What is Morphology?, Slides of English

Download What is Morphology? and more Slides English in PDF only on Docsity!

What is Morphology?

Mark Aronoff and

Kirsten Fudeman

M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S 1

1 Thinking about Morphology and

Morphological Analysis

1.1 What is Morphology? 1 1.2 Morphemes 2 1.3 Morphology in Action 4 1.3.1 Novel words and word play 4 1.3.2 Abstract morphological facts 6 1.4 Background and Beliefs 9 1.5 Introduction to Morphological Analysis 12 1.5.1 Two basic approaches: analysis and synthesis 12 1.5.2 Analytic principles 14 1.5.3 Sample problems with solutions 17 1.6 Summary 21 Introduction to Kujamaat Jóola 22

mor·phol·o·gy : a study of the structure or form of something Merriam-Webster Unabridged

n 1.1 What is Morphology?

The term morphology is generally attributed to the German poet, novelist, playwright, and philosopher Johann Wolfgang von Goethe (1749–1832), who coined it early in the nineteenth century in a biological context. Its etymology is Greek: morph- means ‘shape, form’, and morphology is the study of form or forms. In biology morphology refers to the study of the form and structure of organisms, and in geology it refers to the study of the configuration and evolution of land forms. In linguistics morphology refers to the mental system involved in word formation or to the branch

M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S 3

Some readers may wonder why we have not broken -ation down further into two pieces, -ate and -ion , which function independently elsewhere. In this particular word they do not do so (cf. * reconsiderate ), and hence we treat -ation as a single morpheme. It is important to take very seriously the idea that the grammatical function of a morpheme, which may include its meaning, must be con- stant. Consider the English words lovely and quickly. They both end with the suffix -ly. But is it the same in both words? No – when we add - ly to the adjective quick , we create an adverb that describes how fast someone does something. But when we add -ly to the noun love , we create an adjective. What on the surface appears to be a single morpheme turns out to be two. One attaches to adjectives and creates adverbs; the other attaches to nouns and creates adjectives. There are two other sorts of affixes that you will encounter, infixes and circumfixes. Both are classic challenges to the notion of morpheme. Infixes are segmental strings that do not attach to the front or back of a word, but rather somewhere in the middle. The Tagalog infix -um- is illustrated below (McCarthy and Prince 1993: 101–5; French 1988). It creates an agent from a verb stem and appears before the first vowel of the word:

(1) Root -um- /sulat/ /s-um-ulat/ ‘one who wrote’ /gradwet/ /gr-um-adwet/ ‘one who graduated’

The existence of infixes challenges the traditional notion of a morpheme as an indivisible unit. We want to call the stem sulat ‘write’ a morpheme, and yet the infix -um- breaks it up. Yet this seems to be a property of -um- rather than one of sulat. Our definition of morphemes as the smallest linguistic pieces with a grammatical function survives this challenge. Circumfixes are affixes that come in two parts. One attaches to the front of the word, and the other to the back. Circumfixes are controver- sial because it is possible to analyze them as consisting of a prefix and a suffix that apply to a stem simultaneously. One example is Indonesian ke... -an. It applies to the stem besar ‘big’ to form a noun ke-besar-an meaning ‘bigness, greatness’ (MacDonald 1976: 63; Beard 1998: 62). Like infixes, the existence of circumfixes challenges the traditional notion of morpheme (but not the definition used here) because they involve discontinuity. We will not go any more deeply here into classical problems with morphemes, but the reader who would like to know more might consult Anderson (1992: 51–6).

M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S 3

4 M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S

n 1.3 Morphology in Action

We would like to explore the idea of morphology more deeply by examining some data. These are examples of morphology in action – morphological facts of everyday life.

