Introductory Linguistics Study Guide, Slides of Linguistics

Download Introductory Linguistics Study Guide and more Slides Linguistics in PDF only on Docsity! Honors Project Cover Page Introductory Linguistics Study Guide By Anna Gartsman and Laura Hughes, Class of 2007 May 2007 Department of Linguistics College of Arts & Sciences Northeastern University Advisor: Janet Randall Project field: Linguistics Colleges/Majors: • Anna Gartsman: o School of Computer Science: Computer Science and Cognitive Psychology o School of Arts & Sciences: Linguistics • Laura Hughes o School of Arts & Sciences: Linguistics and Psychology Abstract: A step-by-step guide to the major areas of introductory-level theoretical linguistics (phonetics, phonology, morphology, syntax, semantics, and pragmatics) with a focus on practical problem-solving. Designed to assist students in low-level linguistics classes and tutors of same by providing additional explanations, examples and exercises. Introduction to Linguistics Study Guide Anna Gartsman & Laura Hughes Northeastern University Senior Honors Project 2006-2007 The International Phonetic Alphabet (IPA) Objective: Transcribe English speech sounds in the international phonetic alphabet. What is the international phonetic alphabet (IPA), and why do we need one? The International Phonetic Alphabet is a way of transcribing speech as it’s pronounced, which means it can be used to transcribe any language (so long as there are symbols for the sounds spoken in that language). Because there is a one-to-one correspondence between symbols and sounds in IPA, there are no problems with weird, ambiguous spelling, silent letters, etc. such as we often find in written English. In phonetics and phonology, we use IPA to represent data sets of sounds from different languages. Using words written in IPA, we can focus on the sounds of a language without having to learn the writing systems of different languages—many of which don’t have any writing system at all, or have one that isn’t based on sounds (such as Chinese). You can use your textbook or handouts or the charts provided here to learn the IPA symbols used for English consonants and vowels. Since many languages have sounds we don’t use in English, you may be introduced to more symbols in later problem sets; but these symbols are all you need to start transcribing English speech. Transcription Tips: • Sounds Not Spelling When transcribing words into the IPA, focus on the sounds, not the way that they’re spelled. Say the word aloud. • Normal (Fast) Speech Be careful not to say the word too slowly and carefully, because that may change some of the sounds. The idea is to transcribe the way you usually pronounce the word in normal speech. • IPA Symbols Are Not the Same As Letters An IPA symbol may look like an English letter, but represent a different sound than that letter normally does. This is especially true of vowels. Note, for example, that [e] is the vowel sound in “say” or “weigh”, not in “bed” (that would be []). • Ignore Silent Letters Many English words have silent letters, the “e” at the end of “cape,” for example. Remember, the difference between “cap” and “cape” doesn’t have to do with the “e”—it’s a different vowel between [k] and [p] (that is, [kæp] vs. [kep]). • Your Pronunciation May Vary Even within English, people with different dialects may pronounce words differently. (Again, this is especially true of vowels!) There may be more than one way to transcribe a word, but there is only one way to transcribe the word the way you say it. • Use Your Friends If you’re worried that your accent or dialect is too non-standard, or if you can’t tell what sound you’re saying, ask someone else to say the word. If you’re worried about “priming” them to pronounce the word the way you do, write down the word and have them read it to you. The charts on the following pages give examples of the English consonants and vowels. We transcribed the words into IPA using our own dialect of Standard American English; your mileage may vary. (The symbols in parentheses show alternative symbols for the same sound.) 4 Phonetic Alphabet Chart: Consonants of English IPA Symbol Example (Standard Orthography) Example (IPA) p powers, superhero, cape pawrz, suprhiro, kep b Batman, Robin, lab bæmæn, rabn, læb t toxic, Green Lantern, invisible jet taksk, rin læntrn, nvzbl dt d Doctor Doom, incredible, Alfred daktr dum, nkrdbl, ælfrd k costume, Doctor Octopus, Hulk kastjum, daktr aktps, hlk  Gotham, Magneto, Rogue am, mænio, ro f fortress, Alpha Flight, tough fortrs, ælf flajt, tf v villain, Professor Xavier, Batcave vln, prfsr zevir, bækev  threat, Lex Luthor, stealth rt, lks lur, stl  The Hulk, weather, scythe  hlk, wr, saj s Superman, lasso, spandex suprmæn, læso, spændks z zonk, laser, disguise zak, lezr, dskajz  ( š ) Shadowcat, radiation, Flash ædokæt, redien, flæ  ( ž ) treasure, Mirage trr, mra h Hulk, superhero hlk, suprhiro t ( č ) champion, Watchmen, launch tæmpiyn, watmn, lnt d ( ǰ ) justice, origin, judge dsts, rdn, dd m Magneto, Aquaman, crime mænio, akwmæn, krajm n Nightcrawler, spandex, Robin najkralr, spændks, rabn  super strength, Batarang supr strk, bæræ l Lois Lane, Alfred, Smallville lows len, ælfrd, smalvl r Rogue, Alfred, Nightcrawler ro, ælfrd, najkralr w Wonder Woman, Nightwing wndr wmn, najtw j ( y ) united, slayer junajd, slejr  uh-oh, Batman o, bæmæn  Magneto mænio Weird Consonants to Notice  Glottal Stop: Voiceless stop that’s rare in English. It may almost seem like a short pause instead of a sound. “uh-oh” is the best example, but it can also replace “t” in certain words (like “mitten”) in some dialects.  Lateral Flap: A quick flap of the tongue. Comes out sort of like a cross between a “t” and “d” sound, as in “butter”. j Palatal Glide: Be careful with word-medial [j]. Compare “layer” [lejr] with “lair” [ler]. 5 Phonetic Alphabet Chart: Vowels of English IPA Symbol Example (Standard Orthography) Example (IPA) i superhero suprhiro  invisibility nvzbli e cape kep  incredible nkrdbl æ Batman bæmæn  Buffy bfi  ability bli a Robin rabn u Superman suprmæn  Doctor Octopus daktr aktps o Rogue ro  law l aj ( a ) kryptonite krptnajt aw ( a ) powers pawrz j (  ) Superboy suprbj Weird Vowels to Notice aj, aw, j Diphthongs: Although written with two symbols, these count as one sound.  Schwa: Always unstressed. Usually comes out sounding like [] as in “cut”. Found in longer words when another vowel is stressed.  Low Back Round Vowel: Not everyone has this sound in their dialect. Do you have a difference between “caught” and “cot”? If so, then “caught” would be [kt]. If not, they are both [kat]. 