plural prefix /u-/, though, the /h/ changes. Rather than becoming a voiced /h/ (and there is such a thing: The IPA symbol is [ɦ]), it becomes a glottal stop, [ʔ]. So there's another rule. Finally, take a look at form 11. When you add the plural prefix /u-/, a glottal stop appears between the /u/ of the prefix and the /u/ of the stem, as if by magic! Now, you could imagine a rule that would insert glottal stops everywhere where two vowels come next to each other, but vowels come next to each other all over the place in this data. The only place where you get this glottal stop insertion is between two identical vowels (though, I know, there's only one example). So that's another rule you could write. All of these are phonological rules targeting very specific things, simply because, as you might guess, there aren't a lot of consonants in Kamakawi, so there aren't a lot of consonant classes. Anyway, though, remember this environment of two like vowels, because it'll become important again in just a second.
(3) Ignoring the plurals and singulars for a minute, let's take care of the diminutive real quick. Basically, there are two forms of the suffix: /-i/ and /-ki/. If you look at their distribution, you'll notice that /-i/ comes after stems ending in /a/, /e/, /o/ and /u/, and /-ki/ comes after stems ending in /-i/. And there you have the like-vowel environment again. This is another strategy to keep like-vowels from sitting next to each other (and one reason for not wanting them to sit next to each other could be [yes, you guessed it] dissimilation), but rather than inserting a glottal stop, this particular suffix has two forms, and specifies the particular consonant it wants, which, in this case, is /k/. Now, you don't want to specify this alternation as a phonological rule, though, because it occurs only with this suffix. (P.S.: That's an important point: Knowing when to posit a phonological rule vs. a morphological rule.)

All right, now comes the Million Dollar Birthday Fries question. I didn't think it would be that tough, but I've been told that it was, so I'm making it a Million Dollar Birthday Fries question, since I have the power to do so. Before we get to it, though, let's look at the facts. There are two types of words here. One type of word takes the prefix /apeo-/ in the singular, and the plural is simply the stem. Another type of word has the stem for its singular, and takes the prefix /u-/ in the plural. These environments are mutually exclusive (i.e., you'll never find a word that takes /apeo-/ in the singular and /u-/ in the plural, or that simply has the stem for both singular and plural), so you have to find a way of deciding whether a word is going to be an /apeo-/ word or an /u-/ word. How to do it? Well, you could just say that /apeo-/ words are of class A, and /u-/ words are of class B and be done with it. But there is another way! So, the Million Dollar Birthday Fries question is: Can you predict whether a word will be an /apeo-/ word or an /u-/ word? The answer, of course, is yes. But how?! Here's how. Look at the definitions of the words. In particular, look at 13 and 14. That's the same word you've got there in both 13 and 14, just with different morphology, which produces a slightly different meaning. So that's one clue. Another clue is the stem. Notice that for /apeo-/ words, the stem gives you a plural meaning, whereas with /u-/ words, the stem gives you a singular meaning. Ponder those facts for a minute, and also ponder the difference between these two sentences: (a) "I picked up a rock"; and (b) "That house is made of rock".

All right, here comes the answer. A word will be an /apeo-/ word if it's a mass noun, and it will be an /u-/ word if it's a count noun. What are mass and count nouns? We have them in English. So, you can say, "I have grass on my front line", but not "*I have a grass on my front lawn". You can, however, say, "I have a kind of grass on my front lawn". That's one way to turn a mass noun into a count noun, which is, basically, a noun that can be counted, and which can take the indefinite article. English has an interesting system of derivation, in fact, using the indefinite article. So, you can say, "I swim in water", but not "*I swim in a water". But, consider the following: "I drink water" vs. "I drink a water". The second sentence means, "I drank a glass of water". By using the indefinite article, we turn the mass noun, "water", into a count noun, "glass of water". The same kind of thing is happening in Kamakawi. If you want to say, "My house is made of stone", you use the /apeo-/ form of the noun. If you want to say, "I picked up a rock", though, you use the /u-/ form. For a word like /fale/, though, the meaning is always mass, so you always have to use the /apeo-/ form. Nevertheless, that doesn't preclude the coinage of /fale/ as an /u-/ noun some time in the future. It might mean something like, "a type of grass", like it does in English.

Well, that does it for Kamakawi Plurals and Diminutives, and for the midterm review problems. I hope they were helpful. In case you haven't figured it out by now, I created these languages. I do so for fun. Seems like the natural hobby for a linguist, but they don't all go in for it. Oh well. Anyway, if you have questions, please e-mail me, and I'll get back to you as soon as I can.

University of California San Diego
Linguistics Department #0108
9500 Gilman Drive
La Jolla, CA 92093-0108

Phone: (858) 535-1312
Fax: (858) 534-4789


Campus Web Page:
Personal Web Page:

Back to the Linguistics Home Page

Linguistics Department, UCSD