n 1.3.1 Novel words and word play

If you had been walking down the street in Ithaca, New York, a few years ago, you might have looked up and seen a sign for the music store “Rebop,” a name that owes its inspiration to the jazz term rebop.^1 Rebop was originally one of the many nonsense expressions that jazz musicians threw into their vocal improvisations, starting in the early 1920s. In the 1940s, rebop became interchangeable with bebop , a term of similar origin, as the term for the rhythmically and harmonically eccentric music played by young black musicians. By the 1950s the name of this musical style was quite firmly established as simply bop.^2 Today, the original use of rebop is known only to cognoscenti, so that most people who pass by the store will be likely to interpret the word as composed of the word bop and the prefix re -, which means approximately ‘again’. This prefix can attach only to verbs, so we must interpret bop as a verb here. Rebop must therefore mean ‘bop again’, if it means anything at all. And this music store, appropriately, specialized in selling used CDs. There’s something going on here with English morphology. Of course, rebop is not a perfectly well-formed English word. The verb bop means something like ‘bounce’, but the prefix re- normally attaches only to a verb whose meaning denotes an accomplishment. The verb rebop there- fore makes little sense. But names of stores and products are designed to catch the consumer’s attention, not necessarily to make sense, and this one does so by exploiting people’s knowledge of English in a fairly complex way and breaking the rules so as to attract attention, as verbal art often does. Consider now the following phrases, taken from a Toni Braxton song: Unbreak my heart, uncry these tears. We have never seen anyone unbreak something, and you certainly can’t uncry tears, but every English speaker can understand these words. We all know what it means to unbreak somebody’s heart or to wish that one’s heart were unbroken. If we asked somebody, “unbreak my heart,” we would be asking them to reverse the process of having our heart

4 M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S

6 M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S

n 1.3.2 Abstract morphological facts

Now let’s move to some more abstract morphological facts. These are the kind of morphological facts that you don’t notice every day. They are so embedded in your language that you don’t even think about them. They are more common than the ones we have just looked at, but at the same time deeper and more complex. If you speak English and are concerned about your health, you might say:

(2) I eat one melon a day.

Let’s imagine that we are even more concerned about our health than you are. We don’t just eat one melon a day, rather:

(3) We eat two melons a day.

It is a fact about standard American or British English that we cannot say:

(4) *We eat two melon a day.

However, if we were speaking Indonesian or Japanese, we would say the equivalent of two melon ( three melon , four melon , etc.) because these languages don’t use morphological plurals in sentences like this.

(5) Indonesian: Saiga makan dua buah semangka (se) tiap hari I eat two fruit melon every day ‘I eat two melons every day.’ Japanese: mainichi futatsu-no meron-o tabemasu every.day two- gen melon-obj eat.imperf ‘I eat two melons every day.’ The morphological grammar of English tells us that we have to put an -s on melon whenever we are talking about more than one. This fact of English is so transparent that native speakers don’t notice it. If we happen to be speakers of a language without obligatory plural marking, however, we will notice it because we are going to have a lot of trouble with it. We have now observed something about English morphology. If a word is plural, it takes the suffix -s. Living creatures don’t eat only melons, however:

6 M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S

M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S 7

(6) The evil giant at the top of the beanstalk eats two melons, three fish, and four children a day.

Everyone agrees that fish is plural, but there is no plural marker. Children is also plural, but it has a very unusual plural suffix, -ren , plus an internal change: we say [TIld-] instead of [Tajld]. In other words, it’s not always the case that we mark plural words with an s- like thing; there are other ways in which we can mark plurals. Native speakers of English know this, and they do not need to think about it before making a plural. [Exercise 4] Consider the following:

(7) Today they claim that they will fix the clock tower by Friday, but yesterday they claimed that it would take at least a month.

In this example, we use two different forms of the verb claim. One is present tense, and the other is past. Again, this is not true for all languages. If we were speaking Vietnamese, for example, we wouldn’t make any distinction between claim and claimed – we wouldn’t mark the verb at all. If we were speaking Chinese, we would not distinguish between claim and claimed in a sentence like this, because the adverb zuótian ‘yesterday’ is sufficient to indicate past tense:

(8) jintian tamen shuo tamen xingqi wä ké yà xiu hÜo zhonglóu, today they say they Friday can fix well clock.tower ká shì zuótian tamen què shuo zhì shÜo xu yào yíge yuè but yesterday they however say at least need a month ‘Today they claim that they will fix the clock tower by Friday, but yesterday they claimed that it would take at least a month.’

If we were to leave out zuótian ‘yesterday’, we would need to use the particle le after the verb to show that the action took place in the past. In other words, whether or not a speaker must indicate past tense in Chinese depends on context. Notice what happens in English when we use some other verbs be- sides claim :

(9) Today they say... but yesterday they said... tell us told us know knew

That these verbs and others do not add -t , -d , or - @d to make their past tense is an elementary fact about English morphology. We’ll talk more about verbs like these later in the chapter.

