An Introduction to Linguistics 1

Course Notes for Linguistics 201 Section E, Instructor A. Farudi

Fall 2008

1These lecture notes and exercises are a combination and adaptation of materials written by the follow-ing: Ed Burstynsky, Kathryn Flack, Shigeto Kawahara, John Kingston, Keir Moulton, Monica Sieh, YouriZabbal, and Cherlon Cussery, whose editing and organization of the previous authors’ work I am drawingfrom most heavily. Chapter 6 is authored exclusively by Aynat Rubinstein, used with permission here.

2

Chapter 1

Introduction

1.1 The big picture: Knowledge of language

As we will be in this class, linguists are interested in the knowledge of language that every nativespeaker has. This knowledge is what allows every native speaker to speak and understand hernative language. What do we mean by knowledge? Linguistic knowledge, in the sense in whichlinguists are interested in it, has little to do with the sort of conventional grammatical rules weare taught, for example, in elementary and middle school English courses. It refers to the huge,complex array of facts about how his or her native language works that every normal humanknows—and this is the fascinating part—intuitively, instinctively without having to try. We don’thave to think about our knowledge of language because it is so ingrained in our consciousnessthat we often don’t even realize we have it. The following exercise is intended to illustrate thatyou have knowledge of English that you never knew you had.

Exercise 1: Things you know about English you never knew you knew.Classify the following words into one of the following three categories: (i) English words,(ii)words that are not English but could be, or (iii) impossible English words.

(1) a. bnick impossible wordb. brickc. blickd. bticke. botickf. trickg. tlickh. tnicki. nbidj. glind

k. mgind

How did you know the difference between a word that is possible in English (but not an actualword) and a word that would simply never be possible? You have, presumably, never heard orread these words before, so how were you able to make this distinction? (There is not necessarily

3

one correct answer to this question, but you should try to characterize the way you reacted to thewords above, and how you made your choice for each one.)

(2) Fill in the blanks in the following sentences with any word that makes sense.a. Harry has a blick cat.b. John ate sixteen fertuples.

Now, based on what these sentences show about the words blick and fertuples, characterize thefollowing sentences as grammatical sentences of English or not. If the sentence is grammatical,place a X at the beginning of the sentence; if it it not, place a * beside it. (Example: *John boughta fertuple cupcake.)

(3) a. Sarah also has a blick.b. Because Mary didn’t like them, she only ate one fertuple.c. Harry has a blick cat too.d. John bought many fertuple.

Despite the fact that these words are nonsense, you most likely had strong intuitions aboutwhether or not the above sentences were good sentences of English or not. How did you knowthis? The following question shows that we can produce good and bad sentences involvingnonsense words as well as recognize them.

Fill in the blanks with forms of the nonsense words given.

(4) a. I have one wug, and you have twob. My car is old and grish. It drives reallyc. I like to tunick. I am ??? right now, and I yesterday, too.

Classify each of the sentences below in one of three categories: (i) a good, normal, possiblesentence of English that makes sense; (ii) a sentence that is just nonsense; (iii) a sentence that isnot possible at all in English (say whether you can glean from these impossible sentences somemeaning? What is the intended meaning of each sentence?)

(5) a. What house did you wonder who built?b. Colorless green ideas sleep furiously.c. Ideas colorless sleep green furiously.d. So, like, are you IMing him or what?e. Hamburgers some John devoured quickly they.f. Mary thinks I dont like herself.g. Happy paperclips usually write doughnuts out of the sky.h. There’s many people I wanna talk to.i. Run children the school of out.

What is the difference between a nonsense sentence and a sentence that is simply impossible?What is wrong with the nonsense sentences? What is wrong with the impossible sentences?

4

Can impossible sentences make sense? What does this tell us about the difference between whatmakes a “possible sentence” in English and a “meaningful sentence”?

Your answers in this exercise show that you know things about English that you have never beentaught even things you have never heard or read. It is very likely that you have never heard theword wug, but you recognized that if there were such a thing as a wug and if there were two ofthem you would, instinctively, say wugs. There were also sentences which you have never heardbefore, but you knew they could never be uttered as sentences of English. These were comparedto sentences which were nonsense but nevertheless could in principle be good English sentences,even if in some cases the resulting meaning was silly, or simply nonsensical.

1.1.1 Grammaticality and Grammar: modeling your knowledge

In short, what the above exercise revealed is that we have intuitions about whether words orsentences are possible sentences of our native language. Linguistic intuitions are what allow usto make judgments about what is grammatical or not. Words and sentences that are possible ina particular language, according to the speakers’ intuitions, are called grammatical. This courseinvestigates the source of our linguistic intuitions; that is, it investigates the components of ourlinguistic knowledge that allow us to make judgments about what is grammatical. We will seethat you have a knowledge of your native language (whether English, or Japanese, or Swahili)that enables you to distinguish between grammatical and ungrammatical words, sounds, andsentences. We will talk about this knowledge as a grammar, a certain mental system that allowsus to speak and understand our native language.

1.1.2 Side note: Prescriptive vs. Descriptive Grammar

The grammaticality of a particular sound combination, word, or sentence is determined byspeakers’ intuitions about whether that word, sentence, or sound combination is possible intheir native language. What is grammatical, according to the linguistic definition of the term,is whatever (word, sentence, sound combination, phrase, etc.) a native speaker of a any givenlanguage accepts as part of their language.

Linguistics is interested in describing (and explaining) what is grammatical. However, theterm grammatical is often used in a different sense, which you may be more familiar with.According to this sense of the term, what is grammatical is determined by a set of rules inventedby people who advocate that English should be written and spoken in a certain way. This is aprescriptive grammar, which prescribes rather than describes what is grammatical. For instance,many prescriptive grammar books tell you not to end a sentence with a preposition, as in This isthe person I spoke to.

We are not interested in prescriptive grammar, because it is unscientific and misses the im-portant point about your complex knowledge of language. (Notice that no grammar book evermakes the distinction, let alone tries to describe and explain, the ungrammaticality of the sen-tences in the exercise we did.)

Some descriptively ‘right’ but prescriptively ‘wrong’ sentences include:

(6) a. I aint seen nothing.b. Each person should wash their own dish.

5

c. To boldly go where no one has gone before.d. He is taller than me.

• How do we know the above sentences are grammatical, in the linguistic sense?

Linguists further distinguish between a speaker’s knowledge of language (their grammar) anda speaker’s use of language. Speakers’ knowledge of language is what allows them to produceand understand an infinite number of utterances, including new utterances. This knowledgeof language is often called linguistic competence. A speaker’s use of language, in contrast,is the actual use of language in specific situations and depends on other cognitive factors andinteraction with the world, but it is driven by linguistic competence. A speaker’s language useis called performance. In this course, we will be studying linguistic competence. A speaker’scompetence is, essentially, the result of having a grammar. In fact, for the sake of simplicity, youcan think of the notions knowledge of language, linguistic competence and grammar as equivalents ofone another.

(7) Knowledge of language = linguistic competence = grammar

What is a grammar? Why and how does it give us knowledge about linguistic utterances wehave never heard or read before? We will see that grammars are discrete combinatorial systems.They are discrete in the sense that they have a finite set of individually distinguishable parts. Theyare combinatorial in the sense that we can combine these parts to form and understand infinitelynew and diverse utterances. Grammars are also modular, in the sense that they have differentcomponents responsible for our knowledge of different parts of language — our knowledgeof sound sequences as opposed to our knowledge of sentences, for example. We are going tostudy grammars by looking at these different components. The components we will focus on aredescribed below.

Phonetics: The study of physical sounds.How do we produce sounds? How are sounds transmitted from one person’s mouth to anotherperson’s ear? How do we perceive sounds?

Phonology: The study of how people manipulate sounds.What are the possible sounds in a particular language? What are possible and impossiblecombinations? How do we manipulate sounds?

Morphology: The study of word structure and word formation.How do we make (new) words? Is there an internal structure of words? If so, what does it looklike?

Syntax: The study of sentence structures.How do we combine words to make a sentence? What are the rules for forming sentences thatwe have never heard before?

Semantics: The study of meaning. How do we understand the meaning of individualsentences?

Pragmatics: The study of how context influences our interpretation of meaning.How do we understand the meaning of individual sentences within particular contexts?

6

The grammar is a system that combines small pieces of language into large ones. This allowsus to judge what is and is not a possible sound, word, or sentence even if we haven’t heardor read the combinations resulting in these particular sounds, words, or sentences before. Everynative English speaker’s grammar encodes the information that English words cannot begin withpairs of consonant sounds like bn or mb or mg, all of which were found to be impossible in theexercise above. We will see, when we study sound systems in our next unit, that there is a patternto these impossible consonant combinations, and we will be able to state this as a general rule inyour grammar of English.

Now recall (or look back at if necessary) the second group of questions in the exercise above.You were given nonsense words, but you nevertheless knew certain things about them. Forinstance, you knew, instinctively, that ‘blick’ had the properties of an adjective, like grey or happy,since it occurred in the position that adjectives do: a blick cat. We have evidence that you knewit was an adjective because you judged the sentence Sarah also has a blick, in which blick was usedas a noun, rather than an adjective, to be ungrammatical.

And in the last set of questions, which addressed entire sentences, we discovered that yourknowledge of English allows you to judge that some sentences, sentences you have never heardbefore, could never be possible sentences. The examples also showed that sentences you’ve neverheard before, even some that are nonsense, can still be grammatical.

This ability, to make judgments about and produce sentences you have never heard or pro-duced before, is what we call productivity of language: the ability to produce and understandan infinite number of novel utterances based on your knowledge of English.

1.2 What is language? Human language vs. Animal Communication

Answering this question is closely tied to answering a larger question: what distinguishes hu-mans from all other species? Although the members of many species communicate with oneanother in subtle and elaborate ways, only humans use language for this purpose. So, we canrephrase the original question, and ask how language differs from the ways other species com-municate. The properties that, taken together, distinguish human language from other systemsof communication are listed in (8). Nearly all of the features of human language listed in Table1 can be found individually in the systems of communication used by other animals, so whatmakes a system of communication a language is possessing all of these features.

(8) Design features of language (adapted from Hockett 1960) The features marked out by boldface are generally agreed to be of greatest importance in defining what makes a system ofcommunication a language.

a. Arbitrariness - The form of the signal, whether it’s a sound or a movement, bears norelationship to its meaning.

b. Discreteness - Speech and sign are composed of a small set of auditorily or visuallydistinct units or elements.

c. Productivity - Speech and sign permit an infinite variety of meaningful utterances tobe expressed by combining the discrete elements into new signals.

d. Duality of patterning - The most basic elements of linguistic signals speech sounds(or movements in sign language) are themselves meaningless, but they combine to-

7

gether in particular patterns to form larger meaningful elements, e.g. words or sen-tences, which themselves combine together in particular patterns.

e. Traditional transmission - Particular languages are passed on from one generation tothe next via the physical environment, rather than genetic inheritance.

1.3 Children and Grammar

The last design feature, traditional transmission, refers to the way we learn, or acquire, languageas very young children. Consider an adult—perhaps you yourself are an example—of a personwho, while not American by birth, moved to the United States as a baby. Such a person will be anative speaker of English simply in virtue of having been in an environment in which he or shewas exposed to English at the appropriate age. She will not necessarily be a native speaker ofwhatever language her parents speak natively. This observation brings us to the question of howwe come to gain our knowledge of our language as young children. As long as a child does notsuffer from a pathology that specifically prevents he or she from learning a language, a child canlearn any human language merely from observing that language in use. The infant’s capacityto learn any human language, combined with the huge variety of human languages that existin the world, suggest that what evolved in humans was a capacity to learn a language, not anyparticular language. No other species has the same capacity to learn a human language, even asthe result of rigorous instruction.

Many linguists think that what makes it possible for a child to learn any human language isthat he or she is born with the knowledge of what a human language can be, a potential gram-mar we might say. Put differently, an infant is born with the knowledge, not of any particulargrammar, but with the knowledge of what is a possible grammar. This knowledge, which, byhypothesis, all infants are born with, is referred to as universal grammar.

Universal grammar is the idea that humans are born with a knowledge of a universal grammar,which is the capacity to learn any human language. This hypothesis comes from Chomskyand is widely but not universally accepted, because some claim that not all languages sharecommon traits. Moreover, some argue that learning a language does not require any ‘pre-existing’knowledge about what a possible language can be.

Besides making it possible for an infant to learn any human language, a universal grammarwould also make all languages resemble one another in fundamental ways. The possibility thatsuch resemblances must exist among all human languages also makes the linguist’s analyticaltask easier, because a property of one language is likely to be found in another. This idea haseven led linguists to search for what are called language (linguistic) universals, i.e. propertiesfound in all languages. Because individual languages differ from one another in many obviousways, this search has been very hard, but it has also been very fruitful as well.

To see why linguists think that we are born with a knowledge of what a possible grammaris, consider the following. All children learn language quickly and automatically without beingtaught or being corrected; and, as early as five, children attain a roughly adult-like knowledgeof their language (with the exception of vocabulary, the words). Adults are nothing like childrenwhen it comes to learning language. Adults learn a language slowly, with conscious effort, andwe need instruction, repetition, and correction to learn a language, as you will know if you haveever taken a foreign language class. In comparison, it seems as though children just end upspeaking a language as a natural consequence of growing up. As the linguist Noam Chomsky

8

says, “The child doesnt say, ‘It would be nice to speak a language, I think I’ll learn one’ anymore than he says ‘It would be nice to have a liver, I think I’ll grow one.’ Chomsky is saying thatspeaking your native language comes about as a biological development, one that just happensby virtue of our genetic make-up and they way we grow and develop as humans. That is whywhen we talk about children coming to have knowledge of their native language, we use theterm acquire instead of ‘learn’. Learning a language is what we as adults do when we want tospeak a language. Linguists avoid the word ‘learn’ for children because it implies a consciouseffort. Children do not learn language. They come to know it (acquire it) automatically, usuallydisregarding correction.

Here is some evidence for this view. First, there is overwhelming evidence suggesting thatchildren fully acquire knowledge of a language quickly and completely even though the languagethey hear spoken around them is incomplete. Second, they show signs of forming generalizationsand reformulating incorrect generalizations. Third, they do this without imitating adults and, infact, consistently ignore adult correction. They appear to form their own hypotheses. Here aretwo examples that clearly illustrate how children ignore correction and do not imitate adults.

(9) a. Child: My teacher holded the baby rabbits and we patted them.b. Adult: Did you say your teacher held the baby rabbits?c. Child: Yes.d. Adult: What did you say she did?e. Child: She holded the baby rabbits and we patted them.f. Adult: Did you say she held them tightly?g. Child: No, she holded them loosely.

(10) a. Child: Nobody dont like me.b. Adult: No, say “Nobody likes me.”c. Child: Nobody dont like me. (repeated 8 times!)d. Adult: Now, listen carefully, say “Nobody likes me.”e. Child: Oh, nobody dont likes me.

The universal grammar hypothesis predicts that any normal healthy child with enough ex-posure (recall that by exposure, we don’t mean training or instruction of any sort, but simplyhearing the language spoken around them) will learn the language(s) to which they are exposed.Given this prediction, we might ask whether there are children who dont acquire full knowledgeof a language? Cases are rare, and evidence suggests that children who dont acquire completeknowledge of a language are either deprived of language in their environment or suffer frombiological, neurological or emotional disorder or trauma. If the problem is caught early enough,the child can learn language. Otherwise, the child never seems to learn language fully. Here aresome examples:

VictorA famous example is the ‘wolf child’ Victor of Aveyron, who was abandoned by his parents in aforest in 18th century France. He was discovered at age 12 in the woods near Saint Sernin surRance and taken in by a naturalist named Itard who tried to teach the boy French. Itard failedto successfully teach Victor adequate language skills.

9

GenieGenie was discovered in Los Angeles, in 1970. She was 13 years old and had been locked in acloset and tied to her potty since she was an infant. Her parents punished her for makingsounds and she heard little speech around her. Following her discovery, Genie was put into ahealthy, nonabusive environment and taken into the care of professional psychologists, a few ofwhom worked consistently and patiently with Genie to teach her English. While she madesignificant progress in expanding her vocabulary, that is, in learning words, her ability to formcomplex, novel sentences remained quite lacking:

(11) Examples of sentences produced by Geniea. Mike paint.b. Applesauce buy store.c. Neal come happy. Neal not sad.d. Genie have Momma have baby grow up.

This is particularly revealing in view of what how we have defined a grammar—as a system ofrules that allow us to put parts (words) together to make more complex linguistic strings (in thiscase, sentences).

ChelseaChelsea was diagnosed deaf when she was 31 years old. Until then, she was thought to beretarded. After being fitted with a hearing aid, Chelsea was able to hear. She got a job andlearned a large vocabulary but her ability to form sentences was deficient.

IsabelleThis girl has a similar background as Genie but was discovered at the age of 6.5. Ayear-and-a-half later, she acquired full knowledge of English.

A more critical look...

• What are some possible confounds with using these particular cases of children who failed toacquire language as evidence for the critical period?

From these examples many have reached the conclusion that language, as Chomsky suggests,is biologically determined. There are general criteria set out by the neurologist Eric Lenneberg todetermine if a behavior is biologically determined,that is, if a behavior is instinctive. The idea isthat a behavior is biologically determined if it meets some or all the following three criteria. Wecan use these criteria to help us determine if indeed human language is instinctual.