6 Natural Classes Objective: Define a natural class. Given a natural class, identify the sounds that belong to it. Given a set of consonants or vowels, identify what class, if any, they belong to. What is a natural class? A natural class is a way of describing a set of speech sounds that have something in common. In order for a set of sounds to belong to a natural class, there must be some way of describing them that: • includes all the sounds in the set • does not include any other sounds Examples • The set of sounds {p, t, k} is in a natural class of “voiceless stops”. • The set of sounds {n, , } do not comprise a natural class because there is no natural class that will include all these sounds and only these sounds. One could say that they are voiced stops, and this would be correct, except that would also include /d/, /b/, and /m/, which are not part of the set. • The set of sounds {p, b, m, n} do not comprise a natural class because while /p/, /b/, and /m/ are all bilabial stops, /n/ is not, and the natural class needs to include all the sounds in the given set. Tips: • Look at the charts (next page). When analyzing whether a set of consonants comprises a natural class, a good place to start is to look at the IPA chart and see if the sounds all fit into a particular row (manner of articulation) or column (place of articulation). For example, {m, n, } are all nasals, and {k, , } are all velar sounds. • Don’t forget about voicing. Voiceless sounds are shaded in the consonant chart provided. All vowels are voiced. • Look at combinations. If there is no clear natural class in the manner or place of articulation, try looking at a combination of the two. For example, {f,v} are labiodental fricatives – “labiodental” has to be mentioned in order to exclude other fricatives, such as // and //. • Each sound is a natural class. For example, {p} is the only member of the natural class voiceless bilabial stops. • Vowels have a different set of descriptors than consonants. Vowels are distinguished from each other by frontness (front, central, or back), height (high, mid, or low), tenseness (tense or lax), and roundness (round or non-round). Check the chart for details. All vowels are voiced. 9 IPA English Consonant Chart: Manner of Articulation, Place of Articulation, Voicing* Place of Articulation Bilabial Labio- dental Inter- dental Alveolar Palatal Velar Glottal p t k  Stop b d  Nasal m n  f  s  h Fricative v  z  t Affricate d Approximant w r j M an ne r of A rt ic ul at io n Lateral Approximant l * Shaded sounds are voiceless IPA English Vowel Chart: Height, Frontness, Tense/Lax*, Roundness** Frontness Front Central Back i u High   e o Mid  ,   H ei gh t Low æ a * Shaded sounds are tense **Bold sounds are round 10 Natural Class Exercises I. List all the sounds in each class. a. Voiceless b. Fricative c. Velar d. Alveolar stop e. Velar nasal f. Mid g. Round h. Low back non-round i. High tense II. Determine whether the following sets of sounds are in a natural class. If so, what is the class? a. {f, , s, , h} b. {z, t, d} c. {p, b, m, w} d. {b, d, , } e. {u, , o} f. {æ, a, } g. {i, , } h. {, , } 11 14 Distinctive Feature Chart (Consonants) Feature p b t d k   m n  f v   s z   h t d r l w j consonantal + + + + + + + + + + + + + + + + + + + + + + + - - sonorant - - - - - - - + + + - - - - - - - - - - - + + + + syllabic - - - - - - - - - - - - - - - - - - - - - +/- +/- +/- +/- voice - + - + - + - + + + - + - + - + - + - - + + + + + continuant - - - - - - - - - - + + + + + + + + + - - + + + + nasal - - - - - - - + + + - - - - - - - - - - - - - - - sibilant - - - - - - - - - - - - - - + + + + - + + - - - - lateral - - - - - - - - - - - - - - - - - - - - - - + - - labial + + - - - - - + - - + + - - - - - - - - - - - + - alveolar - - + + - - - - + - - - - - + + - - - - - + + - - palatal - - - - - - - - - - - - - - - - + + - + + - - - + velar - - - - + + - - - + - - - - - - - - - - - - - - - anterior + + + + - - - + + - + + + + + + - - - - - + + + - coronal - - + + - - - - + - - - + + + + + + - + + + + - - Distinctive Feature Chart (Consonants) Feature i  e  æ ,  u  o  a high + + - - - - + + - - - low - - - - + - - - - - + back - - - - - + + + + + + rounded - - - - - - + + + + - tense + - + - - - + - + - - Example Problems: 1. What sound(s) is/are described by this list of features? a. {+labial, +nasal} To find this sound, locate all the sounds that are {+labial}. Then, find out which {+labial} sounds are also {+nasal}. There is only sound that is both {+labial} and {+nasal}: [m]. b. {+palatal, +continuant} As above, locate all sounds that are {+palatal}. This list includes: {, , t, d, j}. Of these, which sounds are {+continuant}? There are several: {, , j}. { t, d} are excluded because they are not continuant. So, the set of sounds described by the features {+palatal, +continuant} is {, , j}. c. {+palatal, +continuant, -consonantal} As in (b), locate the {+palatal, +continuant} sounds: {, , j}. Of these sounds, which are {-consonantal}? {, } are {+consonantal}; only [j] is {-consonantal}. So, the {+palatal, +continuant, -consonantal} sound is [j]. d. {+low, -back} There are only two low vowels, {æ, a}. Only one, [a], is {–back}. 2. What list of features describe(s) this/these sound(s)? Make sure there are no redundant features, and the sound is described uniquely (no other sound is described by this list of features). a. [f] The place feature is a good place to start; [f] is {+labial}. We need to distinguish it from the other {+labial} sounds [p], [b], [m], [v], and [w]. Adding {+continuant} gets rid of the stops [p], [b], and [m], so we only need to deal with [v] and [w]. Both are {+voice} while [f] is {-voice}, so there’s our third feature. Our complete list is {+labial, +continuant, -voice}. b. [r] [r] is {+alveolar}. We need to distinguish it from the other alveolar sounds, [t], [d], [n], [s], [z] and [l]. A nice way to get rid of everything except [l] is to add {+sonorant}. Now we just have to distinguish between [r] and [l]. The only difference is that [l] is {+lateral}, so we add that [r] is {-lateral}. The complete list is {+alveolar, +sonorant, -lateral}. Another way we could have gone instead of using {+sonorant} is to get of [t], [d] and [n] by adding {+continuant} to our list of features. Now we need to deal with [s], [z], and [l]. [s] and [z] are {+sibilant}, so we add {-sibilant} to our list. Now we still need to distinguish between [r] and [l], so we add{-lateral}. Our complete list is {+alveolar, +continuant, -sibilant, -lateral}. This is also correct, but less efficient, because it requires four features instead of three. 15 c. {k, } [k] and [] are both {+velar}; the only other {+velar} sound is []. To get rid of [] we just have to say {-nasal}. Hooray, the features are {+velar, -nasal}. d. {u, } You can start by saying that they are {+high}. Now you only need to differential them from {i, I}. Either {+back} or {+rounded} will do this. Since there are no high, back, non-round vowels and no high, rounded, non-back vowels, either way will work equally well. So there are two correct answers: {+high, +back} or {+high, +round}. 3. What are the distinctive features that differentiate these two sounds? a. {, } This is one of those many pairs where the only difference is that [] is {+voice}, and [] is {-voice}. So the feature involved here is {+/- voice}. b. {s, } The difference between these is place; [s] is alveolar and [] is palatal. There are features for both, so you can either say that the difference is {+/- alveolar} or {+/- palatal}. c. {, } If you look at the feature chart, you can see that the only difference between these two is that [] is {+round}. So {+/- round} is the relevant feature. 4. What are the distinctive features that differentiate these two sets of sounds? a. {t, d} & {s, z} To show the difference between these two sets, we need to describe each set individually. {t, d} are {-continuant, +alveolar, -nasal}. {s, z} are {+continuant, +alveolar, -sonorant}. The distinctive feature is {+/- continuant}: both sets are +alveolar, and even though it is unnecessary to state this for {t, d}, both sets are also {-sonorant}. ` b. {i, } & {u, } {i, } are {-back, +high}. {u, } are {+back, +high}. The distinctive feature is {+/- back}, since both sets of vowels are {+high}. 