M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S 7

M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S 9

of singular, dual, trial, and plural second person subject prefixes in Larike is illustrated below:

(13) Ai - rala iter- lawa pe?a- o? 2 sg.sub- chop.down 1pl.incl.sub- garden finish- qm ‘Did you (sg.) finish clearing our garden?’ Kalu au- ?anu, irua musti iruai - ?anu si?u. if 1 sg.sub- eat 2 du certainly 2 du.sub- eat also ‘If I eat, certainly you both will eat too.’ Kalu iridu - ta- ?eu, au- na- wela. if 2 tri.sub- neg- go 1 sg.sub-irr- go.home ‘If you three don’t want to go, I’m going home.’ Memang iri- hise tapi imi - ta- ?ari?i- truly 3 pl.nonhum- exist but 2pl.sub-neg- see- ri. 3 pl.nonhum.obj ‘They really do exist, but you (plural) didn’t see them.’

n 1.4 Background and Beliefs

This book is a general introduction to morphology and morphological analysis from the point of view of a morphologist. The purpose is not to advocate any particular theory or to give the truth (whatever that is), but rather to get you, the reader, to where you can look for it by yourself. Still, it is inevitable that some of our remarks will be colored by our own beliefs and background. We would therefore like to pre- sent some of our foundational beliefs about linguistics and linguistic methodology. First, we believe that languages differ from one another. You might be thinking, “Of course they do!” But we mean this in a very special way. Some linguists are always looking for ways that languages are similar, and at times, we do that, too. But we believe that if you focus only on the similarities between languages, you miss out on all of the exciting ways in which they differ. What’s more, you may find parallels and similarities where none really exist. We try to approach linguistic analysis with as open a mind as possible, and to do this, it is first necessary to appreciate the uniqueness and diversity of the world’s languages. Our second foundational belief is that languages , which we can write with a small l , are different from Language , with a capital L. There are

M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S 9

10 M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S

thousands of individual languages in the world. But we may also speak of language in general to mean the general phenomenon of Language that encompasses all individual languages. This Language is related to Noam Chomsky’s notion of Universal Grammar, which posits that languages are all alike in basic ways. There is an important distinction between these two uses of the word language and each is equally import- ant to linguistics. Individual languages have features that are not charac- teristic of Language in general. For example, one feature of English is that its regular way of forming plural nouns is to add /z/. We would never claim, however, that this is universally true, or that it is a property of Language. To tie this belief in with the preceding one, we strongly believe that morphological theory and morphological analysis must be grounded in morphological description. If we want to appreciate what morphology really is, it’s best to have some idea of what the morphology of individual languages is like. At the same time, we must have a reason- ably well-thought-out general theory of the morphology of Language, so that we can compare our descriptions of individual languages within a wider context. In short, linguists need to pay equal attention to both small-l language and capital-L Language. Our next belief is that morphology is a distinct component of lan- guages or grammars. If you are not already familiar with some of the controversy surrounding morphology, this needs an explanation. The fact that some languages, such as Vietnamese, do not have morpholo- gically complex words has led some people to conclude that morphology should not be a separate branch of linguistics. The reasoning is that lin- guistics is generally understood to deal with properties of all languages – more precisely, Language with a capital L. If there are languages that don’t have morphology, then morphology is not a property of all lan- guages, and morphological phenomena should be treated in syntax or phonology. We disagree. It has been shown elsewhere (e.g., Aronoff 1994) that there are aspects of morphology that cannot be attributed to syntax or phonology, or anything else. One piece of evidence that morphology is separate from syntax, phono- logy, and other branches of linguistics is that words in some languages are grouped into largely arbitrary classes that determine their forms in different environments. Latin nouns fall into five distinct classes, called declensions, which have little or nothing to do with syntax or phonology, and certainly cannot be explained by either. They are purely morpho- logical in their significance. The uniquely morphological nature of these classes is truly brought home by the fact that Latin nouns also fall into

12 M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S

n 1.5 Introduction to Morphological Analysis

n 1.5.1 Two basic approaches: analysis and synthesis

There are two complementary approaches to morphology, analytic and synthetic. The linguist needs both. The analytic approach has to do with breaking words down, and it is usually associated with American structuralist linguistics of the first half of the twentieth century. There is a good reason for this. These linguists were often dealing with languages that they had never encountered before, and there were no written grammars of these languages to guide them. It was therefore crucial that they should have very explicit methods of linguistic analysis. No matter what language we’re looking at, we need analytic methods that will be independent of the structures we are exam- ining; preconceived notions might interfere with an objective, scientific analysis. This is especially true when dealing with unfamiliar languages. The second approach to morphology is more often associated with theory than with methodology, perhaps unfairly. This is the synthetic approach. It basically says, “I have a lot of little pieces here. How do I put them together?” This question presupposes that you already know what the pieces are. So in a sense, analysis in some way must precede synthesis. Say that you’ve broken a clock and taken it apart, and now you have to put all the little pieces back together. There’s a catch: you don’t know how. You could always go by trial and error. But the most efficient way would be to have some theory of how the clock goes together. Synthesis really involves theory construction. From a morphological point of view, the synthetic question you ask is, “How does a speaker of a language produce a grammatically complex word when needed?” This question already assumes that you know what kinds of elementary pieces you are making the complex word out of. We think that one of the real problems of a morphological theory is that we don’t always have a good idea of what the pieces are. Syntacticians can supply us with some tools: case and number, for example, are ancient syntactic notions that we can use in our morphology. But the primary way in which morphologists determine the pieces they are dealing with is by examination of language data. They must pull words apart carefully, taking great care to note where each piece came from to begin with. We have described analysis and synthesis in terms of the morphologist studying language, but the two notions are equally applicable to speakers themselves. Speakers apply morphological analysis when they read or