(12) Lenneberg’s criteriaa. Biologically-determined behavior follows its own schedule:

-Develops within a critical period or not at all-May arise before it is needed- Emerges in steps

b. Everyone develops the instinct-Emerges without a conscious decision-Emerges in spite of deficits

10

c. Instinct emerges without prodding-Not arrived at by imitation-Not helped by correction-Emerges at the same rate and arrives at the same result

The human language ability meets all three criteria.Of course, however, the knowledge youhave of grammar is not completely biologically determined. That would mean we all wouldspeak the same language, just as we all have the same kind of liver. Of course, that is not right:if we grow up in an English speaking environment we acquire English, if we were raised in aJapanese house, we would speak Japanese. And a child in a mixed English-Japanese-speakingenvironment would acquire both a grammar for English and for Japanese. So human beings havea language instinct, a biological blueprint of what language is supposed to be like, a universalgrammar. By listening to actual speech in their environment, children subconsciously assimilatelanguage specific details. They draw generalizations about what they hear. Children form theiractual grammars by using the biological blueprint in their brains and filling it in with language-specific properties they observe. The acquisition process is unconscious and independent ofinstruction. In this sense, children are somewhat like little language machines. They cant controlthe input, but once they’ve received the input, they know how to process it.

1.4 A myth about grammar

It is some times said that such-and-such a language has “no grammar”, or “less grammar” ora more “primitive grammar”. This is a myth. All languages have a grammar. If a language isspoken or signed, it has a phonetic and phonological system. Since it has words and sentences,it has a morphology and a syntax. These words and sentences have meaning, so the languagethey form has a semantics. The fact is that all grammars are equal. There are no “primitive” or“inferior” languages. This is especially important to keep in mind when considering differentvarieties/dialects of a language. Dialects of a language are, roughly speaking, different versionsof the same language that differ in minimal but systematic ways from each other and which arespoken by specific groups of people within the larger language community as a whole. AfricanAmerican English and the Brooklyn dialect of English are excellent examples of the way thatminority dialects can fall victim to such linguistic stereotypes. While they are often (falsely)perceived as being grammatically inferior to Standard English, in reality they are languages ofthe same complexity, subject to the formulation of linguistic generalizations of the same level ofabstraction as those linguists make about standard English.

1.5 A few words about writing

We perceive language, we speak language, and we write and read it. However, the ability tospeak and perceive language arises long before we learn how to read and write it. Readingand writing are learned skills. As such, they really dont represent any real part of linguisticcompetence. Linguists in general do not analyze or make generalizations about the written word,only about spoken language. Writing systems are highly conventionalized ways of recordinglanguage. They work well enough within a particular language to represent words but werenever designed to capture sounds and consequently speech in other languages. They are also

11

not always consistent, as we can see from the following observations about the relation betweenEnglish spelling and the sounds it is used to represent:

One could say that the spelling ‘ghoti’ can be pronounced the same as the word fish: ‘gh’ in laughis pronounced f ; ‘o’ in women is pronounced i, ‘ti’ in nation is pronounced ‘sh’.

While this is an obviously extreme example, it stands as an irrefutable demonstration of theintricate and highly conventional relationship between letters and sounds in English. Englishspelling will prove inadequate for representing speech sounds, which is something we will do atthe end of the semester when we study phonetics and phonology.

A more critical look...

• Why is this example, while useful in illustrating the point, perhaps not actually a legitimate char-acterization of the spelling-

Suggestions for further reading

Evidence for the language instinct and the critical period:Pinker, Steven. 1994. The Language Instinct.’ HarperCollins. Chapters 1 and 2.

On dialects: what they are, how they develop, how to distinguish a dialect from a language:Fromkin et. al. An Introduction to Language 8th Edition, pp. 409-420

General references:Hockett, Charles F. 1960. The Origin of Speech. Scientific American. 203:88-96.O’Grady, William, et al. 2005.

Contemporary Linguistic Analysis. Bedford-St. Martin’s. Fifth edition. Chapters 1 and 11.

12

Chapter 2

Phonetic foundations

2.1 The sounds of English and their features

As you will recall from the last chapter, the study of phonology focuses on the properties ofsounds and how those properties affect the way that various languages allow sounds to be com-bined into syllables and words. Before we delve into exploration of the sound patterns resultingfrom the combinatorial possibilities in different languages, we have to understand somethingabout the physical properties of sounds. The study of phonetics is the study of the articulatoryand acoustic properties of sounds. Articulation of sounds refers to the gestures made by the vocalorgans in the production of the various sounds employed in speech, while acoustics refers to thenature of the sound energy associated with those sounds—that is, to the properties of the soundwaves emitted by the production of different sounds. We are going to focus in this course onthe articulatory properties of sounds, but, as you should at least be familiar with each kind ofphonetic study, we will begin with a short consideration of the nature of acoustic properties.

2.1.1 Acoustic properties of sounds

Humans produce a wide variety of different types of sounds—yell, laugh, cry, whistle, etc.—but only a particular subset of these sounds are involved in speech. Like all other sounds weproduce, the sounds employed in language are waves of sound energy. All sounds begin withan exhalation of air from the lungs. When this airstream is released through the mouth, itcreates a disturbance in the surrounding air. This disturbance—a small, quick variation in theair pressure spreading through the air—is the sound wave. Obviously, however, each soundin a string of speech sounds is different; otherwise, we would be incapable of perceiving themeaningful difference between a pair of words such as say and day, to name just one of countlessexamples. The differences we perceive among sounds have two sources. The first concernsproperties of the sound wave itself; these properties, as noted above, are the acoustic propertiesof the sound. You are probably familiar, for instance, with the notions of pitch and loudness.Differences in the pitch of different sounds are the result of differences in how quickly or slowlythe change in air pressure that results from a given sound repeats itself as it travels through theair. The loudness refers to the size of the variations in air pressure.

Neither the pitch nor the loudness of a sound or a sequence of sound has any affect on(lexical) meaning in English. Whether I say the word coming with a falling intonation, as when I

13

yell Coming! to announce that I’ll be there in a second, or with a rising intonation, as when I sayComing? to ask you if you are going to join me or not, the word still has the same meaning. Thisis not the case in all languages; in Mandarin Chinese, for example, the single syllable ma has fourdifferent meanings depending on the pitch with which it is pronounced. This is shown in (13),where the diacritics (the marks above the letter a) represent differences in the pitch with whichthe word is pronounced.

(13) a. ma ‘mother’b. ma ‘hemp’c. ma ‘horse’d. ma ‘to scold’

We are not going to study such acoustic differences between sounds in this class, but youshould be aware that they exist, since they are a way of making linguistically significant distinc-tions among sounds not only in Chinese, but also the African language Yoruba, and many otherof the world’s languages.

2.1.2 Articulatory properties of sounds

The other source of differences among speech sounds, which is employed by English and indeedby all of the world’s languages, and is the one that we are going to concentrate on in this course,is the differences in articulation involved in the production of each sound. Articulation refers tothe movements, or gestures, that are made by the various organs used in producing the sounds ofa given language, which include: the vocal cords, the tongue, the lips, and the top of the mouth.[A diagram illustrating the position of these vocal organs, or articulators, in more detail will bedistributed in class.]

A stream of air exhaled from the lungs is manipulated by these organs in various ways as itpasses up through the vocal tract and out through the mouth. Consonant sounds result whenthe air flowing through the vocal tract is obstructed in some ways, and it is the speech organsinvolved in articulation that create this obstruction. Broadly speaking, the resulting sound isaffected by the way the air flows through the vocal tract when a sound is produced (the mannerof articulation), and the configuration formed by the movement of the lips, tongue, and otherrelevant speech organs (the place of articulation). Let’s consider each of these in turn.

2.1.3 Manner of articulation: how the air is moving through your vocal tract

The different manners of articulation distinguish the different ways that air can move through thevocal tract on its journey from the lungs, up the vocal tract, and out of the mouth. The followingare some of the most common manners of articulation. The description of the nature of theairflow associated with each one is followed by an exercise you can try yourself to check yourunderstanding of what your mouth is doing when it produces the type of sound in question.You will notice that all of the sounds exemplifying each type below are consonants. As we willsee below when we move on to consider vowels, this is not an accident. While consonants resultfrom a constriction of the air flow, no such constriction is involved in the production of vowels,which are characterized by a relatively free flow of air through the mouth. Therefore, it doesnot make sense to talk about manner of articulation in the same way that we talk about it withrespect to consonants.

14

Stop: Due to complete closure of the vocal tract, air is blocked from coming out of the mouth orthe nose. No air comes out of the mouth or nose at first, because the tongue or lips are fullyclosed and block it; then the tongue or lips open up and air comes out (sometimes in a strongburst).Examples: The first sound of English pat, bat, catTest: Put your mouth in position to start saying the sound. Try to exhale. If you can’t make airleave your lungs because your vocal tract is completely closed, it’s a stop.

Nasal: The velum lowers, allowing air to come out through the nose, and prohibiting air fromcoming out of the mouth.Examples: The first sound of English nose, moreTest: Rest your fingers on the bridge of your nose and make the sound; if your nose vibrates, itis a nasal. Also, hold your nose and make the sound. If your mouth bizarrely fills up with airand/or you can’t hold the sound for long, then it’s a nasal.

Fricative: The tongue or lips come very close to each other and make a very small opening. Theair hisses through this small opening, making a (relatively) noisy, friction-like sound (yourtongue may also come close to the teeth or the roof of the mouth).Examples: The first sound of English sad, zany, shiny five, vainTest: Make the sound for a while, and put your hand in front of your mouth. If you hearcontinuous hissing, and/or can feel air quickly leaving your mouth, it is a fricative.

Affricate: An affricates is a combination of a stop and fricative. The mouth is closed at first, asin a stop; then it opens a little bit into position for a fricative.Examples: The first and final sounds of judge, church; the final sound of matchTest: If the sound has two parts, and the first part passes the stop test and the second partpasses the fricative test, then it is an affricate.

Approximant: The vocal tract is constricted as in the production of a fricative, but to a smallerextent than required for a noisy (hissing) airstream.Examples: The first sound of rat, way, yellTest: N/A (See below)

Lateral: The tongue touches the roof of the mouth but without contacting the teeth at the sides.Examples: The first sound of laugh, loudTest: N/A (See below)

Tests for distinguishing the final two types of articulations—approximants and laterals—-are notincluded in the list above because it is much harder to pinpoint the nature of the constrictioninvolved in these sounds. This is because they are all articulations where the tongue makes littleor no contact with the roof of the mouth, allowing a lot of air to flow out of the mouth. In fact,the sounds classified as approximants and laterals are some times referred to as semi-vowels,since there is so much air flowing out of the mouth, as is true also of vowels. The practical resultis that there isn’t a good test for any of these sounds. If you cannot clearly identify a sound asone of the others, then it is likely one of these.

15

2.1.4 Place of Articulation: what your lips and tongue are doing

Bilabial: Both of your lips are touching.Examples: The first sound of English bat, pat

Labiodental: Your teeth and lips are touching.Examples: The first sound of English fan, vain

Interdental: Your tongue is between your teeth.Examples: The first sound of English then, this, thick

Alveolar: The alveolar ridge is that part of the top of your mouth that is right behind yourteeth. In alveolar sounds, your tongue touches right behind your teeth.Examples: The first sound of English now, toy, day

Palatal: The palate is sometimes called the hard palate; it is behind your alveolar ridge. If youmove your tongue backwards from behind your teeth, youll feel the roof of your mouth gettingharder. This is your palate. In palatal sounds, your tongue touches this hard area.Examples: The first sound of English show, yes, child

Velar: The velum is sometimes called the soft palate. Its located behind the (hard) palate. Velarsounds are made when the back of your tongue touches this softer part toward the back of yourmouth.Examples: The first sound of English kid, girl, wish; the final sound of English sing

Glottal: The glottis is the space between your vocal chords. Glottal sounds are made when airpasses through this space without any other constriction in the throat or mouth.Examples: The first sound of English how, head; the sound between the two syllables of theinterjection Uh-oh.

In addition to these two properties, sounds are also distinguished by their voicing. Everysounds can be classified as either voiced or voiceless. Voiceless sounds are produced when yourvocal chords are not vibrating. If you put your finger over your throat when you are producinga voiceless sound, you should not be able to feel your larynx vibrating. Voiced sounds areproduced when your vocal chords are vibrating. If you put your fingers over your throat whenyou are producing a voiced sound, you will likely feel your larynx vibrating.

To summarize the discussion so far, every consonant has the following three physical proper-ties:

• Voicing: Whether or not your vocal chords are vibrating.

• Place of articulation: How your lips and tongue are positioned.

• Manner of articulation: How the air is moving and how much air escapes.

These three properties are the physical correlates of the three features that we use to identifyconsonant sounds.

• Voicing (voiced or voiceless)

16

• Place of articulation (bilabial, labiodental, interdental, alveolar, palatal, velar, or glottal)

• Manner of articulation (stop, fricative, affricate, nasal, lateral, approximant, or glide)

We describe consonants (and vowels, as we will see) by their features. For example, [p] is avoiceless bilabial stop and [n] is a voiced alveolar nasal.

2.1.5 Keeping letters and sounds distinct

To maintain this distinction between letters and sounds, we write them differently. When we usewords as linguistic examples and refer to them by their spelling, we italicize them (e.g. cat), ashas been done in these lecture notes so far, or we put them in quotations (e.g. ‘cat’).

As you read in the previous chapter of these notes, and as should now be completely clearfrom the preceding discussion, sounds are completely different from letters. While letters areconventionalized symbols, sounds are physical units distinguishable on the basis of identifiablearticulatory features. To point to just one example, a word like chorus has six letters but only fivesounds (ch-o-r-u-s). Moreover, the sound spelled with a ch in this word can also be spelled witha c, as in cat, or with a c and a k as in sack. (Say each of these words to yourself, concentrating onthe sounds and not the spelling, to convince yourself that the sound is the same in each word.)

To indicate that we are talking about the sounds in a word, we do two things differently.First, we write the words using the International Phonetic Alphabet (IPA), which uses oneunique symbol for every sound. This means that a sound like sh, which is spelled with twoletters, can be written with just one symbol, [S]. The IPA keeps us focused on sounds, not letters.In the word wash, the single sound represented by sh corresponds to two letters. However, sh canalso stand for two sounds, as in the word misheard. We use the symbol S for the sound we hearin wash and the combination of the two symbols [s] and [h]—i.e. [sh]—for the sounds we hear inmisheard.

As you will have noticed in the previous paragraph, the other notational convention youshould follow when writing sounds (as opposed to letters) is to enclose the symbols representingthe sounds in brackets. This is helpful because in some cases the IPA symbols look like regularEnglish letters, as we see in the case of [sh]. Putting the symbols in brackets makes it very clearthat you are using them as sound symbols, not letters.

A note to remember in studying and doing assingments

• When spelling words, write them like this (or ‘this’). Since you will likely be handwriting your assignments, using quotes will likely be more convenient.

• When writing sounds, write them like [Tis] (this), using IPA symbols enclosed inbrackets.

Below are a chart of the IPA symbols for English consonants. We will use these symbolsto transcribe words. Transcribing is the technical term for writing the sounds that words arecomprised of. We will probably be adding a few IPA symbols to our inventory when we look atother languages, but the chart in (145) gives us a good start.

17

(14) IPA symbols for English consonantsBilabial Labiodental Dental Alveolar Palatal Velar

Stop p b t d k gNasal m nFricative f v T D s z S ZAffricate tS dZApproximant (w) r j(=y) wLateral l

The IPA chart is organized according to the features explained above. As you can see, themanners of articulation form the vertical axis of the chart, while the places of articulation arearrayed along the horizontal axis. Finally, within each column, the voiced sounds are located onthe right, while the voiceless sounds are on the left.

This organization makes it easy to find the symbol for the combination of features associatedwith each sound, and to find the symbols for sounds that you have never heard or written before,if you are able to identify the relevant features. We have seen, for example, that [f] is a voicelesslabiodental fricative; therefore, it is located in the cell where the labiodental column intersectswith the fricative row. In (15) there is a list of the consonant sounds and an example word foreach one. Each example word is transcribed using the IPA symbols. Don’t be alarmed for nowif you don’t recognize the symbols representing the vowel sounds for now; we will move on toconsider vowels just below.)

(15)

p pat [paet] T this [TIs] Ã judge [Ã2Ã]b bat [bæt] D the [D@] m mat [mat]t tap [tæp] s sat [sæt] n gnat [næt]d pad [pæd] z zip [zIp] N sing [sIN]k cat [kæt] S wash [wAS] l last [læst]g get [gEt] Z garage [g2rAZ] r rat [ræt]f fat [fæt] h hat [hæt] w what [w2t]v vat [væt] Ù match [mæÙ] y yet [jEt]

A final note about sounds versus spelling: A first glance over this chart shows why it is useful todistinguish words spelled with letters (when we are referring to them as examples) and wordstranscribed in IPA symbols. In most cases, such as (16a) below, the two will look quite different,but in some cases, such as (16b), they will look quite similar. It might be tempting, given thesimilarity of form between some English letters and some symbols of the IPA, to simply startspelling, but remember that this overlap is purely accidental. To keep this distinction clear, it ishelpful to recall that the IPA is used to transcribe English words as well as words in all humanlanguages, many of which are written in an entirely different script [Recall the Persian example,written in Arabic script, we saw in class.]

(16) a. sit [sIt]b. judge [Ã2Ã]

2.2 Vowels

Now that we’ve looked at consonants, let’s move on to vowels.

18

(17) Table 2. IPA symbols for English vowelsFront Central Back

High itense utense,roundIlax Ulax,round

Mid e tense 2lax otense,roundElax Olax,round

Low æ lax Atense

As we saw above when we were considering the nature of the airflow involved in the produc-tion of so-called ‘semi-vowels’, vowels are more complicated than consonants. There are fewerof them, but it’s a bit harder to detect just how your mouth is forming the sound. The chart in(55) contains an example word for each vowel sound listed in (17).

(18)

i beet [bit] u boot [but]I bit [bIt] u book [bUk]e bait [bet] but [b2t] o boat [bot]E bet [bEt] O bought [bOt]æ bat [bæt] A father [fAT2r]

2.2.1 The features of vowels

Consonants and vowels are different in important ways. First, as we already discussed a bitabove, there is always a continuous flow of air through your vocal tract when you are producinga vowel sound, so we don’t have vowels that are stops or affricates. That is why the consonantsclassified as laterals, glides, and rhotics (the approximants) are sometimes called semi-vowels.Second, vowels are always voiced (at least in English; Japanese has voiceless vowels, which wewill see). Third, there is more variation in the way individual speakers of a language producevowels than in the way individual speakers produce consonants.