16 Phonology A phoneme is a distinct speech sound in a language. Lessons Minimal Pairs Overview Example Problems Exercises Phonemes and Allophones Definitions Explanation Summary Complementary & Contrastive Distribution Definitions Explanation Summary Environments Overview Step-by-Step Example Example Problems Exercises Determining the Underlying Phoneme Overview Example Problems Exercises Rules Overview Example Problems Reference: Rule-writing Conventions Exercises Analyzing Data Sets Overview Example Problems Exercises Phonological Processes Overview Example Problems Exercises 19 Minimal Pairs Objective: To identify minimal pairs. Minimal pairs are pairs of words with different meanings and exactly one sound difference. For example, “cat” and “bat” are minimal pairs because only the first sound is different ([k] vs [b]). However, “cat” and “flat” are not minimal pairs, because there are two sound differences: ([k] vs [f] and [l].) The reason we look for minimal pairs is to identify a contrast between two sounds. If two different sounds, placed in the same exact environment, produce different words with different meanings, then those sounds really are different phonemes. See the “Phonemes vs. Allophones” lesson in this section for more information. Remember, the restrictions on minimal pairs are: • The two words have different meanings • Only one sound is different • The words have the same number of sounds • The sound that is different is in the same place in both words Tips • If the words are in English, they will probably be given to you in standard orthography. Remember, DO NOT RELY ON SPELLING! Write the words in IPA. Then you just have to compare the symbols. For example, in standard orthography it looks like “rewind” and “resigned” must have more than one sound change, but if you write them in IPA (rwajnd, rzajnd) you will see that they are actually minimal pairs (Again, your dialect may vary.) • If the words are in another language, you will be given the transcription and a gloss (what the word means). So, you don’t have to worry about trying to do a transcription into IPA. Example Problems 1. Are the following pairs of English words minimal pairs? a. raven, craven IPA transcriptions: revn, krevn These are not minimal pairs—they don’t have the same number of sounds. The idea is to compare one sound with another, but there is no sound to compare [k] to, because it corresponds to no sound in the word “raven.” b. cab, cash IPA transcriptions: kæb, kæ These are minimal pairs. You’re comparing [b] and []. 20 2. Consider this data set from another language, and determine whether the given pairs are minimal pairs or not. a. Tagalog: kahon “box” kaon “to fetch” These two words have different meanings and one sound change (between “h” and “”). They are a minimal pair. b. Inuktitut: ilu (snow)house il (snow)house These two words have the same meaning, so even though there is a sound change, they do not constitute a minimal pair. Exercises 1. Are the following pairs of English words minimal pairs? a. law, jaw b. crime, time c. prison, prism 2. Consider this data set from Thai, and determine whether the given pairs are minimal pairs or not. a. ba (“sheet”), pa (“to go”) b. bryy (“extremely fast”) , myy (“hand”) c. pa (“to go”), pa (“danger”) 21 Complementary and Contrastive Distribution Objective: Distinguish between contrastive and complementary distribution. Know which one indicates phonemes, and which indicates allophones. Definitions Environment: For these simple examples, the environment is simply the sounds before and after the ones we are looking at. Contrastive Distribution: Sounds that are in contrastive distribution can be found in the same environment. The sounds constrast and therefore they are different phonemes. Complementary Distribution: Sounds that are in complementary distribution are always found in different environments. The two sounds complement each other—that is, between them, they cover all possible environments. They are allophones of the same phoneme. Explanation We know that different phonemes can appear in the same surrounding word and create different meanings (minimal pairs). We also know that when there are different allophone variants of a phoneme, the environment determines which one appears (for example, in standard English, [pH] is always pronounced at the beginning of a word, never [p]). We can use this knowledge to determine whether the difference between a pair of sounds is phonemic or allophonic in a given language. If the two sounds can appear in the same environment (that is, with the same surrounding sounds), then they are in constrastive distribution and they are different phonemes. On the other hand, if the two sounds always appear in different environments—if there’s a systematic way to tell when to use one sound and when to use the other—then they’re in complementary distribution and they’re allophones of the same phoneme. Summary • Phonemes are found in contrastive distribution (same environment) • Allophones are found in complementary distribution (different environments) 24 Environments Objective: Make charts showing the environments to compare two sounds, and make a decision about complementary vs. contrastive distribution. If complementary, decide which is the underlying phoneme. Overview In order to determine if a pair of sounds represents different phonemes or different allophones, we need to know whether they are in contrastive or complementary distribution. And in order to determine that distribution, we need to carefully look at the environments in which the sounds are found. If the sounds can be found in the same environment—if both sounds could come before and after the same surrounding sounds—then they’re in contrastive distribution and they’re phonemes. But if we can find some systematic difference in the environments, then we’re looking at complementary distribution (allophones). If there are minimal pairs that contrast the sounds you’re looking at, you know right away that the sounds are different phonemes. If there aren’t, though, you need to carefully look at the sounds before and after. One way to clarify the problem is to draw charts of the immediate environments of each sound. Representing Environment The environment of a sound means the sounds that come immediately before and after it. By convention, we write this as [preceding sound]_[following sound]. For example, the environment of [k] in the word [bækt] would be written s_. Environment-Writing Tips: • Word boundaries—the beginning or end of a word—are represented by the symbol #. The environment of the [k] in [luk] (the name “Luke”) would be written u_#. • A diphthong counts as one sound, so the environment for [k] in [skaj] is s_aj. • Notice that a single word can have the letter of interest multiple times. For example, [luk skajwkr] has three separate environments for [k]: u_# s_aj _ Environment Charts: Step-by-Step 1. Make a list of the environments for each sound of interest. You will typically be comparing two sounds found in a data set. For each sound, go through the data set, writing down every environment where that sound is found. Keep the lists separate. 