Morphology

This section covers the material in Pinker's chapter 5, as well as File 5 in Language Files.

a) Morphology is the study of the structure of words.

Paradoxically, however, the concept of word itself defies simple definition. In English, for example, words tend to be smaller than the sentence, and we combine words to form sentences. One tricky thing, however, is that in many languages, a single word can have "sentence" meaning. Here's an example from Spanish: hazmelo. This "word" is actually a command that is best translated as "do it for me" (do (haz) it (lo) for me (me)). In Swahili, the word atakusumbua means "s/he will annoy you." Nevertheless, if we take English as an example, we have a clear sense that sentences can be broken down into smaller units (words), each of which generally contributes to the meaning of the whole.

For example: Pigs like mud is a sentence containing three words (pigs, like, mud).

If we focus on the word pigs for a second, we notice something interesting. Pigs itself can actually be decomposed into two parts that are easy to grasp intuitively. Specifically, pigs consists of pig + s. So, we have a noun "pig", which lets us know we are in the presence of, well, "pigness" and an additional element "-s" which attaches to the noun and lets us know that we are dealing with not one pig, but rather, with more than a single pig.

By contrast, the word "mud" can't be broken down in the same way. Of course, we can break it down into three "sounds" [m + u + d]. (Actually the [u] is a vowel called schwa but we'll ignore this detail for now.) But these three sounds don't each contribute a meaning to the whole word. In pigs, however, the two parts [pig] and [s] DO each contribute a meaning. The former tells us about the type of thing that we are referring to in the world, and the latter contributes information about number.

Each of these two pieces [pig] and [-s] is called a morpheme, and, for our purposes here, we can define morphemes as the smallest meaningful unit within the word.

Thus, a word like [mud] consists of a single morpheme, while [pigs] has two morphemes. The word [unbelievable] has three morphemes, while the word unidirectionality has four!

• Morphology is the study of word structure and word formation.
• Words, though impossible to define in absolute terms, can be thought of as the units that are combined to form sentences in a language such as English.
• Just as sentences can be broken down into smaller units (words), we can break words down into smaller, meaningful parts.
• The smallest meaningful part of a word is called a morpheme.
• Note: not all words have more than one morpheme. Words that have only one morpheme are also called monomorphemic words (e.g. pig). Words with more than one morpheme are called polymorphemic words, as in foolishness (fool + ish + ness).

c) Glad you asked!

Let's consider another example: governments.

First, we can agree (I hope) that governments is comprised of three morphemes:

• govern
• -ment
• -s

How are these three parts different from one another? Note that govern is the kind of morpheme that can also stand on its own, as in the sentence, "Some people think it is a good idea to govern with an iron fist". Morphemes that can stand on their own are called free morphemes.

The other two morphemes, [-ment] and [-s], can't stand on their own. Rather, they have to be attached, i.e. they have to make up part of a larger word. Morphemes that cannot stand on their own are called bound morphemes.

• Free morphemes can stand alone, while bound morphemes cannot.

There's even more exciting news on the morpheme front. Let's focus more closely on the two bound morphemes [-ment] and [-s] in the word [govern-ment-s]. Intuitively, we can sense that there's a difference between them. While [-ment] is added to the verb "govern" to give us a new word, the noun "government", [-s] simply makes the noun plural. It doesn't change the part of speech and create a brand new word in the same way that adding [-ment] to [govern] did. The difference is that [-ment] is a derivational suffix, while plural [-s] is an inflectional suffix.

Affixes come in two flavors: derivational affixes and inflectional affixes. Here we'll review the TESTS we can use to determine whether an affix belongs to one category or the other.

• it changes the part of speech of the word that it is added to. So, the [-ment] suffix changes verbs to nouns, as in [govern + ment]. The [-able] suffix changes transitive verbs like "read" and "write" into adjectives, as in [read + able], etc. This is a very reliable test. If an affix changes the part of speech, it IS derivational.

• it is not required by the syntax. This test can be confusing at first. What we mean is that we don't need to add, say, the derivational prefix [un-] to the adjective [kind] in order to agree with the noun that [unkind] modifies. Contrast this, for example, with the [-s] ending that we put on the end of the verb in the third person singular, as in "John walk-s a lot." That [-s] is merely there to agree with the subject of the sentence, "John". That is, it's there because the syntax requires it. So, [un-] is derivational, but [-s] is inflectional. Again, this is a reliable test.