Each vowel has four features, as shown in the chart in (17). Each of these features is explainedbelow.

Height (high, mid, or low): This refers to how high in your mouth your tongue is positioned asthe air flows over it and out of your mouth. Try making the sound [i]. Now make the sound [æ].You probably had to lower your jaw a bit and drop your tongue in order to make the [æ] sound.

Backness (front, central, or back): This also refers to your tongue position, but not to how highor low it is. This feature describes how forward in your mouth your tongue is. Make the sound[i] again. Now make the sound [u]. You probably had to move your tongue toward the back ofyour mouth in order to make the [u] sound.

Rounding (rounded or unrounded): This refers to the shape of your lips. Let’s use [i] and [u]again. When you make [i], your lips are open and rather pulled back. However, when you make[u] your lips are rounded. Back vowels are also rounded, except for [A].

Tenseness (tense or lax): This refers to how much the air is constricted when you are producingthe vowel sound. For instance, make the [I] sound. Now make [i]. The back of your throatprobably feels tight when you make the [i] sound and not so tight when you make the [I] sound.

19

That’s because the back of your tongue is closer to the very back of your mouth when you make[i]. Since the airflow is more constricted, this sound is tense. When you make [I] there is morespace between the back of your tongue and the back of your mouth. Since the air flows morefreely, this sound is lax.

Just as we describe consonants by their features, we also describe vowels by their features. Forexample, [E] is a mid front lax unrounded vowel and [u] is high back tense rounded vowel.

2.2.2 Diphthongs

In American English, vowels are often produced by saying two (vowel) sounds consecutively veryquickly, making an articulatory movement form one vowel to another within a single syllable.Not all speakers produce the same diphthongs. For instance, Northern speakers sometimesproduce a diphthong in the place where Southern speakers would produce a monophthong(only one sound). Below are the English diphthongs that we will be concerned with.

(19)AI iceAU blouseoI boy

A note to remember in further studying on your own You will be responsible for learningthe IPA symbols for consonants and vowels, as well as their features. We will do somepractice in class, but it will require time spent outside of class to learn these symbols andtheir corresponding features. Making flashcards and using them to drill yourself is onegood strategy for doing so. Some of the features you will be able to figure out by makingthe sound. Others you will just have to memorize.

2.3 For further reading...

International Phonetic Association. 1999. A handbook of the International Phonetic Association.Cambridge: Cambridge University Press.

Ladefoged, Peter. 2001. A course in phonetics. USA: Heinle and Heinle.

Ladegofed, Peter. 2000. Vowels and consonants. Oxford: Oxford University Press. (Comes with aCD illustrating all of the sounds of the IPA, including sounds from many different langauges.)

20

Chapter 3

Phonology

3.1 Minimal pairs

We have seen that every consonant and vowel sound in English can be classified according to therelevant phonetic features (voice and place and manner of articulation in the case of consonants;height, frontness, tense/laxness, and rounding in the case of vowels). The individual sounds thatcorrespond to each feature combination (e.g. [p], [b], [æ) are called phones. In most of the caseswe have looked at so far, these different phones are also phonemes. A phoneme is the smallestunit of sound that that is relevant to distinguishing meaning in a human language.

Let us consider what is meant by the relevance of phonemes to distinguishing meaning in ahuman language. Clearly, individual phonemes do not themselves have meaning: hearing thesounds [p] or [b] in isolation does not convey a meaning to a speaker of English in the sameway that hearing the words paper or boy does, for example; in fact, it does not convey any suchmeaning at all. Now imagine, however, that you hear the two words pat and bat. Automatically,you understand the meaning of each of these words, and you also recognize, automatically,that the two words have radically different meanings. Structurally, however, these words areextremely similar; more precisely, they are structurally identical except for the first sound in eachone.

Pairs of words such as this one, which are phonologically identical except for a single sound,are called minimal pairs. Below are some further examples of minimal pairs in English:

(20) a. pat batb. boy soyc. rope robed. mean meal

Minimal pairs are very useful to phonologists because they provide a means of identifyingphonemes. The fact that, as we have just seen, changing [p] to [b] at the beginning of the wordpat results in a completely different word, means that p and b are phonemes in English, and wecan say that the difference between [p] and [b] is phonemic in English. It is important to specifythat the difference between these two sounds is phonemic in English, because phonemes must beidentified on a language specific basis. The fact that the difference between [p] and [b] results ina meaning difference in English does not necessarily mean that it results in a meaning distinction

21

in other languages. Similarly, not all phonetic differences are phonemically relevant in English.To see this exemplified, we need first to make a slight digression, in order to familiarize ourselveswith another phonetic property of consonants, particularly, of stop consonants—aspiration.

In phonetics, aspiration is the strong burst of air that accompanies the release of some obstru-ents. An aspirated stop is produced with a more forceful release of the air than a nonaspiratedstop. When you make an aspirated sound, it feels a bit as if you are trying to spit. To feel orsee the difference between aspirated and unaspirated sounds, try placing a hand or a lit candlein front of your mouth, and say the word tore and then the word store. One should either feela puff of air or see a flicker of the candle flame with tore that one does not get with store. (InEnglish, the [t] is aspirated in tore and unaspirated in store.) More generally, in English, aspiratedstops only occur in certain positions within a word. Word initial stops and stops at the beginningof stressed syllables are always voiced in English. The IPA symbol for indicating aspiration is asuperscript h immediately following the aspirated stop, e.g. [ph], [bh]. Thus, we transcribe theword tore as [thor] and the word store as [stor], respectively.

Now that we are familiar with aspiration, we can return to our discussion of minimal pairsand phonemically relevant differences. We can now see that aspiration is an example of a pho-netic difference that is not phonemic in English. In English, saying a word with an aspirated stopand saying a word with an unaspirated stop are simply two ways of saying the same word. Incontrast to the difference between [pæt] and [bæt], saying [pæt] instead of [phæt] does not resultin a different word in English, though pronouncing the word [pæt], with an unaspirated stop, islikely to make the speaker sound as if she has an accent.

However, in some languages, such as Hindi, [pæt] and [phæt] are perceived and understoodas completely different words. This is because in Hindi and other languages similar in thisrespect, the aspirated form of a stop is a separate phoneme and using an aspirated stop forms aword with a different meaning than the same word with an unaspirated stop. This shows us that[p] and [ph] are different phonemes in Hindi, but not in English. The difference between [p] and[ph] is, therefore, phonemic in Hindi but not in English.

It is important to note that, even if Hindi had a word pronounced [pæt] but happened not tohave a word pronounced [phæt], a Hindi speaker hearing [phæt] for the first time would assumethat this was a new word—like the name of a new brand of soap, for example—rather thanperceiving it as “just another way of saying [pæt]”,as English speakers would. The point of thisexample is to show that even if there isn’t an existing minimal pair demonstrating a phonemicdifference between sounds, the fact that these two forms are different possible words is enoughto demonstrate that the difference is phonemic. We can see this in English, where, for examplewe have a word bottom [bAt2m], but no corresponding word *pottom [pAt2m]. (The * here is usedto indicate that the word does not currently exist in English, not that it is an impossible word.)We know, though, that the difference between [b] and [p] is phonemic. We know this becausewe have minimal pairs for these two sounds in English. Pat [pæt] and bat [bæt] are completelydifferent words in English. Because English speakers know that the difference between [b] and[p] is phonemic, if we heard the word [pAt2m] for the first time, we’d assume that it was a newword, rather than another way of pronouncing bottom [bAt2m].

22

3.2 Allophones and complementary distribution

Sometimes a single phoneme can be pronounced more than one way. This happens a lot inEnglish. For example, say the words pat [pæt] and pan [pæn]. Both words have the same vowelphoneme [æ], but the vowels sound different. This is because the nasal [n] at the end of panmakes the vowel sound nasal, too. You should be able to feel your nose vibrating during thevowel in pan. We can note this explicitly in the transcriptions by writing [pan], where the [˜]indicates that the vowel has been nasalized.

We know that [æ] and [æ] aren’t different phonemes in English, for two reasons. First, ‘mis-pronouncing’ [a] as [æ] doesn’t make a new English word. If you heard pat mispronounced as[pat], where it contained the vowel in pan rather than the vowel normally pronounced in pat,it would sound a bit off, but you wouldn’t assume it was a completely different word than theintended pat. Second, there are no minimal pairs that differ only in that one word has [æ] wherethe other has [æ]. When a single phoneme can be pronounced in two or more ways, the var-ious possible pronunciations are called allophones of that phoneme. Allophones are variouspronunciations of a single phoneme.

Recap: phonemes versus allophones

• phoneme: a minimal meaningful unit of sound structure. The occurrence of aphoneme is independent of other factors.

• allophone: the multiple possible pronunciations of a single phoneme. The occur-rence of an allophone is predictable on the basis of the surrounding sounds.

Phonemic analysis involves looking at sets of words, in English or any other language, andtrying to figure out whether pairs of sounds are independent phonemes or allophones of asingle phoneme—that is, whether the difference between a given pair of sounds is phonemic orallophonic.

The best and easiest way to discover and/or prove that the two sounds are phonemes isto find a minimal pair in the data you are given. If there are no minimal pairs proving thatthe difference between the sounds is phonemic, you can instead look for the following kind ofevidence that the sounds are allophones. Allophones are predictable that is, when a phonemecan be pronounced in more than one way, these two pronunciations are not used randomly,wherever a speaker happens to feels like using one or the other. Instead, the occurrence of eachsound is predictable from the context in which the sound is appearing.

To see this more clearly, consider the following words in which [æ] and [æ] appear:

(21)

pad [pæd] man [mæn]hat [hæt] Sam [sæm]wrath [ræT] hang [heN]mass [mæs]

If you weren’t an English speaker, and this data were all you had to decide whether [æ] and[æ] were allophones, the best way to proceed in doing so would be figure out whether the twosounds are in complementary distribution—that is, whether [æ] always appears in a certain

23

context or contexts in words, while [æ] never appears in those contexts and always shows upin a different context. Since the pronunciation of allophones is predictable, finding this sort ofregular, predictable pattern indicates that two sounds are in fact allophones.

Complementary distribution: Allophones are in complementary distribution when theyappear predictably in different contexts that is, when the distribution of one allophonecomplements (i.e. is strictly different from) the distribution of the other allophone.

Returning now to the words above, we should look at the contexts where [æ] and [æ] appearand see if there is a way to predict which of those sounds gets used in which word. When youare attempting to determine the contexts in which sounds appear, it is usually easiest to start bylooking at the segments (the individual sounds) immediately before and after the sounds andsee whether they have anything in common. A helpful way to begin doing this is to make lists ofthe sounds that come before and after the sounds under investigation. In the case of the wordsin (21) above, we have the following lists:

(22) a. Before [æ]: [p], [s], [h], [m]b. Before [æ]: [s], [h], [m]c. After [æ]: [d], [t], [T], [s]d. After [æ]: [n], [m], [N]

The sounds before [æ]/[æ] do not predict which sound is pronounced both [æ] and [æ]can follow [m] and [h]. Therefore, you can’t use the presence of a preceding [m], for instance,to tell you whether you should use [æ] or [æ] in a given context. Either sound could appearimmediately following the sound [m].

However, the two vowels are followed by completely different sets of sounds, so one could usethe sound following the vowel to figure out which pronunciation to use. Looking more carefully,the sounds that follow [æ] are all nasals, and the sounds that follow [æ] are all not nasals. Wecan thus state the following generalization about the distribution of the two sounds, which alsoallows us to predict where they occur.

(23) [æ] always appears before non-nasal sounds; [æ] always appears before nasal sounds.

In many cases the the context determining the pronunciation of an allophone might be morecomplicated. It could be the preceding or following vowel or consonant, rather than simply thepreceding or following sound, for example. It could also have something to do with a sound’sposition at the beginning or end of a word, or in a stressed or unstressed syllable. Sometimes itmight be the preceding and following sounds. And sometimes, there might be more than onefeature on a neighboring sound that determines an allophone’s pronunciation. This method oflisting environments and looking for similarities should help you find all of these.

Possible contexts that can determine the pronunciation of an allophone:

• Following sound

• Preceding sound

• Preceding or following vowel

24

• Preceding/following consonant

• Beginning of a word

• End of a word

• In a stressed syllable

• In an unstressed syllable

It is very important, when you are listing the sounds that can come before or after the soundsyou are looking at, to remember to list word edges. Sometimes one allophone might appearat the beginning or end of a word and the other might appear everywhere else; if you don’tsomehow indicate “beginning of a word” in the list of sounds that occur before the sound inquestion or “end of the word” in the list of sounds that occur after it, you could easily missthis generalization. (The symbol # is commonly used to indicate word edges.) Following isa summary of the steps you can follow to figure out whether two sounds are phonemes orallophones.

Are [X] and [Y] phonemes or allophones? A summary of how to work it out:

Is there a minimal pair of words, differing only in that one has [X] where the other has [Y]?

1. If YES → [X] and [Y] are phonemes.

2. If NO → We don’t have enough information, so we need to look at the contexts in whicheach sound occurs.

Look at the locations in which [X] and [Y] appear in words (their contexts). Are [X] and [Y] incomplementary distribution? That is, does either sound have a consistent, predictable context,such that you could state a generalization along the lines of:

[X] is always before sounds with feature Z; [Y] is never before sounds with feature Zor[X] is always after sounds with feature Z; [Y] is never after sounds with feature Z

1. If YES → [X] and [Y] are allophones

2. If NO → [X] and [Y] are phonemes

Two important things to remember:

1. A phonemic difference between two sounds is a meaningful difference. Trading one sound for theother can make a new word.2. An allophonic difference between two sounds is a predictable difference. Specific properties of thecontext of the sound tell you which of the two sounds should appear there.

25

3.3 Basic phonemes, and phonetic motivation for allophonic changes

The previous section talked about two segments (such as [æ] and [æ]) being allophones in ageneral way. We should now be a bit more specific about the relationship among allophones. Wecan understand one or more allophone(s) of a given phoneme as having undergone a change,while one allophone has not. This means that there is one fundamental sound (which is thephoneme), which is pronounced as itself sometimes and other times, in particular contexts, turnsinto another sound. In the example above, we could say that while [æ] and [æ] are allophones of/æ/—that is, that /æ/ is the basic phoneme /æ/ changes to [æ] in a particular context, namely,before a nasal consonant. We can show the change that occurs using a phonological rule, as in(127).

(24) /æ/ → [æ] / [nasal]

Paraphrased into words, what (127) means is that /æ/ becomes [æ] when it appears before anasal sound. This is the general schema for phonological rules; the meaning of each part of thenotation is summarized below.

Guide to reading phonological rules

general form: X → Y / ( ) Z ( )

• The slanted brackets // indicate the phoneme, the basic sound, the sound that is not pre-dictable. The square brackets [ ] indicate the allophone that has undergone a change. (Weuse this notation to indicate a phoneme or an allophone outside of phonological rules aswell, as you will have noticed in the text above.)

• The arrow (→) means ‘becomes’.

• The slanted line (/) means ‘in the context of’.

• The blank line ( ) indicates where the sound in square brackets is pronounced. (Forexample, if it occurred following instead of preceding the nasal in the example above, itwould mean that the [æ] sound occurs following a nasal consonant.)

• We don’t need a rule to tell us when /æ/ is pronounced as [æ]. [æ] is the basic form andwill be pronounced in all other contexts except the one indicated in our rule.

Phonological rules are a simple way to describe the distribution of allophones. In addition,they allow us to capture the fact that allophonic variation is generally phonetically motivated.That is, most of the time, when there is a change in one of the allophones, the variation ispredictable because the sound appears near sounds that it has something in common with.

In the [æ]/[æ] example, [æ] appears before nasals, and it is itself a nasal vowel. You can tellthat [æ] has something in common with nasals physically, and thus phonetically, because yournose vibrates when you say [æ] in the same way it does when you say [m] or [n]. In this way, ifwe say that [æ] turns into [æ] before nasals, we are really saying that when [æ] is followed by anasal, it becomes phonetically more similar to that following nasal by being pronounced as thenasalized vowel sound [æ]. This is, in the simplest terms, a way in which your brain tells yourmouth what to do in order to make speaking easier.

26

Determining which is the basic phoneme among a set of allophones can be a little tricky, butthe following serves as a useful rule of thumb.

If [X] and [Y] are allophones, and

X appears in a very limited context (only before nasals, for example), while

Y appears in a more general context (such as before every other kind of sound, and at theend of words)

...then the sound with the general distribution (here, [Y]) is the basic phoneme, and

...the sound with the limited distribution (here, [X]) is an allophone of the other.

In general, the basic phoneme is the one that appears in a wider variety of contexts (i.e. thathas a wider distribution), and it turns into the altered form in a limited set of contexts.

3.4 Positions in syllables

Syllables are another structural unit of language. They are difficult to define without someformal terms, which will be described shortly. However, speakers do tend to have an intuitivesense for the location of syllable boundaries. As a first approximation at determining how manysyllables are in a word, you might try clapping or tapping your foot while you say a wordslowly; most people find that they are able to clap once per syllable. Try it with long words suchas de.reg.u.la.tion and pro.no.mi.na.li.za.tion (divided here into syllables with periods).

Once you have a sense for what syllables are, we can begin considering the parts comprisingthem. Words, as we have seen, are made up of vowel and consonant sounds, and syllables arearrangements of vowels and consonants. (Almost) all syllables have at least one vowel in themiddle. This vowel (or these vowels) is (are) called the nucleus. In a very simple word like cat[kæt], which consists of a single syllable, the nucleus is the vowel [æ].