2. Compare each side of one list to the corresponding side of the other list, looking for overlap. Overlap means the same sounds appear in the same position (before or after the sound of interest) on both lists. If there is overlap on both sides, we can conclude the sounds are in contrastive distribution because they could, theoretically, appear in the exact same environment. There’s no difference between the environments in which one sound occurs vs. the other sound, so the sounds themselves contrast. In other words, they’re two different phonemes. 25 If there is no overlap on one or both of the sides (i.e. there is no over lap on either side OR either the right side or left side has no overlap, although the other may certainly have overlap), we can conclude the sounds are in complementary distribution, because they occur in systematically different environments. In other words, they are two allophone versions of the same underlying phoneme. Later sections will show how to tell which is the underlying phoneme and what rule governs the use of the allophone. Example Problems 1. Consider this data from Tongan, a Polynesian language. Are [s] and [t] different phonemes, or allophones of the same phoneme? tauhi “to take care” sisi “garland” motu “island” mosimosi “to drizzle” motomoto “unripe” fesi “to break” sino “body” totonu “correct” pasi “to clap” fata “shelf” movete “to come apart” misi “to dream”ii First off, we can check for minimal pairs, since that’s the easiest way to tell that the two sounds are phonemes. There are none, so we can’t conclude anything yet. We need to figure out if the sounds are in complementary or contrastive distribution. Step 1. Make a list of environments for each sound of interest. Our chart looks like this: s #_i i_i o_i o_i e_i #_i a_i i_i t #_a o_u o_o o_o #_o o_o a_a e_e Note that since we care about variability, there’s really no point in listing all the repeated environments in each list. So we could write the list like this: s #_i i_i o_i e_i a_i t #_a o_u o_o #_o a_a e_e 26 29 Right-hand side comparison t a u o o a e m o o o i There’s overlap on this side, too—[o] occurs on both lists. Since we see overlap on both sides, the sounds are in contrastive distribution, and we can conclude that [t] and [m] are different phonemes in this language. Exercises 1. Consider the following data from English (use the IPA transcriptions, not the standard orthography glosses.) According to this data, are [] and [] allophones of the same phoneme, or different phonemes? bo “both” oz “those” rær “rather” n “thin” farz “fathers” k “think” ej “they” lo “loathe” s “this” mæ “math” 2. Consider the following data from English. According to this data, are [n] and [] allophones of the same phoneme, or different phonemes? mntl “mental” nsttun “institution” lr “linger” hændl “handle” æw “anguish” træk “tranq” jn “yawn” 3. Consider the following data from Ganda. According to this data, are [l] and [r] allophones of the same phoneme, or different phonemes?iv kola ‘do’ lwana ‘fight’ buulira ‘tell’ lya ‘eat’ luula ‘sit’ omuole ‘bride’ lumonde ‘sweet potato’ eddwaliro ‘hospital’ oluanda ‘Ganda language’ olulimi ‘tongue’ wulira ‘hear’ beera ‘help’ jjukira ‘remember’ eryato ‘canoe’ omuliro ‘fire’ effirimbi ‘whistle’ emmeeri ‘ship’ eraddu ‘lightning’ wawaabira ‘accuse’ laira ‘command’ Going Beyond In English, are [n] and [] allophones of the same phoneme, or different phonemes? Give data to support your answer. Does your answer support or conflict with your answer to exercise #2? If there is a conflict, how might you explain it? 30 Determining the Underlying Phoneme Objective: Given two sounds in complementary distribution and their environments, determine which sound is the underlying phoneme. Overview The previous section, “Environments,” described how to decide if two sounds are in contrastive (phonemic) or complementary (allophonic) distribution. If we find that two sounds are different phonemes, then that’s about all we can say about them, but if they are two allophones of the same phoneme, there are still some questions to resolve. One important question is: which is the underlying phoneme? The basic answer to that question is the sound that is in wider distribution. Wider distribution means the sound could occur in more different environments. To determine distribution, we look at the environment lists—specifically, the side with no overlap. Which list is more variable? Example Problems 1. This is the right-hand (no-overlap) side of the environments in example #1 (Tongan data) of the previous section (“Environments”). Which sound is in wider distribution? s i i i i i t a u o o a e This is a very clear-cut case. [t] can occur before all sorts of vowels, and [s] can only occur before [i]. Therefore, [t] is in wider distribution and is the underlying phoneme. 2. Compare the following environments. Which sound is in wider distribution, [s] or [z]? s #_ f_ t_ h_ p_ z _ a_ _ i_ o_ m_ In this case, even though [z] looks like it appears with more sounds, all of these sounds can be combined in one category, namely “voiced” sounds (remember that all vowels are voiced). There is no way to group the environment of the [s] sound (because there is no way to group the word 31 Example Problems 1. In the “Environments” section, example problem #1, we looked at data from Tongan, and decided that [s] and [t] were in complementary distribution. Here are the environments: s #_i i_i o_i e_i a_i t #_a o_u o_o #_o a_a e_e Write a rule in symbol notation which describes when each allophone occurs. Remember that in “Determining the Underlying Phoneme,” example problem #1, we decided that [t] was the underlying phoneme since it occurs in more variable environments (before [a], [u], [o], or [e], while [s] only occurs before [i]). Since [t] is the underlying phoneme, the rule will look like this: /t/ [s] / some environment What environment do we want to put there? Well, we want to say when [s] occurs. Looking at the data, it’s pretty clear that [s] only occurs before [i]. So that’s what we write. /t/ [s] / _i How would we write the rule for this example in feature notation? We’ll have to consult our old phonetics charts to figure out the relevant features of [t], [s], and [i]. /t/ : {+alveolar, -voice, -continuant} /s/ : {+alveolar, -voice, -continuant} /i/ : {+high, -back, +tense} Note that when we write the rule, we only have to include those features of allophone [s] that are different from the features of the underlying phoneme /t/. {+alveolar, -voice, -continuant} {-continuant} / _{+high, -back, +tense} 2. The English plural “-s” can be pronounced as either [s] or [z] depending on the context. Here’s some data: hobbits habts elves lvz orcs rks humans hjumnz trolls trolz wraiths res ponies poniz 34 Write the rule (in feature notation) for when [s] appears and when [z] appears. First, we write the environments as per the previous section. s t_# k_# _# z v_# n_# l_# i_# We’re concerned with the sound on the left (obviously, since the right side of both lists is composed entirely of word boundaries.) Because [z] occurs with both vowels and consonants and [s] occurs only with consonants, [z] appears to be in wider distribution. Therefore we could write the rule in symbol notation as follows: /z/ [s] / {t, k, }_ Now it’s just a matter of converting the rule to feature notation. First of all, what feature, if any, is common to {t, k, } but not {v, n, l, i}? Consulting our chart, we see that {t, k, } are all unvoiced. Interestingly enough, voicing also accounts for the difference between [z] and [s]. {+alveolar, +voice, +continuant} {-voice} / {-voice}_ Exercises Consider the exercises in the previous section (“Determining the Underlying Phoneme”). Can you write the rule for these exercises? Remember to write in both symbol and feature notation. 