• it is not very productive. Productivity is another confusing point. A rule of thumb is that derivational affixes are less productive than inflectional ones. But what does productive mean? Let's look at a concrete example. With the exception of a very reduced set of nouns (child, ox, mouse, deer and a handful of others), the plural suffix [-s] can be freely added to any old noun. So, grab a noun, stick [-s] on the end, and voila, you've got yourself a well-formed plural noun. This is an example of productivity. The plural is productive because it can be added basically across the board to any noun. By contrast, think about the [-hood] suffix in English. It also adds to nouns, in this case to make a new noun, as in [sister-hood]. But can we add it freely across the board? I don't think so. Try adding it to "desk". Do you think [deskhood] sounds really well formed? How about [rockhood] or [concepthood] or [emotionhood]? So, [-hood] can actually only be added to a very reduced set of nouns, not simply to any noun the way that [-s] can. Therefore, we can say that [-hood] is not very productive. (Note, productivity is a gradient or scalar thing. Some derivational affixes are more productive than others. The suffix [-able], for example, is more productive than [-hood]. But, the generalization is still useful. Derivational affixes tend to be less productive than inflectional affixes. This test is useful, but since "productive" is a fuzzy concept, you have to walk with care.

• it appears "before" an inflectional affix (in general). This is just a way of saying that "inflectional" affixes are the last things we add, after we've built the main word up. So, if we have [govern + ment + s], we see that [-ment] is inside of the inflectional [-s]. Be careful, though, because we might have two derivational affixes in a word, as in [humid -ify -er -s] "humidifiers". Don't assume that [-er] is inflectional simply because [-ify] appears before it. Both [-ify] and [-er] are derivational, while our friend [-s], of course, is inflectional. This is a handy rule of thumb, but, as I note, you've got to be careful.

(Note: derivational affixes can be either prefixes or suffixes in ENGLISH. So, [un-] is a derivational prefix, while [-ment] is a derivational suffix.)

• it doesn't change the part of speech to the base that it's added to. So, if we add [-s] to the noun [cat], we've still got a [cats]. BE CAREFUL WITH THIS TEST. Why? Some derivational affixes don't change the part of speech. [-hood] is added to nouns to create new nouns. [re-] is a prefix that can be added to verbs to create new verbs.

• it is required by the syntax. See the comments immediately above. The third singular [-s] suffix of [run-s] will "agree" with the subject of a sentence like "Mary run-s quickly." The basic insight here is that inflectional has to do with relations between words, i.e. with information that is important to the syntax of the sentence.

• it is very productive. Think plural [-s] as a stellar example.

• occurs generally at the margin (outside edge) of a word. This goes back to our generalization that derivation tends to come inside of inflection.

h) Like sentences, words have internal structure!

How do we know this. Well, look at page 140 of Language Files and contemplate the tree for the word [un-use-able]. Note that this tree has hierarchical structure much in the same way that a syntax tree does. The root of the tree in this case is an Adjective node, indicating the part of speech of the whole word "unusable". What we see is that this root dominates a prefix [un-] and another Adjective node. This means that the word was formed by adding [un-] to the already morphologically complex word [use + able]. For its part, [use + able] consists of a verb root [use] and the adjective forming suffix [-able]. That is, we can see that the word is built up this way in our lexicon (the formal term for our mental dictionary which contains morphemes and the rules for their combination):

1) add -able to the root use to form [use[able] ]

2) add un- to the stem [use[able]] to form [un[use[able]]]

How do we know that this is the structure of this word? How do we know that [-un] isn't first added to the verb and that we don't then add [-able]. Simple. We CAN'T add [un-] to the verb [use] because [unuse] is NOT A POSSIBLE VERB. (See how our derivational prefix isn't completely productive.) We can, however, add [un-] to the adjective [useable]. So by this reasoning, we can build up the adjective first and then add [un-].

Ambiguity gives us more evidence in favor of viewing words as having internal structure. Consider the following word:

• un-lock-able

This word is ambiguous because it has two clear readings. It can mean "able to be unlocked" or "not able to be locked". Importantly, we can use a tree structure to disambiguate them, i.e. to reflect their different meanings by accurately representing the different internal structure of the word, depending on which meaning we want to characterize. Look at the trees on p. 141 of Language Files. Make sure you can explain why the top tree represents the meaning "not able to be locked" and why the bottom tree means "able to be unlocked". Be able to draw trees to disambiguate such words.

j) How are words created?

We've talked about how words can be broken down into morphemes, talked about types of morphemes, and talked about how words, like sentences, have internal structure. Now, we need to address the processes that we use to build up morphologically complex words. That's what we'll review here.

• Affixation: a way of building up words by adding prefixes, suffixes, infixes, or a combination of these. Example: [cat-s] is formed by affixation. In particular, by the suffixation of the plural morpheme [-s].