If there are any consonants before the nucleus, these consonants are the onset of the syllable.If there are any consonants after the nucleus, these consonants are the coda of the syllable. Sometimes it is useful to be able to talk about the nucleus and coda together; this part of the syllableis called the rime (because words that rhyme have to have the same final nucleus and coda—forexample, cat and pat ([kæt] versus [pæt]))

The parts of a syllable

Nucleus: The vowel(s) in a syllable. All syllables have nuclei.e.g. cat [kæt], at [æt], see [si], eye [AI], craft [kræft], syllable [sI.l2.b2l]

Onset: Any consonants in a syllable that come before the nucleus. Not all syllables have onsets.e.g. cat [kæt], at [æt], see [si], eye [Ai], craft [kræft], syllable [sI.l2.b2l]

Coda: Any consonants in a syllable that come after the nucleus. Not all syllables have codas.e.g. cat [kæt], at [æt], see [si], eye [Ai], craft [kræft], syllable [sI.l2.b2l]

27

Rime: The nucleus and coda of a syllable. Because all syllables have nuclei, all syllables haverimes.e.g. cat [kæt], at [æt], see [si], eye [Ai], craft [kræft], syllable [sI.l2.b2l]

Finally, sometimes we will talk about complex syllable constituents. A nucleus, onset, orcoda is complex if it contains more than one phoneme. So in the example words above, eye hasa complex nucleus, because there are two vowels in the nucleus [AI]; craft has a complex onsetbecause there are two consonants in the onset [kræft] and also a complex coda because there aretwo consonants in the coda [kræft].

We will represent how syllables the internal structure of syllables using diagrams like thosebelow:

(25) a. σee%%

O

s

N

I

σee%%

O

l

N

2

σb

bb"

""O

b

N

2

C

l

b. σb

bb"

""O

kr

N

æ

C

ft

Here, O indicates Onset, N indicates Nucleus, and C indicates Coda. We connect each pho-netic symbol to the letter that indicates its position in the syllable. Then we connect each part ofthe syllable to the Greek small letter sigma, σ, which is the symbol we use for syllables. We willsee in class that languages vary with respect to what types of sounds they allow in the onset andcoda positions of syllables For instance, in some languages, syllables do not have codas and insome languages complex onsets (onsets that contain more than one sound) are not allowed. Wewill also see how a clustering of sounds can determine which sounds are part of the onset andwhich sounds are part of the coda.

28

Chapter 4

Morphology

4.1 Introduction: productivity and predictability

Linguists are always looking for regularity in language. A linguistic pattern is regular if it ex-tends reliably to a new utterance that the speaker has never said or heard before. We have alreadygained some experience of what it means to look for regularity from our study of phonology. Wefound that the environments of different allophones of a single phoneme were predictable withrespect to one another, and we attempted to write rules that would extend as widely to otherdata in the language as well. Now we are going to begin looking at productivity in the domainof word formation, or morphology, so we will concentrate on the distribution not of phonemesbut of morphemes, the minimal meaningful units of language.

What do the examples from English below tell us about the regularity with which the plural ofnouns are formed in this language?

(26)

Singular Pluralcat catsdog dogstable tableschair chairsstar starsmoon moonsidea ideas

As the chart indicates, the singular of a noun in English consists of just the noun itself, whilethe plural is formed by adding -s at the end of the noun.

This pattern is regular if it extends reliably to a new noun. Consider again the sentence weencountered in Chapter 1 of these notes:

(27) One wug and one wug make two .

Because you would most likely be correct in forming the plural as wugs, linguists wouldcall attaching -s to the end of the noun the regular pattern of plural formation in this language.Because regular patterns reliably extend to new forms like this, they are also called productivepatterns, allowing speakers, and linguists, to predict a new form.

29

Regularity and productivity are important; in addition to being pervasive properties of hu-man languages, they are also properties that make linguistic analysis possible. Once a linguistobserves a particular pattern, he or she assumes that the pattern holds of forms not yet encoun-tered, just like a child learning a language. We have made use of this assumption in doingphonological analysis, and we will make use of it throughout the course; in this chapter we willemploy it in our investigation of morphology.

4.1.1 Irregular forms

While regularity is what gets the study of morphology off the ground, not all English nounsform their plurals by adding -s at the end, as any native speaker knows. The following are someexamples of plural forms in English that are not formed through the addition of -s.

(28) a. ox:oxen

b. child:children, cactus:cacti, octupus:octopi

c. hoof:hooves, leaf:leaves

Unlike the regular nouns, whose plural forms can be predicted, it is not as easy to predictfrom the singular of an irregular noun what the plural would be, and the plural must instead tobe memorized. This difference between speakers’ ability to predict regular forms and their needto memorize irregular ones can even be observed in different locations of brain activity when thetwo kinds of forms are produced (Jaeger, et al. 1996).

The productivity of regular forms shows up in at least three other ways:

1. In early stages of acquiring English, children overgeneralize the regular form to irregularnouns and verbs, forming plurals such as *childs and *hoofs and past tenses such as falled, buyed,and sleeped, instead of the irregular verb forms fell, bought and slept. Like linguists, they assumelanguages are regular until they learn otherwise.

2. When an irregular noun is used as part of a proper name, its plural is formed regularly, e.g.when you have more than one leaf from a maple tree, you have maple leaves, but the members ofToronto’s hockey team are the Maple Leafs. Similarly, when a noun is used as a verb, its pasttense is formed regularly, e.g. when the noun ring is used as verb meaning to encircle, its pasttense is ringed not rang, unlike the verb ring meaning to make a sound.

3. There is some degree of productivity among the irregular forms. For example, if you knowthat the plurals of cactus and octupus are cacti and octopi, then you’re likely to predict that theplurals of rhinocerus and torus are rhinoceri and tori, and these predictions are correct. Similarly,many speakers use brang rather than brought as the past tense of bring, modeling sing:sang,ring:rang, and drink:drank.

Question to consider: Why do you think we would not want a grammar in which all of theplural forms of the regular nouns were memorized as well as the irregular ones?

30

4.2 How to do morphological analysis

In general, when you are asked to do any sort of linguistic analysis, you will be given a set ofdata—words or sentences from some language that generally isn’t, but occasionally is, English—and asked to find patterns in it. Here is a set of data from Cree, an Algonquian language spokenin Canada. On the basis of this data set, for example, you might be asked to determine how tosay “my (whatever)”.

(29)

Cree EnglishÙi:ma:n canoeniÙi:ma:n my canoeso:niya moneyniso:niya my moneywiya:S meatniwiya:S my meat

Look for pairs of words with small, relevant differences. For example, if you’re trying tofigure out how to say that something is “my (thing)”, you can look for two words that differ onlyin that one is “my (thing)” and the other is just “(thing)”. The first pair of words is like this:the first is a word for ‘canoe’; the second is a word for ‘my canoe’. It seems likely, then, that thedifference between the two Cree words corresponds to the difference between the two Englishphrases. That is, the difference between these two words:

1. Ùi:ma:n

2. niÙi:ma:n

...is just the ni- at the beginning of the second word. So you might think, very sensibly,that adding ni- to the beginning of a word is how you make it “my (thing)”. You can test thishypothesis on other pairs of words whose English translations differ only in the desired way, likethese:

1. so:niya ‘money’

2. niso:niya ‘my money’

Again, the only difference between the two Cree words is the ni- at the beginning, so again,it looks like adding ni- to something makes it “my (thing)”. This is confirmed by looking at thelast pair of words, which show the same pattern:

1. wiya ‘meat’

2. niwiya ‘my meat’

Let’s summarize the steps to go through in doing morphological analysis when you don’tknow anything about the language:

• Find a pair of words whose English translations differ only in a single way.

• Find the corresponding difference in the non-English words. Often some letters will beadded, or the word will be changed in some other systematic way. Making this change inthe non-English word therefore produces the relevant change in the English meaning.

31

• Check your theory: find another pair of foreign words whose English translations alsodiffer only in this relevant way, and make sure this pair of foreign words changes in thesame way as the last.

We will encounter kinds of morphology problems that are more complex than this, but thisbasic method of looking for forms with minimal differences and figuring out how to describethat difference is always a good approach with which to begin.

4.2.1 Morphemes

If your friends asked you what the smallest meaningful element in your language is, your firstanswer might be “words, obviously.” This answer seems obvious to us because we ordinarilythink of utterances as being composed by stringing words together. But let’s now go back andlist all the words in the first sentence of this paragraph and see whether they are indeed thesmallest meaningful elements:

(30)

If youryour languagefriends isasked youryou firstwhat answerthe mightsmallest bemeaningful wordselement obviouslyin

Why do you think some of the words in these two lists are in italics? The answer is that eachword in italics is composed of smaller meaningful pieces, called morphemes:

(31)

Words Morphemesfriends = friend + sasked = ask + edsmallest = small + estmeaningful = mean + ing + fulwords = word + sobviously = obvious + ly

4.2.2 Some semantics: compositionality

As English speakers we know that the meaning of the word smallest is the combination of themeanings of small—the state of being of of a lesser degree along some dimension of size inrelation to something else—and -est—being the most extreme value of some scale—, and soon for the other words in 31). How do we know that smallest can be broken down into themorphemes small and -est in this way? You might answer: because that’s what smallest means:the state of being the most limited in dimension. But from a linguist’s point of view, there is

32

a much better answer. It is that small and -est must be morphemes because each can combinetogether with other morphemes to make other words, e.g. smaller (= small + er) and largest (=large + est). (These examples also show us that -er and large are morphemes, too.) This is a betteranswer because we can discover what morphemes are using this method without needing toknow much about what the morphemes small and est or their combination “smallest” mean.Themost we need to know about the meaning of smallest is that part of its meaning is shared with partof the meaning of smaller and the other part of its meaning is shared with part of the meaning oflargest. Smallest is also partly different in meaning from both smaller and largest. Both the sharedand unshared meanings among these three words can be specified precisely by breaking themdown into their constituent morphemes. Specifically, we would hypothesize that the similarityin meaning between smallest and smaller is conveyed by the part they both share, small, and thatthe difference by the parts that differ, i.e. -est vs -er. In the same way, we would hypothesize thatthe similarity in meaning between smallest and largest is conveyed by their shared part, -est, andthe difference in meaning is conveyed by their nonshared parts, small and large.

We have been making an assumption above about the meanings of morphemes: that theystay the same wherever they occur. Thus, the differences in meaning between small, smaller.and smallest are the same as the differences in meaning between large, largerer. and largest andthe difference in meaning between smallest vs largest is the same as the difference in meaningbetween small versus large and smaller versus larger. A consequence of these assumptions isthat the meaning of a word is made up of the meanings of the morphemes it contains. Thisis a property called compositionality. (We’ll also see soon that compositionality applies to theanalysis of sentences as well, where the meaning of a sentence is made up of the meaning of itsparts.)

Compositionality is a hypothesis about language, but we will use it as an assumption for themost part. Since it is a hypothesis, compositionality is subject to verification. The meanings ofsome words, for instance, are different from the meanings of the morphemes that they contain.In essence, these words are non-compositional. This commonly happens in what are calledcompounds, such as blackboard or hotdog, which are neither a black board nor a hot dog, respectively.The process of compounding the morphemes black and board or hot and dog results in a specializedmeaning.

4.2.3 Types of morphemes

Morpheme A morpheme is the smallest unit of meaning in language. In contrast to phonemes,which are the smallest pieces of words that affect meaning, morphemes are the smallest pieces ofwords that in themselves contribute meaning. The word sightings, for example, has 3 morphemesin it: sight-ing-s.

To break a word into morphemes, try starting at the beginning of the word and seeing howfar into the word you need to go to find a sub-part of the word that has some meaning. Forexample, in the word unloved, the first two letters un- are independently meaningful in a waythat just the first letter, u-, is not; un- means something like ‘not (whatever)’, and changes themeaning of the word it attaches to in a predictable way; sub-parts of un-, like u- or -n-, don’t havethis property. This means that un- is a morpheme.

Once you’ve found the first morpheme, ask yourself whether there is another meaningfulsub-part of the word after that first morpheme. Again, -love- is independently meaningful; so isthe last part of the word, -d. So unloved has three morphemes: un-love-d. Some words just have

33

one morpheme, of course you can’t break down the word love into any meaningful sub-parts,for instance.

(32) Exercise Break the following words down into their morphemes:rewrite, delight, fewest, impossibility, refastening, deconstruct, review

Did you experience any difficulties in doing this exercise? What is a conclusion you can drawabout the nature of morphemes from these difficulties?

4.2.4 Key Morphological Concepts

We define different kinds of morphemes based on various properties of those morphemes, in-cluding in particular where they occur in words. Below is an explanation of some key terms andconcepts:

Key Morphological Concepts:

• All morphemes are either free or bound.

Free - A free morpheme is one that can stand on its own—that is, it forms an entire word in itsown right.Examples: the, cat, run, pretty, trapezoid

Free morphemes may appear with bound morphemes attached to them. Crucially, though, theydon’t need to have other morphemes attached to be well-formed.

Bound - A bound morpheme cannot stand on its own. Rather, it must be attached to a freemorpheme whenever you say it.Examples: re-, un-, -est, -er

For instance, we would never say the prefix re- all by itself. It doesn’t mean anything inisolation. It has to be attached to a verb.

• Some morphemes are roots; others are affixes; and others are stems.

Root - The primary piece of meaning in a word, to which affixes can be added.In English, a rootis often a word itself, a free morpheme.Examples: cat, pretty, magnet

For example, in the word remagnetize the root is magnet, and the other morphemes change itsmeaning.

Affix - A morpheme which attaches to roots (or stems), changing their meaning in regular ways.Examples: re-, un-, -est, -er, ing, -s

Affixes are bound morphemes. Can you explain why?

Stem - A form to which an affix is attached. This can be a root alone, or a root that already hasone or more affixes attached but that can still take more.Examples: drink, swim (can take e.g. -er); drinker, swimmer (can take e.g. s)

Question: What’s the difference between a root and a stem? Can a morpheme be both?

34

Types of affixes:Affixes are generally either prefixes or suffixes, but there are others which we will find are morecommon in languages other than English.

Prefix - An affix that goes before a stem.Examples: re-, un- (re-read, un-loved)

Suffix - An affix that goes after a stem.Examples: -est, -er, -s (quick-est, quick-er, read-s, book-s)

Infix - An affix that appears inside a rootExamples: -fuckin’- (fan-fuckin’-tastic) -bloody- (abso-bloody-lutely)

English has only a handful of infixes, but they are more common in other languages. It isimportant to note that the parts on either side of the infix are not morphemes themselves (e.g.fan and abso are not morphemes in these words). The infix literally goes inside the root,breaking up the morpheme.

Circumfix - An affix that surrounds a stem. It is broken into two parts, usually one part is aprefix and the other part is a suffix.Example: In Tuwali Ifugao, spoken in the Philippines, the circumfix ka–an turns a verb into anoun, and surrounds the root/stem.

1. baddang ‘help’ verb root/stem

2. ka-baddang-an ‘helpfulness’

Compound: A complex word formed by combining two or more individual words from variousparts of speech (for part of speech, see below).Examples: fire engine fire + engine Noun + Noun bluebird blue + bird Adjective + Noun

As noted above, for some compounds, the meaning is not derived from the meanings of theindividual words.

Allomorph: Allomorphs are versions of the same morpheme that change slightly depending ontheir environment (preceding or following sounds). An example is the English indefinite article,which has two allomorphs, a and an. Another is the different realizations of the pluralmorpheme as [s] or [z]. Recall that we now have a way of capturing formally when each formoccurs.

Question: Can you predict when [a] and [an] are used?

4.2.5 Morphological rules

Just as was the case with phonological rules, the point of writing a morphological rule is todescribe what is going on morphologically in such a way that someone could use your rule tobuild new words. The productive nature of language allows us to predict new forms. A goodtest for whether your rule is right is to try to use it and make sure it gives you the right result.

Morphological rules can be statements that tell you the following three (or four) things:

35

1. What kind of morphological category you’re expressing.

2. What change takes place in the stem to express this category.

3. Where in the stem this change takes place.

4. Special conditions, if any, on this change (e.g. it might only occur in certain circumstances,on certain kinds of words, etc.).

We can look at a simple example rule that makes the English plural form of a noun (i.e. thattakes the root cat and adds an -s to the end to make cats). If we were to explain the rule usingplain English, it would look like this:

(33) To make the plural form of a noun, add s to the end of the noun.

We can break down this rule to show very explicitly which parts of it are doing which of thefour necessary things, like this:

1. To make the plural form of a noun,

2. add s

3. to the end of the noun.

Note that in this case (well, for the purposes of our discussion, at least), the plural is alwaysformed by adding s, so we don’t need any special conditions that is, no part (iv) above. Wecould also show all this with a rule using the following notation.

(34) Plural Noun → Root Noun + -s

In (34), → means “consists of”. So, in plain English, the notation in (34) says, “A plural nounconsists of a root noun and s”.

You’ll write rules based on data; here, the data is the pair of words cat and cats. You canthen take the rule and the data and make sure the rule produces (or predicts) the data. Here, forexample, you can take the singular form cat and apply the rule ‘add s to the end of the noun’ toit. This produces cats, as it should. You should also check the rule to see if it works with otherforms, such as dog/dogs, and it does.

Writing morphological rules and making generalizations

Look at the data below from Michoacan Aztec.

(35) Michoacan Aztec1.nokali ‘my house’ 9. mopelomes ‘your dogs’2. nokalimes ‘my houses’ 10. ipelo ‘his dog’3. mokali ‘your house’ 11. pelo ‘dog’4. ikali ‘his house’ 12. nokwahmili ‘my cornfield’5. kali ‘house’ 13.mokwahmili ‘your cornfield’6. kalimes ‘houses’ 14. ikwahmili ‘his cornfield’7. nopelo ‘my dog’ 15. ikwahmilimes ‘his cornfields’8. mopelo your dog’ 16. kwahmili ‘cornfield’

36

Now let’s write a morphological rule for Michoacan Aztec words from the list above. We aregoing to try out various ways of writing the rule, with the aim of ending up with the best wayto describe the words in the data.