1. Consider the following data from English. According to this data, what is the rule that predicts the occurrence of the [n] vs. the [] sound? mntl “mental” nsttun “institution” lr “linger” hændl “handle” æw “anguish” træk “tranq” jn “yawn” 35 2. Consider the following data from Ganda. According to this data, what is the rule that predicts the occurrence of the [r] vs. the [l] sound? kola ‘do’ lwana ‘fight’ buulira ‘tell’ lya ‘eat’ luula ‘sit’ omuole ‘bride’ lumonde ‘sweet potato’ eddwaliro ‘hospital’ oluanda ‘Ganda language’ olulimi ‘tongue’ wulira ‘hear’ beera ‘help’ jjukira ‘remember’ eryato ‘canoe’ omuliro ‘fire’ effirimbi ‘whistle’ emmeeri ‘ship’ eraddu ‘lightning’ wawaabira ‘accuse’ laira ‘command’ Going Beyond 1. In example problem 2, when we developed a rule for plural pronunciation, we decided [z] was the basic underlying phoneme. Given what the rule turned out to be, what’s an alternate explanation for why [z] is in wider distribution than [s]? Could [z] and [s] actually be equally ranked (neither one more “underlying” than the other, or you can’t tell)? 36 ALLOPHONES ONLY: Step 4: Write the rule We need to fill in a rule of the form /underlying phoneme/ [allophone version] / environment Step 4a. Decide which allophone is the underlying phoneme. The allophone that is in wider distribution is the underlying phoneme. To determine distribution, we look at the side of the environments where we found no overlap. Sounds following [s] Sounds following [t] [i] [a] [o] [e] [u] # Right away, we can see that [t] is in wider distribution because while [s] only comes before certain vowels, [t] comes before certain vowels and word boundaries. Step 4b. Write the rule. We already have all the information we need to write the rule in symbol notation. Symbol Notation /t/ [s] / _ [i, e] “/t/ becomes [s] before [i] or [e]” For many rules, especially those with multiple sounds, it will make more sense to write at least part of the rule in feature notation. What’s another way we can say “[i] or [e]”? Consult your vowel chart. What do they have in common (that [a], [o], and [u] do not have in common)? Partial feature notation /t/ [s] / _ [-back, -low] “/t/ becomes [s] before non-back, non-low vowels” We can also put the whole rule into feature notation by describing [t] and [s] in terms of their features. First we must uniquely describe [t], then we can describe [s] in terms only of the feature(s) that make it distinct from [t]. (Note that because we are working with another language, there may be additional sounds/features to take into account. These will usually be provided for you if they are relevant. For the purposes of this problem, we will assume the features we developed for English are sufficient.) Full Feature Notation /+alveolar –voice –continuant/ [+continuant] / _ [-back, -low] “voiceless alveolar stops become continuant before non-back, non-low vowels” 39 Another Example Question Consider the following data set from Tongan. Are [m] and [t] different phonemes, or allophones of the same phoneme (if they are allophones, provide the rule)? tauhi “to take care” sisi “garland” motu “island” mosimosi “to drizzle” motomoto “unripe” fesi “to break” sino “body” totonu “correct” pasi “to clap” fata “shelf” movete “to come apart” misi “to dream”vi Following the step-by-step guide: Step 1: Write the environments. t #_a o_u o_o #_o a_a e_e m #_o i_o o_o #_i Step 2: Compare environments for overlap. When we look at the left side of each sound, we notice that there is overlap – they share [#, o]. When we compare the right side of each sound, we notice that there is overlap there too – they share [o]. Step 3: Decide on phonemes or allophones. Since there is overlap on both sides (contrasting distribution), this part is easy – these two sounds are distinct phonemes in this language. Exercises 1. Consider the following data from English. According to this data, are [n] and [] phonemes or allophones (if allophones, provide the rule in symbol and feature notations): mntl “mental” nsttun “institution” lr “linger” hændl “handle” æw “anguish” træk “tranq” jn “yawn” 40 2. Consider the following data from English (use the IPA transcriptions, not the standard orthography glosses.) According to this data, are [] and [] allophones of the same phoneme, or different phonemes (if allophones, provide the rule in symbol and feature notations)? bo “both” oz “those” rær “rather” n “thin” farz “fathers” k “think” ej “they” lo “loathe” s “this” mæ “math” 3. Consider the following data from Ganda. According to this data, are [r] and [l] phonemes or allophones (if allophones, provide the rule in symbol and feature notations): kola ‘do’ lwana ‘fight’ buulira ‘tell’ lya ‘eat’ luula ‘sit’ omuole ‘bride’ lumonde ‘sweet potato’ eddwaliro ‘hospital’ oluanda ‘Ganda language’ olulimi ‘tongue’ wulira ‘hear’ beera ‘help’ jjukira ‘remember’ eryato ‘canoe’ omuliro ‘fire’ effirimbi ‘whistle’ emmeeri ‘ship’ eraddu ‘lightning’ wawaabira ‘accuse’ laira ‘command’ 4. Consider this data from Tongan, a Polynesian language. Are [s] and [t] different phonemes or allophones of the same phoneme (if allophones, provide the rule in symbol and feature notations)? tauhi “to take care” sisi “garland” motu “island” mosimosi “to drizzle” motomoto “unripe” fesi “to break” sino “body” totonu “correct” pasi “to clap” fata “shelf” movete “to come apart” misi “to dream”vii 41 You may notice that some consonants are dropped (some may not be, depending on dialect). So, a lot of people pronounce these words in the following manner: underlying representation surface pronunciation “fast task” [fæst tæsk] [fæstæsk] “camp fire” [kæmp fajr] [kæmfajr] “first grade” [frst red] [frsred] “second semester” [sknd smstr] [sknsmstr] What happens here is that the medial consonant in a cluster ([t] or [d] in this case) gets deleted in surface pronunciation. This is the process of deletion. The rule for this would be: /t, d/ Ø / C _ C 2. Consider the following words in Spanish: escuela “school” esperar “to wait” esfera “sphere” estructura “structure” And the following loanwords from English: estandar “standard” esmoquin “smoking jacket” What happens to the English words when they are borrowed into Spanish? It looks as if they are made to comply with the Spanish standard rules of pronunciation (which require an initial “e” [] before a sC cluster). So, an “e” is inserted to comply with Spanish pronunciation. This is the process of epenthesis. The rule for this would be: Ø []/ #_sC 3. Consider the following words from English (pay attention to the vowels): jab jam sag sang dad Dan You may notice that the vowel [æ] in the first column of words is slightly different from the same vowel [æ] in the second column. The [æ] in the second column is nasalized. The reason for the nasalization becomes clear when we look at the environments. The environment on the left side is the same, so the difference must be on the right. In the first column, the sounds on the right are all voiced oral stops; in the second column, the sounds are all nasal stops. So, the vowel assimilates to its environment (becomes nasalized when it’s next to a nasal sound): /æ/ [+ nasal] / _ [+nasal] C 44 4. Consider the following data from Kitharakaix ( is a voiced velar fricative): /ka + pandi/ [apandi] a small grasshopper /ke + tue/ [etue] a pole /ka + cuma/ [acuma] a small piece of iron /ke + kundi/ [ekundi] a group What happens when a [ke] or [ka] prefix is added to a word that starts with a voiceless sound? The [k] in the prefix becomes voiced, so that it is different from its environment. Notice here that the vowel between the two consonants does not seem to play a role. This is sometimes the case with phonological data sets – sometimes, the environment that triggers a change is not the immediate environment. So, in this case of dissimilation, the rule is: /k/ [+ voice] / _ V [-voice]C Here, it is the voicing that is important, so we ignore, in partial feature notation, the change from –continuant to +continuant that also occurs. Exercises 1. Consider the following data from a dialect of English sometimes spoken in Eastern Massachusetts (“:” next to a vowel indicates a long vowel); what phonological process could account for the transformation from underlying to surface form? Provide the rule: underlying form surface form orthography [s:] [s:r] sawing [ad] [ad] idea [wtn] [wtn] Washington [kl] [kl] cloth 2. Consider the following data from English; what differences in pronunciation do you find among the basic word and the word with an attached suffix? Pay specific attention to the last sound in the root word. What phonological process could account for the transformation? Provide the rule: Root Root + morphology hand handed long longer limb limber 45 3. Consider the following data from English; keep in mind that all of these words have the same underlying prefix: “con-”. What phonological process could account for the transformation from underlying to surface form of the prefix “con-”? Provide the rule: conjoin commerce contraction compose contemplate comply conspire combine configure conclude 4. Consider the following data from English; keep in mind that all of these words have the same underlying suffix: “-al”. What phonological process could account for the transformation from underlying to surface form of the suffix “-al”? Provide the rule: Noun Adjective Noun Adjective person personal pole polar region regional circle circular autumn autumnal single singular cause causal uvula uvular 46 Morphology Terms and Definitions Objectives: To separate English words into the morphemes that compose them. To identify the root and the affixes. To identify whether an English morpheme is free or bound, and if bound, whether inflectional or derivational. Morpheme A morpheme is the smallest unit in language which holds meaning. • A basic word like “cat” is a morpheme – the individual parts of the word (the sounds /k/, /æ/, /t/) don’t have any meaning of their own. • A word like “unsinkable” can be broken down into three morphemes: “un-“ which means “not”; “sink” which means, well, “sink”; and “-able” which means “able to be...” Types of Morphemes (Free and Bound) Free morphemes can stand alone as their own word. English Examples: cat, apple, sing, wonder, green, happy, of, above Bound morphemes cannot stand alone as their own word but, instead, join to other words. English Examples: un-, -ing, -able, re-, de-, -ist, -er, -s (plural) Types of Bound Morphemes in English (Inflectional and Derivational) Inflectional morphemes denote number, gender, tense, or case. (They don’t really change the meaning of the root word, except to make it plural, past tense, etc.) English Inflectional Morphemes (complete set) Deri vational morphemes change the meaning of the root word. For example, adding the derivational morpheme “un-” to the root word “happy” turns it into “unhappy,” a different word with a different meaning. Some, but not all, derivational morphemes change the lexical class of the word; for example, “-ize” turns a noun into a verb. (Example: “energy”, a noun, plus “-ize” results in “energize,” a verb.) It is possible to create novel words using derivational morphemes. (Somebody had to be the first to use the word “prioritize,” for example.) Derivational morphemes that can and are frequently used to form new words are considered productive. There are too many derivational morphemes list them all. For some common ones, see “Common Derivational Morphemes Chart”; but try to do the exercises in this chapter before you study the chart. That attach to... nouns verbs adjectives plural –s (or –es) (dogs, churches) present tense continuous –ing (running, jumping) past tense –ed (walked, slept) past participle –en (eaten, taken) comparative –er (happier) superlative –est (happiest) 49 Roots Every word has a single root—a free morpheme onto which other, bound morphemes may be attached. • Exception: Compound words (like “pancake” or “blackboard”) have two roots (free morphemes). • Bound Roots: There are certain words which clearly have morphology, but which don’t have a free root. (For example, “antagonist” and “antagonize” have the “-ize” and “-ist” suffixes as we know them, and they are certainly related in meaning, so it seems senseless to consider them separate, monomorphemic words; but the problem with separating them into “antagon” and “-ist” or “antagon” and “-ize” is that “antagon” cannot stand alone.) In our example problems, every word should have a free root, but if your teacher gives problems like this, and you need to find out how they are to be handled in your class. Affixes An affix is any morpheme that is added onto a root—in other words, any bound morpheme is an affix. In English, we have two kinds of affixes: Prefixes: Attach to the beginning of a word, like “un-” or “dis-” or “re-”. Suffixes: Attach to the end of a word, like “-tion” or “-ing” or “-ist.” Other languages might have additional types of affixes, such as: Infixes: Inserted in the middle of the root. In English we actually do have one infix, but only one: forms of the word “fucking,” as in “abso-fucking-lutely!” • Note: Don’t make the mistake of identifying an affix as an infix just because it is neither at the beginning nor the end of the word. In “activation” (act+ive+ate+tion), “-ate” is NOT an infix—it’s just the first suffix. Remember, infixes only occur when the root is actually broken into two pieces to surround the morpheme. Circumfixes: Single morphemes that are broken into two parts and placed on either side of a root. There is no example of this in English. • Note: Don’t make the mistake of identifying a pair of different morphemes as a circumfix. A word like “unreliable” (un+rely+able) may have a root surrounded by morphemes on either side, but “un...able” is not a circumfix—they’re two different morphemes which can be separated, mixed, and matched. 50 Subcategorization Frames Objective: Write subcategorization frames for various derivational morphemes. Overview A subcategorization frame tells you what lexical category a morpheme takes, and what lexical category it produces. For example, Take the suffix “-ize,” found in words like “prioritize,” “energize,” and “ionize.” All of these are verbs, so we know that the suffix “-ize” takes words and turns them into verbs. What kind of words does it take? Well, look at the roots this suffix is acting on: priority + ize energy + ize ion + ize “Priority,” “energy,” and “ion” are all nouns. From this evidence, we see that “-ize” takes nouns and turns them into verbs. So the subcategorization frame looks like this: -ize: N -> V Step-by-Step Suppose we’re trying to find the subcategorization frame for the suffix –tion. 1. Brainstorm some words that use this morpheme. I came up with “creation,” “action”, “retention,” and “station.” 