M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S 13

hear a complex word that they have never encountered before. In order to understand it, they pull it apart and ask themselves whether they recognize any of the pieces. Speakers use synthesis whenever they create new forms from pre-existing pieces.

Read the caption in the following Motorola ad carefully. It contains an example of morphology in action – a striking morphological fact. Comment on it, relating it to the discussion of analytic and synthetic approaches to word- formation.

Who’d of thought that an electronic chip inside your car could help you avoid curbs, other cars, and best of all, Earl in repair.

M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S 15

This isn’t the whole story, as Principle 2 tells us:

(15) Principle 2 Forms with the same meaning but different sound shapes may be instances of the same morpheme if their distributions do not overlap.

In Kujamaat Jóola, for example, the stem /baj-/ has two possible shapes, [baj-] and [b@j-], but their distributions don’t overlap. [b@j-] occurs in the presence of a morpheme with an underlyingly tense vowel, but [baj-] does not. This non-overlapping distribution allows us to conclude that the two forms are instances of the same morpheme. When two or more instances of a given morpheme occur with different shapes, we call them allomorphs. Allomorphs were introduced above in section 1.2. The regular plural marker in English has several allomorphs – voiceless alveolar fricative /s/, voiced alveolar fricative /z/, schwa plus voiced alveolar fricative /z/, syllabic alveolar nasal /n/, and Ø – as shown in (16):

(16) seat-/s/ shade-/z/ hedg-/@z/ ox-/y/ fish-Ø

As in the previous example, the distributions of these forms do not over- lap, and they all have the same meaning. We can infer that they are instances of the same morpheme.

(17) Principle 3 Not all morphemes are segmental.

Normally, when we think of morphemes, we think of forms that can be pronounced in some sense, e.g., chicken , the , un- , -ize. But some morphemes can’t be pronounced on their own. They are dependent on other morph- emes for their realization. In English, for example, vowel alternations may serve to differentiate basic and past forms of the verb. We refer to these alternations as ablaut (as in 18):

(18) run ran speak spoke eat ate

We know that there is a past tense marker distinguishing the words in the second column from those in the first. But what is it? It is not the /æ/

16 M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S

of ran or the /o/ of spoke but rather the difference between these vowels and the vowels of the basic verb, which is not segmental at all. We must look at both the present and past tense forms of these verbs, because it is the contrast between them that is important. Another type of non- segmental morpheme in English is shown in (19):

(19) breathN breatheV clothN clotheV houseN houseV

In each pair, the noun ends in a voiceless fricative ([T, s]), while the verb ends in a voiced fricative ([D, z]). Assuming that the noun is basic, we say that the morpheme that marks the verbs consists of the phonological feature [+voice]. [Exercise 9] Although Principle 3 says that we can apply the term morpheme to the non-segmental alternations seen in (18) and (19), it is nonetheless the case that doing so is awkward. Pairs like run ∼ ran or breath ∼ breathe are more easily explained as processes than as concatenation of morphemes. In the next chapter we will further address this issue. In section 1.2 we briefly mentioned classical problems with morphemes in the context of infixation and circumfixation. The existence of non-segmental alternations such as those in (18) and (19) is another classical problem. The contrast between forms was crucial in (18) and (19). The notion of contrast can be further extended, leading to Principle 4:

(20) Principle 4 A morpheme may have zero as one of its allomorphs provided it has a non-zero allomorph.