We could do what we did for the English plural, where we said a word was made up of aroot noun plus a suffix. In the case of Michoacan Aztec, we could write rules like these:

(36) a. Word → no + Nounb. Word → i + Nounc. Word → Noun + mesd. ...e. ... (and so on for each word)

But does this rule tell us anything besides how you make the words with the morphemes no-,i-, and -mes in them? No! All these rules are doing is listing the possible forms with thosemorphemes in them. What we want to do is make a more general statement about how wordsare built in this language.

Question: What do we mean by ‘listing’ things? Why is a list like the one above bad in linguis-tics?

The first thing we do is categorize all the morphemes and the morphological class they belongto. This is a good use of a list because we are grouping morphemes into categories. This isinformation we cannot predict about Michoacan Aztec. The name in linguistics for the part ofa language that is not predictable but must be listed is the lexicon. The lexicon is like a largemental dictionary, but containing more detailed information.

(37) Morphemes and morphological classes (information listed in lexicon):a. Possessors: no, mo, ib. Plural: mesc. Root Nouns: kali, kwahmili, pelo

Question: Why can’t we predict this information about the language?

Now that we have this information listed, we can write a much more general rule.

(38) Morphological rule, Version 2Word → Possessor + Root + Plural

Something is wrong with this rule, however. Not all words follow this pattern. If you lookback at the data in (35), you’ll notice that some nouns don’t have a possessor and some nounsaren’t plural.

(39) Morphological rule, Version 2Word → (Possessor) + Root + (Plural)

The parentheses indicate that morphemes are optional: the word does not need to have eithera possessor or a plural. Version 3 can account for all the data given and is a lot simpler thanVersion 1, because it makes a generalization about the data.

37

4.2.6 English prefixes and suffixes

So far our morphological rules have been pretty simple, adding an affix like plural or possessorto a noun to make a plural or possessed noun. Some affixes can actually change the part ofspeech of the newly formed word, as the suffix -able does in the following example.

(40) -able V + able = Adj Eg. read + able = readable

Moreover, when we look at a lot of English affixes we discover that we have to be very preciseabout what words they can affix to. For instance, -able can only be attached to verbs. In example(63) below, the words in the left hand column below can all take -able because they are verbs rootsbut the words on the right cannot.

(41)

readable *redabledesirable *roosterablefixable *linguisticsablereformable *morphemeabledownloadable *shoeable

What’s more, the suffix -able changes the part of the speech of the verb it attaches to, turningit into an adjective. We know the resulting word is an adjective because it sits in the same spotthat adjectives do; for example, a very readable novel.

Below is a list of some affixes in English. The table tells you what the affix means, what itattaches to (that is, what part of speech it attaches to) and what part of speech it forms onceattached. Note that this is not a complete list of English affixes. Only included are some of themore common ones.

38

Affix Meaning Examples Attaches to a... Forming a...ex- former... ex-president, ex-con N Ndis- not... dishonest, disloyal, dis-

satisfiedAdj Adj

in- not... incompetent, incom-plete, intolerable

Adj Adj

mis- ...in a wrong manner mistake, misunder-stand, misspell

V V

re- ...again rework, rethink, reeval-uate, redo

V V

un- not... unhappy, untrue, un-sure, unconscious

Adj Adj

un- do the opposite of... untie, unwrap, un-cover, undo, unfold

V V

-able able to be ...ed lovable, fixable, break-able, washable

V Adj

-(i)al pertaining to... national, musical, pres-idential

N Adj

-ate make... activate Adj V-ation act of ...ing relaxation, meditation,

realizationV N

-dom state of being... wisdom, freedom,boredom

Adj N

-er one who ...s baker, teacher, owner,wanderer

V N

-ful full of... graceful, joyful, play-ful,hopeful

N Adj

-hood state of being a... sisterhood, childhood,neighborhood

N N

-ion act or result of ...ing protection, compensa-tion, reflection

V N

-ish like a... boyish, childish, fool-ish, sheepish

N Adj

-ity the quality of being... sanity, activity, passiv-ity, masculinity

Adj N

-ive tending to... assertive, persuasive,reflective

V Adj

-less without... penniless, priceless,hopeless

N Adj

-ly like a... friendly, womanly,manly, cowardly

N Adj

-ly in a ... manner slowly, happily, hur-riedly, foolishly

Adj Adv

-ment act or result of ...ing adjournment, govern-ment, movement

V N

39

-ness quality of being... happiness, firnmess,kindness

Adj N

-ous characterized by... famous, poisonous,rancorous

N Adj

-ship state of being (a) ... championship, kinship,governorship

N N

-y ...like mealy, pulpy, mousy,icy, fruity, fiery

N Adj

You will notice that even though some affixes have the same form, they attach to differentgrammatical categories. For example, one prefix un- attaches to an adjective to form a newadjective and another prefix un- attaches to a verb to form a new verb.

4.2.7 Drawing morphology trees

In addition to rules, we often use tree drawings in linguistics to represent the internal structureof words, sentences, etc. Trees are diagrams that look sort of like mobiles, where all the partsof e.g. the word hang off the bottom, and they’re grouped together two at a time until you’vejoined together all the parts of the word.

A tree for a word containing two morphemes, like cat-s, is really simple:

(42) Nll,,

N

cat

Suf

s

This shows that the noun cats has two pieces: the root cat (which is a noun) and the suffix-s. Cat is underneath an N; this shows that cat is a noun. -s isn’t under a part of speech labelbecause it doesn’t have a part of speech on its own. Here is it under Suf for suffix. The top Nin the tree—representing the whole word cats—is made by joining the N above cat and the -s;there’s an N where these two join together because the word they make—cats—is still a noun.

Trees are more illuminating when we’re combining more than two morphemes. If you’rebuilding a word like untraditional, with three morphemes, we need to combine them two at atime—first you make traditional, and then you add the prefix un- to it. Doing it the other wayaround would make the non-word untradition first, and that would be a problem. At every stageof the morphological tree, the word you’ve formed has to be an actual word. We can show theorder that morphemes combine in a tree, like this:

(43) Adjaaaa

!!!!Pref

un

Adjb

b"

"N

tradition

Suf

-al

Again, this tree has the root tradition under a part of speech label N. Tradition and the suffix-al join to form an adjective, denoted by the Adj at the top of the lines connecting N and -al. The

40

new adjective traditional combines with the prefix un- to form yet another adjective, untraditional.There are some important things to remember when you are drawing morphology trees:

• Roots are always directly below a part of speech label.

• All words created by adding a prefix or suffix to a root (or stem) are labeled with their partof speech.

• Affixes are added to stems one at a time. That is, no part of speech label has more than twolines below it (we call these lines branches).

The incorrect ordering of the affix attachment is shown in (44).

(44) * Adjaaaa

!!!!Noun?

bbb

"""

Pref

un-

N

tradition

Suf

-al

In addition to the fact that “untradition” is not a word, the prefix un- does not attach to nouns;it can only attach to adjectives or verbs. Of course, there’s the famous “Uncola” add campaign.This term stuck precisely because we don’t expect un- to attach to nouns.

4.2.8 Inflection and Derivation

Affixation is a morphological process whereby an affix (a bound morpheme) is combined witha root or a stem to produce a complex word different from the underlying root or stem. Thedifference may be in its phonological form, grammatical function, or semantic meaning. Forexample, when the noun winter combines with the suffix -ize, the result is the verb winterize whichhas a different phonological form, grammatical function, and meaning from the root winter.

(45)winter winterize

Part of Speech Noun VerbMeaning The coldest season of the year’ ‘To make ready for the winter’

Compounding is a morphological process whereby two or more words are combined to forma new word. The new word necessarily contains more than one root. A compound word obtainsits function (its use in a sentence) from one of the words that make it up. For example, thecompound ice-cold is made up of the noun ice combined with the adjective cold and it functionsas an adjective.

The morphological operations used in word formation are traditionally divided into twotypes: inflection and derivation. In other words, morphological processes are distinguished asbeing either inflectional or derivational in nature. What follows is a description of what it meansfor a morphological operation to be inflectional or derivational.

41

Inflection

A morphological process is inflectional if it marks a word so that the form of the word reflects itsuse in a sentence. The range and degree of inflection is language specific. There are languageswith little or no inflectional morphology and others with rich inflectional morphology. Englishhas very little. Consider the free morpheme play in English, which is a verb (V) (we’re notconcerned here with the noun play). It has several forms depending on its use in a sentence.

(46)

Forms Examples Category Grammatical Informationplay John wants to play the piano. V Infinitiveplays John plays the piano quite well. V Third person present tenseplayed John played the piano when he was younger. V Past Tenseplaying John is constantly playing the piano. V Progressive Tense

These forms each reflect some grammatical information about the lexical item play, in particu-lar, its tense. The idea is that in each sentence in (70), only one lexical item is being used, namelyplay. However, in each sentence, play surfaces with a different morphological form to signal itsgrammatical function in that sentence. However, it keeps its original meaning and category. Thedifferent morphological forms of play are called its inflected forms.

In inflectional affixation, the grammatical category of the new form is always the same as thegrammatical category of the stem. So, when we add the past tense morpheme -ed to a verb, thenew form is still a verb.

Because the processes of adding the types of suffixes shown in (70) only serve to distinguishthe forms in a paradigm, they are called inflectional processes or inflectional morphology (orsimply inflectional). The words play, plays, played, and playing are clearly different words but theyare all forms of the same verb play. The words in (70) are related to play in a way that the wordsin (71) are not.

(47)playable V → A This tune is quite playable by a beginner.player V → N The piano player dazzled us with her performance.

The words in (71) are also derived from play using affixation but these words are all distinct incategory and meaning and consequently they constitute different lexical items. The verb becomesan adjective and a noun, respectively. The examples in (71) illustrate derivational affixation,which is discussed more below.

As a final example of an inflectional paradigm, consider the Russian lexical item kosk- ‘cat’.The words in (72) are the inflectional paradigm for kosk- ‘cat’ in the singular.

(48)

koska ‘cat’ the form used when ‘cat’ is the subject of the sentencekoski ‘of a cat’ the possessive formkoske ‘to/for a cat’ the form used to mark that the ‘cat’ is getting somethingkosku ‘cat’ the form used when ‘cat’ is the direct object of the sentencekoskoj ‘by a cat’ the form used to mark that the ‘cat’ is the agent

The familiar English inflections include the plural on nouns and the tense on verbs. In manylanguages, including Russian, nouns are inflected depending on their grammatical role in asentence; for instance, they are inflected one way if if they are the subject of the sentence, andanother way if they are the object of the sentence. Inflectional affixes do not really change the

42

meaning of the word to which they attach, so the new word that is formed is always in the samegrammatical category as the root.

Derivation

Derivational morphological processes are primarily about forming new lexical items. Forming anew word may entail a change in category or a change in meaning or both. We have seen thatsome affixation is inflectional (70) while some is derivational (71). Let’s take a look at some moreexamples of derivational affixation.

(49)

Root Root’s grammatical category Affix New Word New word’s grammatical categorywash verb -able washable adjectivechild noun -hood childhood nounreflect verb -ion reflection nounkind adjective -ness kindness nounevaluate verb re- reevaluate verb

Unlike inflectional affixes, which do not change the category of the words, derivational affixescan change the category. For instance, we see that when -able attaches to a verb, it forms anadjective. Derivational affixes do not necessarily change the category, though. For instance, wesee that when -hood attaches to a noun, it forms another noun.

Inflection is generally more productive than derivation. For instance, almost all English verbscan be inflected with the progressive affix, -ing, and almost all (countable) nouns have a pluralform. However, not all nouns can take the derivational affix -hood; nor can all verbs take the affix-able. We can see this from the ungrammatical forms illustrated in (84).

(50) *evidencehood, *concernhood, *goable, *sleepable

An interesting fact is that derivational affixation has to happen before inflectional affixation.Consider the word neighborhoods. Why is it that we can’t say *neighborshood to mean the samething? This is because -hood is a derivational affix and -s is an inflectional affix. In the (ungram-matical) word neighborshood, the inflectional affix has attached to the root before the derivationalone, and in general, this ordering of affix attachment is banned throughout the world’s languages.

To sum up, we’ve seen that languages combine small units of meaning —morphemes—in order to form larger units of meaning. We’ve learned that morphemes fall into differentcategories and that we can model how these categories combine. Additionally, we’ve seen thatthere are a set of guidelines and restrictions that govern how morphemes combine.

Suggestions for further reading

Jaeger, Jeri J., Alan H. Lockwood, David L. Kemmerer, Robert D. Van Valin, Brian W. Murphy,and Hanif G. Khalak. 1996. A positron emission tomographic study of regular and irregularverb morphology in English. Language. 72.451-497.

Pinker, Steven, and Alan Prince. 1991. Regular and irregular morphology and the psychologicalstatus of rules of grammar. Berkeley Linguistics Society 17.230-251.

43

44

Chapter 5

Syntax

5.1 Structures: the hidden organization of language

Syntax is concerned with the ways in which words are assembled into phrases, and in whichphrases are assembled into sentences. Phrases, which exist above and beyond the words thatmake them up, are part of an invisible layer of linguistic structure. This structure is “hidden”, ofcourse, because in speech or in writing we just “line up” words one after the other.

A very basic way of describing sentences would be to give an ordering of the parts of speechin that sentence, as follows.

(51) I watched the movie about snakes on a plane in August.

(52) Sentence =

NounI

++

Verbwatched

++

Detthe

++

Nounmovie

++

Pabout

++

Nounsnakes

++

Pon

++

Deta

++

Nounplane

++

Pin

++

NounAugust

But we are going to see that this is not enough to capture speakers’ knowledge of sentences.Instead, when words are put together with other words into sentences there is a structure (hi-erarchical relationship) between them. Again, this structure is “hidden” from us, but we haveevidence that it is there. Sentences are made up of words, but we can show that there is anotherlevel of organization in between—that is, that sentences are made up of phrases, and phrases aremade of words. By the end of this section you will see that what we really need to do is representthe sentence above with a tree, as in the following diagram.1

1This sentence is in fact ambiguous (think about what was “on the plane”: the snakes, or the movie?). Linear ordercannot help us in understanding the source of this ambiguity, but phrases can! The tree below represents only one ofthe meanings of this sentence.

45

(53) A tree structure for the sentence I watched the movie about snakes on a plane in August.Sentencehhhhhhhhhh

((((((((((N

I

•hhhhhhhhhhhhh

(((((((((((((•PPPP

����V

watched

•PPPP

����Det

the

•PPPP

����N

movie

•aaaa

!!!!P

about

•aaa

!!!N

snakes

•b

bb"

""P

on

•QQ��

Det

a

Noun

plane

•ZZ��

P

in

Noun

August

This might look very complicated at first, but don’t worry; we will talk about how we proceedstep by step to building trees like that above. All you need to know right now is that certain wordsare grouped together before being grouped with other words. The top of the tree is a sentenceand each of the branches (ultimately) ends in a word. This tree has all the information the linearorder representation in (52) has (part of speech, order of words) but it also has something more.The words are grouped together into little mini-trees, and at the top of each mini tree there is asymbol •. This symbol represents phrases: the kind of thing that words are grouped into. Thesephrases are put together with other words and phrases to form sentences.

Syntax is the study of these kinds of structures. It is concerned with the ways that words areassembled into phrases (the little trees marked with •), and phrases are assembled into sentences.The next section will describe what kinds of word groups count as phrases, and how to identifythem. That is, we will see that there is “hidden structure” in sentences. This is one of the mostamazing things about language and linguistics.

5.2 Words, phrases, and constituency tests

We can see a relationship between words and phrases by noting that whenever we find a singlenoun in a very simple sentence, it can be replaced with a group of words—all related to each otherand crucially containing a noun—and result in a new, also grammatical sentence. To exemplify,let’s start with the simple sentence in (54).

(54) John runs.

46

We could replace the single noun John with a group of related words, as in (55).

(55) The skinny little guy in the neon tank top runs.

The sentence is still fine—-and, if you know the John I know, it means the same thing. Thegroup of words that replaced the noun is called a noun phrase, a phrase that is built around acentral noun (guy, in this case). The entire phrase can be traded for a single noun (John) and thewhole sentence still makes sense. Notice that while we can replace this noun phrase (abbreviated‘NP’) with a noun and still have a good sentence, we cannot replace it with any other part ofspeech. (Remember that an asterisk before the sentence indicates that it is ungrammatical.)

(56) a. *Swim (V) runs.b. *Fast (Adj) runs.c. *Down (P) runs.

At this point, we should also consider two other types of phrases: verb phrases and prepo-sitional phrases (PP). A verb phrase (VP) is a group of words that can substitute for a singleverb, or that can have a single verb substituted for them; similalry, a prepositional phrase is agroup of words that can substitute for a single preposition, or that can have a single prepositionsubstituted for them.

In another simple sentence, John ate, where ate is the verb, we can put in various groups ofwords—each one crucially containing a verb—-in place of ate; both ate and the phrases it can bereplaced with are VP’s.

(57) a. John ate potatoes.b. John ate at the dinner table.

We can see that prepositions also have a relationship with prepositional phrases: anywherea preposition can occur on its own— in a simple sentence like John fell down, for example—youcan have a longer group of words including a preposition. This group is, of course, a PP:

(58) a. John fell down the stairs.b. John fell down the dangerous winding stairs.

While an intuitive definition of a noun phrase like “A noun phrase is a group of words thatdescribe a noun, and can go in sentences where nouns can go” may be a useful guide, we wouldlike to establish more explicit criteria for determining whether or not a particular word or groupof words is a constituent, first of all, and, if it is a constituent, what kind of constituent it is. Wewill use formal tests called constituency tests to prove whether groups of words are particularkinds of phrases. Let’s define a constituency test as in (59).