2. Break down these examples, removing the morpheme. Here, we realize we need to throw out “station” since it seems to be a free morpheme on its own, not “sta + tion”. (What’s a sta?) The rest look like this: create+tion act + tion retain + tion 3. Identify the lexical categories of the words, with and without the morpheme, and compare. “Create,” “act” and “retain” are all verbs; “creation”, “action” and “retention” are all nouns. So – tion takes a verb, and turns it into a noun. The subcategorization frame looks like this: -tion: V -> N (evidence: “activation,” “action,” “retention”) Tips • Some morphemes have more than one meaning (and therefore, more than one subcategorization frame). Much like words like “bat” have multiple meanings, many morphemes look and sound the same, but are really different, and may have different subcategorization frames. There are, for example, two kinds of “un-” and two kinds of “- er” (see example problems). • Roots might change spelling (or even pronunciation) when a morpheme is added. You might have to do a little thinking to realize that, for example, the root of “detention” is “detain.” • Be aware of the meaning of your morpheme (and choose evidence accordingly). For example, “underwear” can’t be used as evidence for the subcategorization frame of “un-” 51 Common Derivational Morphemes Chart Morpheme Subcategorization Frame Meaning Examples -able V -> Adj able to be Xed washable -al N -> Adj professional anti- N -> N against X antiwar -ate Adj -> V make be X activate de- V -> V do reserve of X deactivate dis- V -> V do reverse of X disestablish -er V -> N one who Xs baker, swimmer, singer -ful N -> Adj full of X beautiful, bountiful -ify N -> V make be X clarify, beautify in- Adj -> Adj not X indecent, inorganic -ity Adj -> N the state of being X normality, curiosity -ish Adj -> Adj sort of X greenish, prettyish -ism N, Adj -> N belief in or support of X egoism, creationism, universalism -ist N, Adj -> N one who does, believes in or supports X artist, egoist, creationist, universalist -ive V -> Adj active -ize N -> V terrorize, prioritize, energize -less N -> Adj without X clueless, penniless, careless -ly Adj -> Adv in an X manner hurriedly, happily -ment V, Adj -> N establishment, merriment mis- V -> V not X, not X correctly misanalyze, misunderstand -ness Adj -> N prettiness, niceness -ous N -> Adj dangerous re- V -> V X again reheat, recreate -tion/-ation/- ion V -> N creation, action, decision un- (1) V -> V do reverse of X untie, unlock, undo un- (2) Adj -> Adj not X unhappy, undecided -y N -> Adj lucky, sexy 54 Morphology Trees Overview: Draw morphology trees. Overview Morphology trees are diagrams used to illustrate: (1) what are the individual morphemes in a given word, (2) what order the affixes were added to the root, (3) what each affix is doing to the word—that is, the subcategorization frame of the morpheme, and the lexical category of each word along the way. Step-by-Step Suppose we’re trying to do a morphology tree for the word “activation.” 1. Divide the word into all its component morphemes. For “activation”, that’s: act + ive + ate + tion 2. Give the lexical category of the root, and the subcategorization frames of the morphemes. (If you can’t do them all now… you may be able to fill them in later.) The root is “act,” which is a verb. So we’ll label “act” with its lexical category, and write the subcategorization frames off to the side. V act + ive + ate + tion ive: V-> Adj (evidence: creative, explosive) ate: Hmm. I can’t think of many words that end in “ate.” I’ll leave this one for now. tion: V->N (creation, action) 3. Connect the root with first morpheme that’s attached, and label with the lexical category of the new word. We only have one option for which morpheme is attached first, because there is only one morpheme next to the root! (For more complicated cases, see “How do I tell…?”) The new word, “active,” is an adjective. Adj V act + ive + ate + tion 4. Keep going until you have reached the top! “Activate” is a verb: 55 V Adj V act + ive + ate + tion Finally, “activation” is a noun. Here is the final tree: N V Adj V act + ive + ate + tion Subcategorization frames: -ive: V->Adj -ate: Adj->V -tion: V->N How do I tell what order the morphemes go on in? Generally, the morphemes have to connect in order: in the above example, “activation” is not formed by combining “act” and “-ate” first, for instance. However, what happens when there are two possible morphemes that can connect next? Let’s take, for example, the word “unbelievable”. This word is composed of three morphemes: un + believe + able. We know that believe is a verb, so we can start making the tree: un: V V un: Adj Adj able: V Adj (“doable”, “imaginable”) V un + believe + able However, here we run into a problem. Does “un-” connect first, or does “-able”? There is a shortcut that can be used here because one of these morphemes connects to the root to make a word, and one of them does not. That is, “believable” is a word, but “unbelieve” is not. Therefore, we know that “-able” connects first. 56 Now, what order do we add the morphemes in? Recall that “-s” is inflectional (the plural morpheme for nouns) and that inflectional morphemes always get added last. So we know that much. Next question: Is this “-ing” the inflectional “-ing” or the derivational “-ing”? If it’s inflectional, then the whole word “misunderstanding” will be a verb (since inflectional “-ing” always works on verbs). If it’s derivational, then the whole word will be a noun (since derivational “-ing” turns verbs into nouns). Actually, the word “misunderstanding” could go either way (a noun in sentences like “It was all a big misunderstanding” and a verb in sentences like “She was misunderstanding everything I said.”) But the word “misunderstandings” must be a noun, since only nouns can be plural (in other words, “-s” is N -> N). So in this case, “misunderstanding” is a noun and “-ing” is derivational. So we know that “misunderstandings” is going to be a noun, and “misunderstanding” is also going to be a noun. Now we just need to decide what to add to the root first: the “-ing” (V- >N) or the “mis-” (V->V). Both take verbs, so either one could work on “understand”. But if we add “-ing” first, we’ll get a noun, and “mis-” can’t work on a noun. On the other hand, if we add “mis-” first, we’ll still have a verb, and “-ing” can work on it. So the only thing we can do is add “mis-” first and then add “-ing.” To sum up: we start with “understand”, a verb; add “mis-” and we have “misunderstand”, still a verb; add “-ing” and we have “misunderstanding”, a noun; and finally, add the inflectional plural “-s” and we have “misunderstandings”, a noun. N N V V mis + understand + ing + s e. unlockable The morpheme breakdown: un + lock + able Here, it is crucial to realize that “unlockable” has two meanings. One meaning is “not able to be locked”, as in “The lock is broken, so this door is unlockable”. The other meaning is “able to be unlocked” as in “That door has a flimsy lock, it’s unlockable with a credit card”. Each meaning has a different tree structure, but both trees have a lot in common: both versions of the word are adjectives, both have “lock” as the root (and we know “lock” has to be a verb and not a noun because “-able” does not attach to nouns, and neither does “un-”). The difference is in the attachment of the morphemes. One of the meanings corresponds to “un-” attaching first, the other meaning corresponds to “-able” attaching first. 