Fish is an example of a word with a zero plural: one fish, two fish-Ø. We can say that it has a zero plural, and that this zero plural is an allomorph of the usual plural [z], because other words in the language, like frogs , have non-zero plurals. This is an analytic procedure, not a theoretical point. We cannot posit a zero unless it contrasts with some

Answers to morpheme-breakup exercise: a. pass/word b. spray/able c. child/hood/s d. auto/bio/graph/y e. co-/educ/at/ion/al

18 M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S

One form contains the meaning ‘big’ but not the morpheme -wewe. This is (21o) ko·yameilama ‘big female pig’. We recognize the piece ko·yame- ‘pig’, which also appears in (21m–n, p). Based on the minimal data we have, we can only deduce that like -wewe, -ilama means ‘big’, but that it attaches only to a certain class of noun. Both (21o) ko·yameilama ‘big female pig’ and (21n) ko·yamewewe ‘big male pig’ appear to have the same stem, but since one refers to a female animal and the other to a male animal, such a situation would not be unprecedented. Based on (21b–c) ikalsosol ‘his old house’ and ikalci·n ‘his little house’ we isolate the pieces -sosol ‘old’ and -ci·n ‘little’. This analysis is affirmed when we look at other words in the data set, such as (21e–f) komitsosol ‘old cooking-pot’ and komitci·n ‘little cooking-pot’, which contain the same pieces. We can also isolate komit- ‘cooking-pot’. In all, we can isolate the following morphemes:

(22) ikal- ‘his house’ komit- ‘cooking-pot’ ko·yame- ‘pig’ petat- ‘mat’ -ci·n ‘little’ -sosol ‘old’ -ilama ‘big’ (occurs with stem meaning ‘female pig’) -wewe ‘big’ (occurs with stems meaning ‘his house’, ‘cooking- pot’, ‘mat’, ‘male pig’) -meh plural marker

This exercise was fairly simple in the sense that there were no allomorphs, and the morphology was entirely morphemic – it did not interact with any non-segmental phenomena. However, there were a few difficulties. One was the fact that we did not have enough data to break up ikal- ‘his house’, and yet, since the English gloss clearly has two parts, you may have been tempted to break it into two parts, too. A second difficulty was the presence of both -ilama and -wewe ‘big’. Again, we did not have enough data to understand fully what their difference is. Occasionally uncertainty is something that morphologists have to accept when working with published data sets and written grammars. Some- times there are gaps in what is presented. Morphologists doing field research have the advantage of native speaker consultants whom they can ask. But in order to ask the right questions, it is important that morphologists alternate data collection with data analysis and not wait to get back home to analyze their findings.

M O R P H O L O G Y A N D M O R P H O L O G I C A L A N A L Y S I S 19

A final observation is that this data set was not presented in the IPA. For example, you probably were not familiar with Nida’s convention for marking long vowels: a raised dot, as in -ci·n ‘little’. This fact in itself should not have posed any problems. It is often possible to isolate morphemes, particularly when there are no allomorphs or phonological interactions between them, whether or not we fully understand the transcription system. That was the case here. However, the use of a non-standard transcription system may make a problem set seem more daunting. As explained in the prefatory remarks to this book, we chose to retain non-standard transcription systems despite the difficulties they present because as a linguist you will be faced with them time and time again. We hope that the experience you gain in this book will help you deal with such systems in your own research. Our next sample problem set comes from French. It addresses different issues than the Aztec data discussed above. The French adjectives in the first column are masculine, and those in the second are feminine. Your task is to determine how masculine and feminine adjectives are differen- tiated and to outline a possible analysis. You may ignore changes in vowel quality.

(23) Masculine Feminine a. gros [g‰o] grosse [g‰os] ‘fat’ b. mauvais [movE] mauvaise [movEz] ‘bad’ c. heureux [ø‰ø] heureuse [ø‰øz] ‘happy’ d. petit [p@ti] petite [p@tit] ‘small’ e. grand [g‰Ñ] grande [g‰Ñd] ‘big’ f. froid [f‰wa] froide [f‰wad] ‘cold’ g. soûl [su] soûle [sul] ‘drunk’ h. bon [bõ] bonne [bOn] ‘good’ i. frais [f‰E] fraîche [f‰ES] ‘fresh’ j. long [lõ] longue [lõg] ‘long’ k. premier [p‰œmje] première [p‰œmjE‰] ‘first’ l. entier [Ñtje] entière [ÑtjE‰] ‘entire’ m. gentil [ZÑti] gentille [ZÑtij] ‘kind’ n. net [nEt] nette [nEt] ‘clean’

As with the Aztec set, you should limit yourself to the data provided, although some of you may know French. One way to begin is to see whether there is a single morpheme, which may or may not have allomorphs, that signals the difference between masculine and feminine. There is not. Masculine and feminine adjectives are differentiated by an alternation between Ö and [s] in (23a), Ö and [z]