(59) Constituency testA constituency test is a way of testing whether a group of words forms a phrase.

There are different constituency tests to determine whether groups of words are differenttypes of phrases. The constituency test for noun phrases is the pronoun test. To apply this test,you replace a group of words that you think might be a noun phrase with a pronoun; if theresulting sentence is grammatical, and means the same thing as the original sentence (that is,if the resulting sentence is a shorter way of saying the original one), the group of words youreplaced is a NP.

47

(60) Pronoun testReplace a group of words with a pronoun, (e.g. he, she, him, her, it, they, or them). If anyof these replacements results in a grammatical sentence with the same meaning as theoriginal sentence, the group of words is a noun phrase. If none of the replacements hasthis result, the words are not a noun phrase.

(61) a. I watched the movie about snakes on a plane last week.b. I watched it last week.

The new sentence is grammatical and means the same thing as the original sentence, so thewords in boldface in the first sentence form a noun phrase.

(62) a. Most people like the guy in that movie about snakes.b. Most people like him.

The new sentence is grammatical and means the same thing as the original sentence, so thebolded words in the first sentence form a noun phrase.

(63) a. I watched the movie about snakes last week.b. *I watched the it last week.

The new sentence is not grammatical, so the words in boldface in the first sentence do notform a noun phrase.

Now, consider the following example, in which the group of words away his homework follow-ing the verb has been replaced by the pronoun it.

(64) a. Bob threw away his homework.b. Bob threw it.

The new sentence that results from this replacement is grammatical, so it is not marked witha *. However, it does not mean the same thing as the original sentence, so the bolded words inthe first sentence do not form a noun phrase.

It is important to understand that phrases can be just single words. We can prove this usingthe pronoun test on a sentence like Bob likes cheese. The sentence Bob likes it. is grammatical andhas the same meaning as the original sentence. Since cheese can be replaced with the pronoun it,cheese is a one-word NP. It is also a noun, of course.

There is also a constituency test for verb phrases: the do-form test.

(65) Do-form testReplace a group of words with a form of do so. If the replacement results in a grammaticalsentence with the same meaning as the original sentence, the group of words is a verbphrase. If the replacement does not have this result, the words do not constitute a verbphrase.

(66) a. I’m going to watch the movie.b. (I said I′d watch the movie, and) I′m going to do so.

48

The new sentence is grammatical, so the bolded words in the first sentence form a verbphrase. The part in parentheses is added to provide more context, so that the sentence with ‘doso’ sounds more natural. Let’s look at another example.

(67) a. I watched the movie about snakes in August.b. (I wanted to watch the movie in July but,) I did so in August.

The new sentence is grammatical, so the bolded words in the first sentence form a verbphrase.

(68) a. I watched the movie about snakes in August.b. *I did so about snakes in August.

The new sentence is not grammatical, so you can feel fairly secure in concluding that thebolded words in the first sentence do not form a verb phrase in the original sentence.

Finally, the constituency test for prepositional phrases is the there/then test.

(69) There/then testReplace a group of words with there or then. If the replacement results in a grammat-ical sentence with the same meaning as the original sentence, the group of words is aprepositional phrase. If the replacement does not have this result, the words are not aprepositional phrase.

In the example illustrating application of the There/then test in (70), the new sentence result-ing from replacing the PP in August with then is grammatical, so this test provides evidence insupport of the bolded words in the first sentence forming a prepositional phrase.

(70) a. I watched the movie about snakes in August.b. I watched the movie about snakes then.

Similarly, the new sentence formed in (71b) is grammatical, so the test provides support thatthe bolded words in the first sentence form a prepositional phrase.

(71) a. I watched the movie in Brooklyn.b. I watched the movie there.

In contrast, in (72), the new sentence is not grammatical, so the bolded words in the firstsentence do not form a prepositional phrase. Also, we should take note of the fact that here to isnot a preposition. It indicates the infinitival form of the verb watch.

(72) a. I wanted to watch the movie about snakes in Brooklyn.b. *I wanted there/then.

There is another constituency test that can in many cases be used to successfully identifyNPs, VPs, and PPs. When one of the other tests doesn′t work, the movement test often does. Themovement test works as follows.

(73) Movement testIf a group of words can be moved together to the front of a sentence and result in a newgrammatical sentence, then that group of words is a constituent.

49

As the following examples show, this test works for NPs, VPs, and PPs.2

(74) a. I watched the movie about snakes on a plane last week.b. The movie about snakes on a plane I watched last week.

(75) a. Students face many new challenges in college.b. In college students face many new challenges.

(76) a. Students face many new challenges in college.b. In college students face many new challenges.

(77) a. I always enjoy going for long walks in the countryside.b. Going for long walks in the countryside I always enjoy.

As you may have noticed, we can’t use this test just as it is used above to target verb phrases,because moving a verb phrase to the front of the sentence without doing anything else willalways result in an ungrammatical sentence in English. Consider our first example above:

(78) a. I watched the movie about snakes on a plane last week.b. *Watched the movie about snakes on a plane last week, I.

But there is another test, similar to the movement test, which we can use to target VPs inaddition to NPs and PPs. To apply this test to a string of words XP to see if it is a constituent,put that group of words in a construction of the form It is/was XP that/where/who/etc.. In thecase of our current example, for instance, we can put the VP into this construction and get thegrammatical sentence Watch the movie about snakes on a plane is what I did. This construction iscalled a cleft-construction, so we will call this the cleft-formation test.

(79) Cleft-formation testIf a group of words can occur in the position following It is/was in a sentence of the formIt is/was VP/NP/PP that/where/who..., then that group of words is a constituent.

Here are some examples:

(80) VP Clefta. I watched the movie about snakes.b. Watched the movie about snakes is what I did.

(81) NP Clefta. I watched the movie about snakes on a plane.b. The movie about snakes on a plane is what I watched.

(82) PP Clefta. I watched the movie about snakes on a plane in the theatre.

2An example like (82b) below might sound odd at first, but it gets better if you play around with the intonation,particularly, if you add a pause after the fronted NP, and/or put it in the right context. e.g. The movie about snakes on aplane, I watched last week. The one about sharks on a ship, I’m going to watch tonight.

50

b. In the theatre is where I watched the movie about snakes on a plane.

Identifying phrases in a sentence can often be slightly easier than this. There is an importantproperty of phrases that helps us identify them:

EVERY N IS PART OF AN NP.EVERY V IS PART OF A VP.EVERY P IS PART OF A PP.

The technical term for this relationship between N and NP, V and VP, and so on, is head.Every NP has as its head an N, and so on for each phrase. So we can generalize this rule to saythat any phrase XP (where X stand in for any part of speech) has an X in it that is, has a head X.

This means that when you are asked to find all of the phrases in a sentence, if the sentencehas 3 nouns, 1 preposition, and 1 verb (like the one below), you should find 3 NPs, 1 PP, and atleast one VP. Try to find the phrases in the sentence:

(83) IN

watchedV

theDet

movieN

atP

noonN

Here are the phrases, built up one by one:

1. Noun phrases (NPs) in bold:

IN

watchedV

the movieDet N

atP

noonN

2. Prepositional phrases (PPs) underlined:

IN

watchedV

the movieDet N

at noonP N

3. Verb phrase (VP) underlined again:

IN

watched the movie at noonV Det N P N

4. Verb phrase (VP) underlined again:

IN

watched the movie at noonV Det N P N

5. Sentence (S) underlined again:

I watched the movie at noonN V Det N P N

There are a few things that we should take note of here. The first is that there are twoVPs, even though there is only one verb. We have the larger VP, watched the movie in Brooklyn,and the smaller VP, watched the movie. We will cover this in more detail when we outline ourphrase structure rules for verb phrases. Also, we see that VPs start with verbs and PPs start withprepositions. However, we see that NPs start with either a noun or a determiner. Finally, thesediagrams represent the fact that phrases often contain other phrases; for example, the PP has anNP inside it, and so does the smaller VP. The big VP has both a VP and a PP in it.

51

5.3 Phrase structure rules: an overview

In studying the structure of sentences, a major goal is to provide a general description of whatsentences, NPs, VPs, PPs, and other phrases look like in different languages. We do that bywriting general rules for the structure of each of these kinds of phrases. These rules are calledPhrase Structure Rules (PSRs).

Let’s say we′ve figured out (as we likely have by now) that an English sentence may beginwith a noun phrase and then a verb phrase. We can write a schematic rule that says this:

(84) S → NP VP

In words, what this sentence says is “A sentence consists of a noun phrase followed by a verbphrase.” To be really explicit, the pieces of this rule mean the following:

(85) a. S = “a sentence” → = “consists of”b. NP = “a noun phrase”c. VP = “a verb phrase”

The left-to-right order in the rule corresponds to left-to-right order in a sentence. That’s wherethe “noun phrase followed by a verb phrase” part comes from—the order of the constituents afterthe arrow.

Syntax rules can, of course, build things other than sentences:

(86) NP → Adj N

This rule says that a noun phrase consists of an adjective followed by a noun, like in the nounphrase pretty girls. We know, though, that not all English noun phrases have adjectives in them.You can start a sentence with just a noun, as in Girls laugh. or Mary laughs. Other times, you canput several adjectives in a noun phrase: small smart pretty girls laugh, etc. Because adjectives areoptional and because of our ability to continually add adjectives, we are going to end up revisingthe NP rule, so don′t commit this one to memory.

Our constituency tests showed us we also have verb phrases, VPs. Here′s an attempt at aphrase structure rule for a verb phrase such as watch a movie.

(87) VP → V NP

We have the verb watch and the NP a movie. This phrase structure rule itself has anotherphrase in it (as did the Sentence phrase structure rule). This means that one phrase structurerule leads to another, and that′s why we can have phrases inside of other phrases.

Note: The VP watch a movie provides another reason for us to revise our NP rule. It beginswith the determiner a, and presently, our NP rule doesn′t indicate a place for determiners.

5.4 Syntactic Trees

Throughout the rest of this chapter, we are going to be using our phrase structure rules togenerate syntactic trees, so before we go into any more detail about the nature of phrase structurerules, let’s learn the basics of tree-drawing.

Any phrase structure rule can be diagrammed as a tree, as shown in the table for the ruleswe proposed in the previous section.

52

(88)

Phrase structure Rule Corresponding tree diagramS → NP VP S

ll,,NP VP

NP → Adj N NP@@��

Adj NVP → V NP VP

@@��V NP

(Note: These phrase structure rules are just examples to get us started. We are probablygoing to keep S → NP VP, but the other two will be changed slightly, so don’t commit them tomemory.)

You can put these altogether when you have a whole sentence:

(89) SXXXXX

�����NP

QQ

��

Adj

pretty

N

girls

VPPPPP

����V

watch

NPH

HH�

��Adj

interesting

N

movies

A syntax tree is a representation of the structure of a single sentence of the way that theindividual words are grouped into phrases and eventually form the full sentence. A tree showsevery phrase that’s in the sentence, so if your project is to draw a tree, it’s a really excellent ideato start by labeling every word with its part of speech, and then using constituency tests to findall of the phrases in the sentence.

Question: Do the constituency tests for NPs and VPs confirm that we have grouped thewords in this sentence correctly? If not, we’ve made a mistake.

Very importantly, the groupings of words into phrases and into the sentence in a syntax treeabsolutely must be an accurate representation of the phrase structure of the language. You’llusually be asked to draw a syntax tree after you’ve written phrase structure rules for a language.Other times, you might be given a tree and asked to write the rules. If you just make sure thatyou’re identifying all phrases when both writing phrase structure rules and drawing trees, andthat the phrases in your tree can be generated using your phrase structure rules, this shouldn’tbe a problem.

Important properties of trees:

• Every word must be labeled with its part of speech.

• Every phrase present in the sentence must be correctly drawn in the tree.

• Every word must be connected into the tree.

• The groupings of words into phrases and phrases into sentences as shown in the tree mustbe consistent with the language’s phrase structure rules.

53

Let’s close this section by reviewing some terminology we saw in class when we went overthe handout containing terms for describing the formal properties of tree structure. Recall thatthe category that appears to the left of the arrow in the phrase structure rule is the mother andthe categories that appear to the right of the arrow are the daughters. Daughters that have thesame mother are sisters.

Consider the phrase structure rule S → NP VP and its corresponding tree:

(90) S motherPPPP

����NP daughter VP daughter

The S node is the mother and its daughters are the NP and VP nodes. NP and VP are sisters.

5.5 Phrase structure rules

5.5.1 The structure of noun phrases

Now let’s return to the issue of why (86), repeated below, doesn’t make for a good NP rule.

(91) NP → Adj N

We’ve already said that it doesn’t include determiners, it doesn’t allow for more than oneadjective, and it doesn’t tell us which elements are optional. We will also see that NPs can havePPs and sentences inside of them. Here are some more examples of NPs.

(92) a. cats

b. black cats

c. the black cats

d. the small black cats

e. the small black cats on the couch

f. John′s small black cats

g. the small black cats that John owns

Let’s take the NPs in (a) through (c) first. Here is an attempt at a phrase structure rule:

(93) NP → (Det) (Adj) N

This rule says that an NP consists of an optional determiner, an optional adjective, and a noun(which is required).

This looks good, right? We would draw the following tree for (92c):

(94) NPaaaa

!!!!Det

the

Adj

black

N

cats

54

This is what is called a flat phrase structure. That is because below the NP level, all theother categories are lined up (technically, they are all daughters to the NP). But there is evidencethat phrase structure is not flat in this way. There is another kind of constituency test thatdistinguishes different constituents within an NP. You can often replace certain parts of an NPwith the word one(s).

(95) a. the red shoes → the red ones

b. the silly linguistics teachers → those silly ones

c. the small green bottles of beer → the small ones

d. some intelligent students of language → some intelligent ones

These replacements suggest that the adjective and the noun, and even the noun itself, form aconstituent to the exclusion of the determiner. But we still want the whole NP to come togetheras an NP. To do so, we are going to add in another “level” to our phrases. We have heads (thecategories at the bottom that are just above the actual words) and we have the phrases, the XPson top of the phrases. Let’s say there can be a level in-between, and we will notate this with anapostrophe/prime symbol next to the category of phrase. So, in the tree below, we have N′ (read“N-bar”).

(96) NPaaaa

!!!!Det

the

N′aaa

!!!Adj

small

N′b

bb"

""N′

N

bottles

PP@@��

P

of

NP

N′

N

beer

All of this tree below the first N′—-the subtree from the tree above below—is replaceable byone(s):

55

(97) replaceable by ones: N′aaa

!!!Adj

small

N′b

bb"

""N′

N

bottles

PP@@��

P

of

NP

N′

N

beer

With this extra level of structure we can account for why the word ones can replace bottlesof beer but leave the adjective small. This is because it forms a constituent to the exclusion ofthe adjective (and the determiner). In general, these extra levels of structure—which we will callBar-levels—are where we attach modifiers, such as adjectives and some prepositional phrases.The number of bar-levels is, in principle, infinite. In this way, it is recursive (it can recur overand over again). This is why we could keep adding adjectives in an NP, as in (92d) above.So we are going to replace our current NP with the following:

(98) a. NP → (Det) N′

b. N′ → A N′

c. N′ → N

The first PSR, (98a), says that an NP consists of an optional determiner and necessarily an N′.The next two rules state what an N′ can consist of. It can consist of an adjective and another N′,or it can consist of just an N. It is the rule in (98b) that allows us to have an infinite number ofadjectives. Because we have N′ on both sides of the arrow, this phrase structure rule can keepcalling itself, as in (99).

(99) several small black scary angry catsNP

PPPP����

Det

several

N′PPPP

����A

small

N′aaaa

!!!!A

black

N′aaa

!!!A

scary

N′

ZZ

��

A

angry

N′

N

cats

56

We can also have prepositional phrases inside of NPs, as we saw in the case of the small blackcats on the couch in (92e), and which we see in the tree in (96). The rules for this are given below,followed by a tree diagram of this particular NP.

(100) a. N′ → N′ PPb. PP → P NP

(101) NPaaaa

!!!!Det

the

N′aaaa

!!!!A

small

N′aaa

!!!A

black

N′H

HH�

��N

cats

PPb

bb"

""P

on

NPZZ��

Det

the

N′

N

couch

Some NPs indicate possession, and the possessor goes in the same place as the determiner,as in (92f), Johns small black cats. Our rule for possessive NPs is similar to the NP rule we alreadyhave.

(102) NP → NP’s N′

In this rule, the possessor is represented by “NP’s” and the thing that is possessed is repre-sented by “N”.

(103) NPaaaa

!!!!NP’s

John

N′HHH

���A

small

N′

ZZ��A

black

N′

N

cats

The possessor is represented by an NP instead of an N′ because it can contain a determiner,as in the pretty girl’s small black cats. Unlike adjectives, determiners and possessors do not recur:an NP can have only one determiner or one possessor.

57

(104) a. * The some catsb. * A every dogc. * Freds the house

Finally, we need to account for (92g) the small black cats that John owns. Here, we have asentence—that John owns—inside of our noun phrase. This sentence is different from the othersentences weve looked at because it begins with what we call a complementizer, that. (Of course,we can also say this NP without the complementizer, as in the small black cat John owns. In thiscase, we would say that the complementizer is null, or silent.) Complementizers are words suchas that, if, and whether. The rules that account for complex noun phrases—that is, noun phrasesthat contain sentences inside them— can be written as follows.

(105) a. N′ → N′ CPb. CP → C S

(106) NPaaaa

!!!!Det

the

N′aaaa

!!!!A

small

N′aaa

!!!A

black

N′HHH

���N′

N

cat

CPHHH

���C

that

SZ

Z�

�NP

N′

N

John

VP

V

owns

With the exception of the VP rule, all the other rules used to generate this tree should befamiliar at this point. So to summarize, these are the rules we need to account for the structureof NPs:

(107) a. NP → (Det) N′

b. NP → (NPs) N′

c. N → A N′

d. N′ → Ne. N′ → N′ PPf. PP → P NPg. N′ → N′ CPh. CP → C S

58

5.5.2 The Structure of Verb Phrases

We’ve spent a lot of time looking at how noun phrases are formed. Since we have some back-ground, dissecting verb phrases will be much easier. Let’s start with a list of verb phrases wewould like to account for.