• Compounding: a way of building up words by combining two words. Examples: [blackbird], [air conditioner]. Note that compounding has its own special stress pattern. So, the words [WHITE HOUSE] are a noun phrase in which "white" describes "house", while [WHITE house], with more stress on [WHITE] is a compound. Note also that while [air conditioner] is a compound, the word [conditioner] is itself built up by affixation!

• Reduplication: We don't have this in English. But...reduplication is a way of building up words by repeating either a part or all of the base word. An example of total reduplication is found in the Indonesian word [rumahrumah] 'houses', which is built by reduplicating the root for house, [rumah]. Partial reduplication is found in the Philippine language Tagalog, where the future tense is expressed by reduplicating part of a verb. So we take the verb [bili] and reduplicate only the first two sounds to get the future form [bibili].

• Morpheme-Internal Changes: Sometimes "adding" morphemes results in changes to a part of the word. We have some cases of this in English, though not "productive" ones. Think about the past of the verb [ring]. Is it [ring-ed]? No. Rather, we change the vowel in the middle from [i] to [a] and we get the (irregular and thus non-productive) past form [rang]. That's an example of a word that is built up by a morpheme-internal change.

• Suppletion: Suppletion is about irregularity taken one step further. Suppletive forms are morphologically complex forms in which the connection between the base and the newly created form is not obvious at all. Examples are helpful. We know that the regular past is formed by suffixing [-ed]. (We'll return to this later in the course for a closer look). Anyway, what happens when we make the past of [go]? The answer is that we get [went], which has NOTHING to do with the root [go]. This is a case of suppletion.

• Blending: Blending is like compounding in that it involves combining words. BUT, a part of each word gets lost along the way. Here's some examples:

• motel (a blend of motor + hotel)
• smog (a blend smoke and fog)

l) Typology

Linguists classify languages in terms of way that they put morphemes together. Some languages have fairly "impoverished" morphological processes (English is one), while other languages have incredibly rich, complex morphological processes (Navajo is one). This section will review the basis ways that linguists classify things.
 
 

m) Analytic languages

What we mean by this term is that a language tends to not make words out of multiple morphemes. In a completely analytic language, every word would be made up of a single morpheme, and these one-morpheme words would then be combined to form sentences. That's why analytic languages are called isolating languages. I actually think "isolating" is a better, clearer term. Chinese is a very analytic language, and English is also fairly analytic. Analytic languages often have a rigid word order in their sentences, because they don't have lots of inflectional affixes to supply information about who is doing what to whom. So, if English had a rich case system, we might be able to say:

• man-object bites dog-subject
• dog-subject bites man-object

and have them mean the same thing. The word order wouldn't matter, because the subject and object affixes would give us the important information about who the biter and bitee are. This is, in fact, the way Latin works (and lots of other still "living" languages). But, since English doesn't have such affixes, we rely on stricter word order to supply the "biter" "bitee" information.

This is a cover term for languages with richer morphologies. That is, in synthetic languages, words tend to be morphologically complex, i.e. words tend to be built up by adding affixes. Here's a list of types of synthetic languages.

• Agglutinating. In agglutinating languages, the rule of thumb is that the morphemes that are joined together are easy to spot or pick apart. This is like your Turkish homework, where it was very easy to pick out each morpheme and say what part of speech it was. Hungarian is given as an example in Language Files on p. 166
• Fusional. Fusional languages combine morphemes in such a way as to make picking out the individual morphemes more difficult. Your book cites Spanish and Russian as examples. The point of the spanish explanation is to make you see that the stem meaning speak is hard to isolate. Is it "habl" or is it "habla". So, if the morphemes are easy to pick apart with nice, crisp boundaries, the language is more agglutinative. When the borders start looking messier between morphemes, we're talking fusional.
• Polysynthetic. Polysynthetic languages are best viewed as the opposite of analytic languages. If analytic languages represent one end of a continuum in which words tend to consist of a single morpheme, in Polysynthetic languages, words always consist of a combination of many morphemes which often result in full sentence meanings.

o) A point of confusion: stems and roots.

One thing that confuses people is the difference between a stem and a root. In simple terms, here's the difference. A root is always a single morpheme. By contrast, a stem can be (though it doesn't have to be) larger than a single morpheme. It's like squares and rectangles. A square is a rectangle, but not all rectangles are squares. If I have a word [cat-s], the morpheme [cat] is a root. It is also a stem. But, if I have a word like [unuseable], I have a root [do] to which I add a suffix [-able] to get a stem [doable]. Now, I can add another affix, [un-] to get the word [undoable].

Now, let's toss in the word base. A base is maximally general cover term for any morphological material that I add more material to (say via affixation) or that I copy (in reduplication) etc... So, in [undoable], [do] the root is the base to which I add [able]. Then, [doable] (the stem) is also a base to which I can add [un-]. Get it. If not, email me!