(108) a. jumpb. jump aroundc. jump into the poold. playe. play the pianof. should play the pianog. heard that John plays the piano

Looking at (108a)-(108c), we see that VPs can contain a single verb, a verb followed by apreposition, or a verb followed by a preposition that is followed by a noun phrase. The followingrules capture this:

(109) a. VP → VP (PP)b. VP → V

Let’s draw a tree for something like (??), a verb phrase which contains a prepositional phrase:

(110) VPaaa

!!!VP

V

jump

PPHHH

���P

into

NPcc##

Det

the

N

pool

We have a VP, instead of just a V, as the first daughter because we wil want to be able toaccount for phrases like jumped off the diving board into the pool. Here, we have two PPs, and ourrule in (117a) would allow us to incorporate both of them by recalling the VP.

Transitive and intransitive VPs

Of course, VPs can also have noun phrases in them. As we saw in class, some verbs must befollowed by an NP, like those in (111) below. In contrast, others cannot be followed by an NP, likethose in (112). (Remember that the * indicates that the sentence is ungrammatical.)

(111) a. The children devoured the candy.b. * The children devoured.c. The students brought their books.d. * The student brought.

59

(112) a. The baby giggled.

b. * The baby giggled her mother.

c. * The baby giggled a giggle.

d. The parents chatted.

e. * The parents chatted childcare/recipes/many words.

We’ll call verbs such as those in (111), whose meaning seems to require an NP object, transi-tive verbs, and verbs such as those in (112) intranstive verbs. We will reflect this in our rules inthe following way:

(113) a. VP → V[Trans] NP

b. VP → V[Intrans]

Following the rule in (113a) to make the tree corresponding to the transitive verb phrasedevour the pizza, we get the structure in (114).

(114) VPH

HH�

��V

devour

NPZZ��

Det

the

N′

N

pizza

In the case of the verb phrase pouted in (112), we have an intransitive verb, so we follow therule in (113b) to get a tree like (115) below.

(115) VP

V

pout

Of course, transitive VPs, those verb phrases whose verb has an NP as immediate sister tothe verb, can also have PP modifiers, like in devoured the pizza quickly in the kitchen or received theletter at the post office.

Using the rule in (117a) above, we can draw the tree corresponding to these phrases. Firstexpand the higher VP according to the rule in (117a) above (and the PP according to the PP rulein (100b) further above). So you get:

60

(116) VPHHH

���VP PP

bbb

"""

P

in

NPQ

Q�

�Det

the

N

kitchen

Now, expand the lower VP according to the rule in (??) to finish constructing the tree:

(117) VPXXXXX

�����VP

HHH���

V

devour

NPZZ��

Det

the

NP

N′

pizza

PPb

bb"

""P

in

NPQ

Q�

�Det

the

NP

N′

N

kitchen

Auxiliary verbs

Sometimes VPs contain auxiliary verbs (what you probably learned as helping verbs). The rulefor a VP involving an auxiliary verb—-which include, as we talked about in class, modal auxil-iaries such as may, might, must, can etc. and aspectual auxiliaries such as have, was, is etc. Ourrule for VPs involving an auxiliary verb is shown in (118), with a corresponding tree in (119).

(118) VP → AUX VP

(119) VPaaa

!!!Aux

should

VPHHH

���V

play

NPZZ��

Det

the

N′

N

piano

Finally, some VPs contain sentences, just as some NPs do, as in (108g), heard that John playsthe piano.

61

(120) VP → V CP

(121) VPaaaa

!!!!V

heard

CPaaa

!!!C

that

Saaa

!!!NP

N′

N

John

VPHHH

���V

plays

NPZZ��

Det

the

N′

N

piano

5.5.3 Sentential subjects

When a VP includes a CP, such as in the sentence above, the CP is the object of the transitiveverb. CPs can also be subjects of sentences, as in the sentence That Mary shops without couponsbothers John. Here, instead of starting our sentence with a noun phrase, we are starting it with acomplementizer phrase that Mary shops without coupons. The rule we will use for such sentencesis S → CP VP. Using this rule, the tree for the sentence That Mary shops without coupons bothersJohn.

(122) Shhhhhhhhhh

((((((((((CP

PPPP����

C

that

SPPPP

����NP

N′

N

Mary

VPaaaa

!!!!VP

V

shops

PPH

HH�

��P

without

NP

N′

N

coupons

VPQ

Q�

�VP

V

bothers

NP

N′

N

Tom

62

5.5.4 Coordination

The final rule we will introduce now is the rule for coordinating phrases. As we saw when wedefined constituency tests, two phrases of any category can be joined together with a conjunctionsuch as and or or. Since this rule can target any category of phrase, we represent the phrase withα, to indicate that the rule is not specific to any particular category of phrase.

(123) αP → αP CON αP

This rule says that any category can consist of two members of itself. So, we can conjoin twoVPs to form a larger VP or two NPs to form a larger NP, etc. Let’s see how this works.

(124) SXXXXX

�����NP

N′

N

Theresa

VPXXXXXX

������V

likes

NP

N′XXXXXXCC

������N′Q

Q�

�A

fine

N′

N

cuisine

Conj

and

N′

ZZ��A

good

N′

N

wine

Here we have two N′s conjoined, fine cuisine and good wine. This is the only time when a nodewill have three branches. Here we have two N′ daughters, as well as and. It is very important toremember that whenever we have conjunction, the two daughters are necessarily identical to themother. So, we could not ever have a VP mother that has as its conjoined daughters a VP anda PP, for instance. We don’t ever put the exact rule αP → αP CON αP into our syntactic trees.Rather, we use the particular category that α represents; so in (124) above, we use N′.

5.5.5 Inventory of English Phrase Structure Rules

The following is a summary of all of the PSRs we have seen here that we will be using for English.You should be able to provide example phrases and sentences that use the PSRs. Memorize thefollowing!

63

English Phrase Structure RulesSentence PSRsS → NP VP

S → CP VP

CP → C SNP PSRsNP → (D) N′

NP → (NP’s) N′

N′ → N′ PP

N′ → A N′

N′ → N

N′ → N′ CP

VP PSRsVP → VP PP

VP → V NP

VP → AUX VP

VP → V

VP → VP CP

PP PSRPP → P NP

Conjunction PSRαP → αP CONJ αP

5.6 How to do syntactic analysis (and find phrase structure rules)

There will be times when you’ll be given some sentences in a language that isn’t English, andyou will be asked to write phrase structure rules that describe the language. Sometimes you willalso be given some English data that we haven’t studied, and asked to write some new phrasestructure rules that describe how the new data works. These phrase structure rule writing tasksare pretty similar.

Start by labeling every word with its part of speech. You’ll need to know all of the words’parts of speech eventually, because you write rules that say which parts of speech (or whichkinds of phrases they contain, rather than literally what words they contain, this can really helpwith seeing similar structures in different-looking sentences.)

Let’s take a moment to concentrate on figuring out the parts of speech of non-English words:

64

• If you are given a non-English sentence with a line under it where every word has anEnglish translation (like below; this line, and the words in it, is called a gloss), every non-English word will have the same part of speech as its English gloss.

• If you are just given non-English sentences and their English sentence translations, andasked to figure out which non-English word corresponds to which English word, again,once you figure this out, you can basically trust that all non-English words will have thesame part of speech as their English translations.

So in this French example below, the parts of speech are as indicated; we know this becausethey’re the same for the French words as for these words’ English correspondents:

(125) DetUne

Nfille

Adjjolie

Vmange

Detle

Nfromage

‘A pretty girl eats the cheese.’

Let’s say that the task is to write all the phrase structure rules you possibly can for French basedon this data (you’ll usually have more than one sentence).After the words are all labeled with their parts of speech, find all the phrases in the sentence.Another thing you can count on, in foreign data you’ll receive, is that the non-English wordswhich correspond to the words in an English noun phrase also form a noun phrase. So youcan find all the phrases in the English translation, then use this information to find all the phrases in theforeign sentence. That is, in the English sentence, a pretty girl is a noun phrase (you should beable to prove this using the pronoun test). This information can be used to find a French nounphrase, which is composed of the French translations of these words: une fille jolie (literally, a girlpretty). Identifying all of the French NPs in this sentence in this manner gives you the followinginformation:

(126) DetUne

Nfille

Adjjolie

Vmange

Detle

Nfromage

a girl pretty eats the cheese‘A pretty girl eats the cheese.’

Now you have examples of some NPs, and can describe the individual NPs like this:

NP = Det N Adj (une fille jolie)NP = Det N (le fromage)

The information from these schematic structures of individual NPs can be used to write asingle rule for the structure of a French NP, which says that an NP consists of a determiner, anoun, and an optional adjective:

(127) NP → Det N (Adj)

Using what we learned about English NPs, we could write this rule as:

(128) NP → Det N′

65

Since, at this point, we don’t have evidence that the determiner is optional, we won’t put it inparentheses. Now we need to write the rules for N′. If this were English, it would be N′ → A N′

and N′ → N. But, as we saw above, some adjectives comes after the noun, not before it. So, ourrules need to be:

(129) a. N′ → N′ Ab. N′ → N

Next, going back to the goal of writing as many phrase structure rules as possible basedon this sentence, look to see what other kinds of phrases you can find in the English sentence.Remember that, as we have discussed, every N is part of a NP, every V is part of a VP, and everyP is part of a PP. You can use this guideline to make sure you find every single phrase there isto find in the data, starting with English and then finding the corresponding phrases in French.So since you know that eats is a verb, you know there must be a VP around it somewhere; thedo-form test can be used to reveal the VP:

(130) [NP A pretty girl] [VP eats [NP the cheese]]

This means that the French translation of eats the cheese also forms a VP, so we now know thatwe can divide the sentence as follows:

(131) DetUne

Nfille

Adjjolie

Vmange

Detle

Nfromage

‘a girl pretty eats the cheese’‘A pretty girl eats the cheese.’

This lets us write another rule, this one for the structure of the VP:

(132) VP → V NP

One might wonder why this rule isn’t “VP → V Det N”. The short answer is that phrasestructure rules always refer to the biggest things inside them. Phrases have the potential to bebigger than individual words, so phrase structure rules refer to phrases rather than to individualwords when possible.3

Now that we have grouped every single word in the sentence into a phrase, we can see anddescribe the structure of a French sentence:

(133) S → NP VP

Remember, we generally try to write phrase structure rules for NPs, VPs, PPs, and Ss. Sowhen you are trying to figure out what’s going on in some new language, you might stop atsome point and ask yourself things like, “Do I have any evidence for what an NP looks like inthis language?” Hmm, I should find all the English NPs and then look at their translations.” Thesame for a VP, and for a PP, if you find them in the English sentences. Languages will almostalways have VPs and NPs, and of course they always have Ss, but you won’t necessarily alwaysget data about PPs.

Finally, a good way to double-check the rules you have written is to pick at least two of thesentences you were given and try to construct them using your rules. If you can, you are in greatshape; if you can’t, you know you need to fix something before you continue.

3This is not always possible, since phrases some times do consist of single words, as we have seen.

66

5.6.1 On combining sets of phrase structure rules into single rules

Sometimes you can combine rules into a single, more efficient rule; other times, you can’t. It’snot strictly wrong to leave rules uncombined, so when in doubt, don’t combine them. But it isgenerally better to have one or two general rules than sixteen specific, redundant rules, so whenit’s clearly safe, it will make me happier (and you might get a couple more points, and also yourhand will get less tired) if you combine rules.A case where you cannot combine rules:

Say you have the following rules:

(134) a. VP → V NPb. VP → V PP

Its not correct to combine these to form “VP → V (NP) (PP)”, because you dont have evidencethat a sentence can have both NP and PP after a V, or that a sentence can have just a V and neithera following NP nor a following PP, or for what order the two would be in if they could co-occur.

A case where you can combine rules:

If you have the following rules:

(135) a. VP → Vb. VP → V NPc. VP → V PPd. VP→ V PP NP

...then it is okay to combine them to “VP → V (PP) (NP)”, because you know that a single VPcan have both NP and PP (or neither), and you know which order they go in when they combine.

5.7 Structural ambiguity

A sentence is ambiguous if it has more than one meaning.4

Ambiguity can arise from different sources. We have already seen examples of ambiguity inthe lexicon—i.e. of morphological ambiguity. If the meaning of a sentence is ambiguous due toan ambiguity in the meaning or category of a word, the word and thus the sentence are said tobe lexically ambiguous. If the meaning of the sentence is ambiguous due to an ambiguity insyntactic structure, i.e., if a sentence has more than one syntactic structure associated with it, thesentence is said to be structurally ambiguous.

(136) a. Bob caught a mole.b. Bob chased the dog with a stick.

The sentence in (136a) is an example of lexical ambiguity. The word mole has two meanings:it can refer to a dormant spy or to a burrowing mammal. So, the sentence in (136a) could haveone of two meanings given in (137).

4The notes in this section are based heavily on those of Youri Zabbal.

67

(137) Bob caught a mole.a. Bob caught a dormant spy.b. Bob caught a burrowing mammal.

Context decides which meaning you understand in any given discourse. If the topic underdiscussion is espionage or something related, then the sentence in (137) has the same meaning asthe sentence in (137a). However if the discourse is about farm animals, then the sentence in (136a)has the same meaning as the sentence in (137b). Since both meanings are available, it is necessaryfor the grammar to account for both meanings. The grammar can do this if the string of soundsthat make up the word mole has two separate entries: (i) a noun meaning a dormant spy and (ii)a noun meaning a burrowing animal. Two separate meanings sharing a single string of soundsare called homophones (or homonyms). The context will determine the meaning attributed tothe sounds. We will not be spending very much time on the subject of lexical ambiguity; we will,instead, concentrate on structural ambiguity.

Going back to the sentence in (136b), we have another interesting ambiguity. This sentenceis an example of structural ambiguity. It could be Bob who has a stick or it could be the dogthat has a stick. So, that sentence is ambiguous between two meanings. Each meaning can beparaphrased as follows.

(138) a. Bob chased the dog with a stick.b. With a stick, Bob chased the dog. (Paraphrase of meaning 1)c. Bob chased the dog who carried a stick. (Paraphrase of meaning 2)

Again, context decides which meaning you have in any situation. If the context is Bob,angry, chasing a dog out of his flowerbed, the likely meaning for the sentence in (138a) is thatparaphrased in (138b). However, if the context is Bob, happy, playing with a dog, the meaning forthe sentence in (138a) is paraphrased in (138c). Since both meanings are possible, it is necessaryfor our grammar to account for both. Please keep in mind that we are only interested in theoriginal sentence in (138a). The sentences in (138b-c) were used to paraphrase the two possiblemeanings, but we will always be drawing trees to represent the original sentence.

(139) a. Meaning 1: Bob chased the dog and a stick was used as the instrument in the chasingb. Meaning 2: Bob chased the dog and the dog had a stick

The ambiguity arises from where the prepositional phrase PP with a stick joins in the syntacticstructure. This is called structural ambiguity.

A sentence or phrase is structurally ambiguous if its words can be grouped together in dif-ferent ways. Ultimately this just means that a sentence or phrase is structurally ambiguous ifit has more than one tree structure (just as was the case for words in morphology). Each treestructure has a meaning associated to it. Sometimes, the meanings of two different structureswill be virtually indistinguishable; however, the structural ambiguity still exists.

(140) If a sentence has more than one possible corresponding tree structure, it is structurallyambiguous.

In order to understand structural ambiguity, we will need to have an understanding of theconcept of modification.

68

(141) Modification: A PP or A modifies (roughly, describes) the phrase or bar-level projectionthat it is a sister to.

Here is an example of structural ambiguity in the noun phrase.

(142) Green eggs and ham

a. NPPPPPPSS

�����NP

N′

ZZ

��

A

green

N′

N

eggs

Conj

and

NP

N′

N

ham

b. NP

N′PPPP

����A

green

N′

N′PPPP

����N

eggs

Conj

and

N′

N

ham

The structure in (142a) corresponds to the meaning in which the eggs are green but notnecessarily the ham. This is because the adjective green only modifies the N′ containing the Neggs. The structure in (142b) corresponds to the meaning that both the eggs and the ham aregreen. The adjective green modifies the conjoined N′ that contains both nouns, eggs and ham.This structural ambiguity arises because the adjective can have as its sister either a small N′

that contains just eggs or a larger N′ that contains eggs and ham. When we are speaking, wewould signal which meaning we intended with a pause. So, if we say green eggs [pause] and ham,this would correspond to the first meaning and if we say green eggs and ham [pause], this wouldcorrespond to the second meaning. Try this with a friend. Notice that this string of words cannotmean that the ham is green, but not necessarily the eggs. Why do you think this is so?

Now let’s look at structural ambiguity in the verb phrase. This brings us back to sentences likeBob chased the dog with the stick.The sentence has two structures, each with a different meaning,corresponding to the two potential sisters for the PP with the stick. If the PP is sister to the VP,the resulting meaning is Meaning 1 above, that is, Bob chased the dog while brandishing a stickin his hands. The rules used is VP → VP PP, and the resulting tree is as follows.

69

(143) Bob chased the dog with the stick. SXXXXXX

������NP

N′

N

Bob

VPXXXXX

�����VP

HHH���

V

chased

NPll,,

Det

the

N′

N

dog

PPHHH

���P

with

NPcc##

Det

a

N′

N

stick

If the PP is sister to N′ inside the NP, we get the meaning that the dog had a stick, say, in itsmouth, as Bob was chasing it. The rule used is N′ → N′ PP.

(144) Bob chased the dog with the stick. SPPPP

����NP

N′

N

Bob

VPaaaa

!!!!V

chased

NPaaa

!!!Det

the

N′HHH

���N′

N

dog

PPHHH

���P

with

NPcc##

Det

a

N′

N

stick

70

Chapter 6

Semantics

6.1 The meaning of a sentence: its truth conditions

Language is first and foremost a communicative tool. We say meaningful things all the time, butusually don’t stop to think about precisely what was said, conveyed, or implied, unless in a courtof law, when translating between languages, etc. The meaning of what we say corresponds tothoughts, ideas, or messages that we have and want to share with others. It is independent of theparticular language we choose to use; rather, it is what is common to the content of a text in theoriginal and all its translations into different languages. It is meaning in this broad sense that wewill get a glimpse of in this chapter.

If I say sentence (145) out loud in class, everyone (who was listening when I uttered it) willmost likely have a similar reaction. Or if not the same reaction, then definitely the same ideaabout what I was trying to say when I uttered it. One way to describe the meaning of a sentenceis thus to define its truth conditions the circumstances under which it is true. This is thedirection we will take in this class.

(145) The name of the person sitting closest to the door starts with a “D”.

Exercise: on the difference between truth conditions and truth values.

What are the truth conditions of sentence (145)? Try framing your answer as follows: The sen-tence in (145) is true if and only if.

At present, that is—right here and right now, is it true or false? Can you think of a scenarioin which the truth value of the sentence would be the opposite of what it is right now? Wouldthe meaning of (145) change in that case?

Defining truth conditions of particular sentences in this way is really not very exciting. Whatis exciting is that we can explain how these truth conditions are derived, based on the words thatmake up the sentence. Furthermore, we can do this for every single sentence of English (or anyother language for that matter). This is a compositional approach to the meanings of sentences:the idea that the meaning of a sentence is based on the meanings of its parts and how they areput together.

How may we convince ourselves that the meaning of a complete sentence is based on themeaning of its component words? Think about someone who hears sentence (146) but doesn’tknow what cardamom means, or misheard ate for made in that sentence. A change in just oneword affects dramatically the interpretation assigned to the entire sentence.

71

(146) A blonde girl made the Persian cardamom cookies.

But how exactly are the parts of sentences composed? And what is the end result of puttingthem together? Let’s start answering these questions by looking at adjective-noun combinationslike blonde girl and Persian cardamom cookies. We will make some assumptions about the meaningsof adjectives and nouns, by themselves, and then see how they come together.

6.1.1 Adjective meanings

Here are some adjectives (can you provide evidence that they are adjectives?):blonde well-known skillful formerblue-eyed shiny young quietBelgian fake married goodtall successful true translucent

Exercise: discovering that not all adjectives behave the same.

Let’s take a closer look at these adjectives. For each one we will ask three questions, as listed inthe columns in the table below. If, when filling a particular adjective in the blank, the answer toa question is “yes”, write that adjective in the column below the question. As you will soon see,not all adjectives behave alike, so not all adjectives will end up in the same columns. Note thatthere will be words that will end up in more than one column.

Example:

A blonde pianist .

1. is blonde? YES.

2. is a pianist? YES.

3. is not a pianist? NO.

A(n) ADJ pianist/diamondis ADJ? is a pianist/diamond? is not (necessarily) a pianist/diamond?blonde blonde

The picture that emerges from the table is that there are (at least) three types of adjectives: (i)adjectives that appear in both of the first two columns, like blonde, (ii) adjectives that appear onlyin the second column, like skillful, and (iii) adjectives that appear in the third (and possibly also

72

the first) column, like fake. Moreover, most adjectives, at least judging by the sample we havehere, belong to one of the first two kinds.

Adjectives like blonde, blue-eyed, Belgian, shiny, married, and translucent are called intersective.This is because the way they combine semantically with the noun that they modify (in ourexamples, this is simply the noun they precede) can be modeled by set intersection.

A note to remember in studying and doing assingments Important note to the reader:the following discussion assumes familiarity with terms such as ‘set’, ‘set union’, ‘Venndiagram’, ‘subset’, and ‘cardinality of a set’. Make sure to read the excerpt on basicconcepts of set theory before you continue reading!

We will assume that both adjectives and nouns denote sets of entities. Entities are individu-als or things that have the property described by the adjective or the noun. To take some concreteexamples, the adjective blonde will denote the set of all blonde things, the noun pianist will denotethe set consisting of all people, male or female, that are pianists, and the noun diamond will de-note the set of all objects that have the composition of a diamond. Throughout our discussion wewill assume a very limited context, say Amherst, or UMass, or even just our classroom, in whichall these sets are quite small, and definitely finite. We happily ignore the rest of the universe.

Here are some diagrams that illustrate the sets we have described:

(147) diagram to be distributed in class

Now, what does the phrase blonde pianist denote? Blonde pianists are people that are bothblonde and are pianists. In our ‘set talk’, these are entities that are members of the set of blondethings and of the set of pianists. In other words, they are the entities that are found in theintersection of these two sets. In our small universe, there is only one such entity, namely Jenn.

(148) diagram to be distributed in class

In order to derive the meaning of a phrase (intersective) adjective-noun compositionally, allwe had to do was find the set denoted by the adjective and the set denoted by the noun, andintersect them. This doesn’t work for other kinds of adjectives, specifically those that are notlisted in the first column of the table above.

Let’s think about tall. Somewhat surprisingly, even within the limited context of UMass, it isimpossible to make a list of all the tall people; it depends! Compare tall female students with tallstudents. Who we consider to be tall depends on the set denoted by the noun (female studentsand students in general, respectively). A female student that is relatively tall with respect to othergirls on campus could be considered a tall female student because she is tall for a girl. This sameperson, though, would probably not qualify as a tall student if she is shorter than most guys.

73

The only thing that is guaranteed for an adjective like tall is that if something is a tall X, then thatthing is also an X (this is the question that defines the second column in the table above). Sincethe adjective-noun combination in this case is a subset of the set denoted by the noun, we willcall adjectives of this kind subsective.

(149) diagram to be distributed in class

Other adjectives in our sample that work this way include well-known, successful, skillful, young,true, quiet, and good.

One thing that intersective adjectives and subsective adjectives have in common is that theentities they describe are always in the set of the noun they combine with. This is seen clearlyin the two Venn diagrams on this page. The third type of adjectives, the ones listed in the thirdcolumn of our table, are different. If something is a fake diamond, that thing is not a diamond, andif someone is a former pianist, that person is not (necessarily) a pianist anymore. We will call theseadjectives non-intersective. Notice that adjectives in this group do not behave uniformly withrespect to the first question that forms the first column in our table: something fake is definitelyfake, but a former president, for example, is not former.

Although adjectives form a natural class syntactically, we should be convinced by now thatthey are quite a diverse bunch semantically. In fact, if we had more time and access to a largecorpus of English text, we would soon find that there are adjective-noun combinations that do notclearly follow any one of the composition rules discussed in this section. However, even in thosecases, linguists have been able to discover the rules that govern the composition of the adjectivewith the noun and make hypotheses regarding what is a possible meaning for an adjective andwhat is not. This project continues to be a lively area of research in semantic theory.1

6.1.2 Quantificational determiners

Sets and operations on sets are used in many parts of semantic theory because the meaningof the basic building blocks of sentences, namely nouns, adjectives, and as we will see in thissection, also verbs, can be modeled as sets of entities. We are not claiming that ALL items inthese categories can be modeled as sets (in fact, we know that this is wrong), only that SOME ofthem can. The previous section showed interesting interactions between adjectives and nouns inadjective-noun combinations. In this section we will focus on interactions between nouns, verbs,and determiners, in sentences like (3) (and the sentence with ALL and SOME above!).

(150) Every vote counts.

In this sentence, vote is a noun, counts is a verb, and every is a determiner. The way we haveset things up, the meaning of vote is the set of all individual votes. Similarly, we will associatethe verb counts with a set of entities, the set which has as members all the things that count.What about the meaning of the determiner?

1A well-known paper that deals with some of these questions is Kamp and Partee (1995): Kamp, Hans, and BarbaraPartee. 1995. Prototype theory and compositionality. Cognition 57: 129-191.

74

In our discussion of parts of speech in previous chapters, we gave the following character-ization of determiners: words that form a closed class, appear before nouns (or adjective-nouncombinations), at most one before each noun. We are now ready to supplement this morpho-logical/syntactic characterization with some generalizations about the meaning, or semantics, ofthese words. To do that, let’s look at the following sentences, in which a determiner (in italics)seems to be relating two sets of entities in the sentence the one denoted by the noun and the onedenoted by the verb. The relationship between the two sets has to do with quantities, so we willcall these determiners quantificational.

Exercise: quantificational relationships

Example:Every girl in the kingdom tried on the shoes.Among girls in the kingdom, the quantity that tried on the shoes equals the number of the girls.

1. Exactly one girl had a small-enough foot.Among , the quantity that equals .

2. No shoe fit her stepsisters.Among , the quantity that equals .

3. Few leather boots were on sale.Among , the quantity that equals .

4. Both stepsisters were upset.Among , the quantity that equals .

Side note: both is special among these determiners because it seems to be a combination oftwo independent quantificational elements: every/all and exactly two. The closest paraphrase ofsentence (v) would be “among the two stepsisters, all were upset”. Can you think of anotherdeterminer that behaves like this? (Hint: look for a negative counterpart to both.)

Let’s express these quantities using sets. We have assumed that the noun (phrase) and verb(phrase) each denote a set. So the question we need to answer, for each quantificational deter-miner, is “what does this determiner tell us about how these two sets are combined?”.

Our noun will be vote and our verb will be counts. The meaning contribution of the determinerevery in every vote counts is that all the votes are included in the set of things that count. In adiagram:

(151) diagram to be distributed in class

In other words, if every vote counts is true, then there cannot be a single vote that is outsideof the set of things that count. The relationship between the two sets in this case is the subsetrelation: the set denoted by the noun (vote) needs to be a subset of the set denoted by the verb(counts).

75

What about no, as in no vote counts? The determiner no tells us that there is not even onesingle vote that is among the things that count. The sets cannot have members in common, theyneed to be disjoint. In a diagram:

(152) diagram to be distributed in class

Some determiners specify conditions on the (non-empty) intersection between the noun-setand the verb-set. For example, exactly one specifies that the intersection should include exactlyone entity, at least one specifies that it should contain one or more entities, and some requires onlythat the intersection be non-empty (without specifying its cardinality).

(153) diagram to be distributed in class

Whatever the specific conditions a determiner imposes on the relation between the two sets, italways results in the same semantic value for the sentence overall: a statement of truth conditions.

. . .is true if and only if:Every A B A is a subset of BNo A B A and B are disjointSome A B A and B have some members in commonExactly one A B A ∩ B has exactly one member in it

Looking back at the goals we set up at the beginning of this chapter (§2.1), we can happilysummarize that set theory has given us tools to achieve both things:

• A compositional analysis of sentence meaning.

The meaning of a sentence like No bus came is derived from putting the meanings of busand came together according to the instructions in the determiner (no).

• A truth-conditional perspective on sentence meaning.

This sentence will come out true “if and only if the intersection between the set of busesand the set of things that came is empty”.

This is quite nice!

76

6.1.3 Sentential connectives

Now that we know something about how to arrive at the truth conditions of a sentence, we canapproach complex sentences—sentences, such as those we saw in the previous chapter on syntax,which contain sentences inside them. As we saw, this is something that speakers do all the time,for example by conjoining two sentences with a coordination like and or or.

(154) diagram to be distributed in class

Since we are committed to the idea of compositionality, we need to discover how the meaningof a conjunction (meaning, a sentence that has the structure shown in (154) is derived from themeaning of the sentences it contains. In other words, we need to find what meaning is associatedwith the coordination and.

The sentence above is true if and only if it is true both that Juno got pregnant and that theperfect couple wanted a baby. The composite sentence, the one labeled “sentence 3”, is true if andonly if both “sentence 1” and “sentence 2” are true. In all other cases i.e. if sentence 1 is true butsentence 2 is false, or vice versa, or if both are false—then sentence 3 is false. It seems reasonable,then, to identify the meaning of and with the logical connective ∧ of propositional logic. Itsmeaning is given by the truth table below. The truth table lists, for each possible combination oftruth values for the “inner sentences” p and q, the resulting truth value for the conjunction p∧q.

(155) Truth table for conjunction:p q p∧q0 0 0

0 1 0

1 0 0

1 1 1

The only row under p∧q which contains the value 1 (for true) is the one in which the valuesfor both p and q are also 1 (true). This seems compatible with our intuitions about the meaningof and. Furthermore, we learn from the table that switching the order of p and q does not resultin any change in the meaning of the conjunction. This also seems right (at least for this example,we will qualify this assumption in the next section), because intuitively the two sentences in (156)are indeed equivalent.

(156) a. Juno got pregnant and the perfect couple wanted a baby.b. The perfect couple wanted a baby and Juno got pregnant.

Are there other logical connectives in addition to ∧ that correspond to natural languageexpressions?

One obvious candidate is ∨, disjunction. It is often identified as the meaning of the coordi-nation or, as in the sentence we will do some shopping for dinner or we will go out. What are yourintuitions about the truth conditions of this sentence?

77

It seems to be clearly false in case both inner sentences are false, for example in a situationwhere we don’t go shopping and skip dinner. This corresponds to the first row in the truth tablebelow. If only one inner sentence is true, the disjunction seems true (second and third rows).What if both sentences are true? In the truth table for (inclusive) disjunction this option alsoresults in truth, so or really means and/or according to this table, not either or.

(157) Truth table for disjunction (inclusive):p q p∧q0 0 0

0 1 1

1 0 1

1 1 1

Whether or not this truth table represents a desirable meaning for or is actually somethingresearchers in semantics are still debating. We will not be able to get into the details of this debatehere, except to note that it poses very interesting questions regarding the relationship betweensemantics and logic as well as the interaction between meaning and intonation (think about theeffect of emphasizing or in a sentence like we will do some shopping OR go out).

6.1.4 Beyond semantic meaning

In the previous sections we laid out the foundations for a compositional, truth-conditional, se-mantics for natural language. In principle, one would hope that this semantics could be extendedso as to derive meanings for all (and only) the grammatical sentences of a language. All onewould need to do is look at more words, discover their basic meaning, and explain how theycombine with the words around them. But would that be enough?

To conclude our discussion of sentence meaning, I would like to give you some reasons todoubt that this may be enough. The evidence for this conclusion will be of two types. First, wewill see a clear case in which the “pure” semantic meaning falls short of what speakers actuallycommunicate with their utterances in context. Second, we will encounter words whose basiccontent is not fixed, but rather changes based on the context in which they are uttered. Bothphenomena make a similar point: words are always interpreted within a context of conversation.The context of conversation systematically influences the content of what is said, and should alsobe part of any complete theory of language meaning.

Case 1: Sometimes and means more than ∧

Recall our example of conjunction (in 156 above, repeated here):

(158) a. Juno got pregnant and the perfect couple wanted a baby.b. The perfect couple wanted a baby and Juno got pregnant.

These two sentences seem equivalent, nothing more than variants of each other.Now, what about the sentences in (159)? Here too there are two sentences conjoined by and,

but suddenly switching the order of the inner sentences makes a big difference in meaning (let’signore the fact that, technically speaking, what comes after and is not a full sentence here).

(159) a. Juno got pregnant and married.

78

b. Juno married and got pregnant.

Question: How do you interpret these sentences? Can you think of a situation in which (159i)would sound natural? Can you think of one for (159ii)? Are (159i) and (159ii) interchangeable inthese situations?

I think they are not interchangeable. In everyday speech we often understand and to meanand then, indicating some temporal sequencing between the conjoined sentences. So (159ii) mostnaturally describes a situation in which the pregnancy followed the marriage, and vice versa for(159i). However, although the temporal sequencing is somehow part of the meaning of thesesentences, it is not captured by the truth conditions delivered by our semantics (and it seems itshouldn’t be). The truth table we have for conjunction takes nothing like temporal sequencinginto consideration.

How is this discrepancy to be explained? What is the source of the temporal sequencingeffect? It is common practice in linguistics to turn to pragmatics for the rescue in such cases. Apragmatic explanation involves taking into consideration additional factors on top of the wordsin the sentence, such as the identity of the participants in the conversation (speaker and hearers),their goals, and basic principles of cooperative communication. Think of it as a process of fine-tuning the semantic meaning to fit a particular context of conversation. We will think about whatthis process might be when we discuss these examples in class.

Case 2: Who is “he”?

Another example of context influencing the meaning of utterances comes from the use of pro-nouns. These are words like I, you, they (also me, my, myself etc.) which do not, out of context,refer to any one entity in particular.

Let’s compare pronouns with other kinds of noun (phrases). A proper name like Canada orJones always refers to the same entity (keeping the universe of entities constant). A descriptionlike the girl sitting next to the door may refer to different girls in different contexts, but the wordsthemselves are enough to guide us to a particular entity in a given context. Pronouns are differ-ent: in a sentence like He likes expensive cars, we only know the meaning of he if we know whohe refers to. If we’re talking about Jim, the guy Sandra met last weekend, then we know that hemeans Jim. Jim is the referent of he. But he could in principle refer to any male person in thecontext.

Pronouns are actually quite tricky. We have just seen a case in which nothing in the sentencecould help us figure out who he refers to; the interpretation of the pronoun was influenced byextra-linguistic, contextual factors. But look at sentence (160): is the pronoun (him) also free inits interpretation here?

(160) John loves him.

Not quite. In fact, the pronoun cannot be used in this sentence to refer to John. If it were, wewould have to say (161):

(161) John loves himself.

So, there seem to be both extra-linguistic and purely linguistic (in the example above, syntac-tic) pressures on the interpretation of pronouns.

79

6.2 For further reading...

Heim, Irene and Angelika Kratzer. 1998. Semantics in generative grammar. Blackwell.

Lyons, John. 1995. Linguistic semantics: an introduction. Cambridge: Cambridge University Press.

80