59 Let’s draw the tree with “un-” attaching first: Adj V V un + lock + able Which meaning does this correspond to? We make the word “unlock” first, and then attach “- able”. “-able” attaches to verb X, and makes an adjective that means “able to be Xed”, so this structure means “able to be unlocked”, which is the second meaning discussed above. The other option is to attach “-able” first: Adj Adj V un + lock + able In this structure, we make “lockable” first, and then attach “un-” which, when attached to an adjective X, means “not X”. So, this structure makes “not lockable”, which is the first meaning discussed above. Exercises Make morphology trees for the following words: a. unmentionable b. revelations c. undeniably d. untieable e. deactivating 60 Zero Derivation Overview You know that “-ed” and “-ing” are inflectional morphemes which assign tense to verbs, as in “I waited for you all day, but you never arrived!” But what about, “This is the most anticipated film of the year”? What lexical category do you assign to “anticipated” in this context? How can you explain this? One way to explain the adjectival “anticipated” in “the most anticipated film of the year” is to suggest that this is a new, derivational “-ed” which transforms verbs into adjectives. But there is an alternative explanation, which is that a new, non-pronounced (or null, written as Ø) morpheme is being added to the word. Ø can take any subcategorization frame you please. Adj V V anticipate + ed + Ø (as in “the most anticipated film of the year”) -ed: inflectional verb tense marker Ø: V Adj Adding a null morpheme, or zero derivation, is the favored explanation because the practice of using a word in a different part of speech than usual is so widespread in English. Consider the following three sentences, where the word “baby” changes its lexical category without any attachment of morphology: • He held the baby awkwardly. (noun) • The baby mice were adorable. (adj) • Her mother likes to baby her. (verb) Example Problems 1. Show how to derive the adjective and noun versions of “-ing” words, as in “the slumbering giant” or “Swimming is my favorite activity.” Adj V V slumber + ing + Ø (as in “the slumbering giant”) -ing: inflectional verb tense marker Ø: V Adj 61 • The original or underlying lexical category of a given word isn’t always clear. Remember the word “heat” from the examples--it might originally be a noun or a verb. If you can manage to consult one, an etymology dictionary may resolve the question. When in doubt, you may zero-derive or just assume that the word originally has the lexical category that would make life easiest for you; as always, your mileage (and your professor’s) may vary. Exercises 1. Business jargon provides a rich source of zero derived words. Provide tree structures for the following words: a. “actioning” as in, “I’ll talk to Todd in accounting and see about actioning that for you.” b. “table” as in “I think we should table this discussion for now.” 2. Provide tree structures for the first two words in the sentence “Verbing weirds language.” 64 Morphology in Other Languages We can divide languages into the following groups according to how they handle morphology: Isolating (or analytic): Almost every word is a free morpheme. Example: Vietnamesex khi tôi dên nhà ban tôi, chúng tôi bát dâu làm bài. when I come house friend I Plural I begin do lesson When I came to my friend's house, we began to do lessons. Synthetic: Words consist of many morphemes. Types of synthetic languages: Agglutinative: A sequence of morphemes are added in a specific order, and each morpheme represents one unit of meaning. Agglutinating languages tend to be very regular (with some exceptions, e.g. Georgian). Example: Turkishxi Turkish adds morphemes (mainly suffixes) to words, in a specific order, to indicate person, number, aspect, and other features of language: ev house evin your house evinde at your house Evindeyim I am at your house Fusional: A single morpheme encodes several units of meaning. Example: Latin The Latin word “bonus”, meaning “good man” has the suffix “-us” which indicates masculine gender, nominative case, and singular number. Changing any of these features requires a completely new suffix. For instance, to say “good men”, thereby changing only the number (from singular to plural) requires the use of the suffix “-i”: “boni”. In an agglutinating language, only the appropriate affix indicating number would be changed. Agglutinating language puzzles Agglutinating languages puzzles consist of examining a set of data from an agglutinating language, and determining which affixes indicate which feature, and what order these affixes attach to the root word. Given a set of phrases and their glosses, we can try to figure out the morphological structure of a language. For example, let’s say that we have the following information: 65 MFL 1xii English Gloss MFL phrase I see her kurakax He saw her kuratix I ate tixoki You (sg) wrote (to) her elalix She will eat it elao They (masc) see her kurafax I ate them (fem) tixokir You (sg) didn’t eat them (inanimate) tixoliæ We see them (masc) kuramaz They (fem) wrote elani He will write it elato She eats tixoa We write elama They (masc) will see them (inanimate) kurafo What appear to be the distinctions that are identified in the English glosses? First, there are three different roots (“see”, “eat”, “write”). There are several tenses (present, past, future), and pronouns that sometimes differentiate between genders (“I”, “He”, “You”, “She”, “We”, “They (fem)”, “They (masc)”). There are also direct objects (“her”, “them (fem, masc, inanimate)”, “it”). So, let’s find all the corresponding morphemes. To do this, we need to analyze the MFL phrases that differ in only one element (do you recall “minimal pairs” from phonetics/phonology? The principle is the same). So, for example, the only difference between “I ate” and “I ate them (fem)” is the presence of the direct object “them (fem)”. So, we can compare the MFL phrases “tixoki” and “tixokir”. The only difference between them is the final “r”, so we can assume that the direct object “them (fem) is indicated by the morpheme “r”. Now, we can try to find the root word for “eat”. Look at “I ate” and then “She eats”. There are two differences between these phrases: one difference is the pronoun, and the other is the tense of the verb. So, when we compare the two phrases, the root is what they have in common: “tixoki” ad “tixoa” have “tixo” in common, so that is probably the root. Next, let’s look at the first two examples. There, the root is probably “kura” (for “see”) because that is what those two phrases have in common. Further, we can compare “he saw her” with “I ate”, since we know both of the roots. “kuratix” and “tixoki” have one thing in common (if the roots are disregarded), namely the “i”. Since both of these phrases are in the past tense, and there is no other similarity, we can conclude that “i” indicates past tense. 66 Exercise: MFL 2 Find the following morphemes in the data set below: Subjects Objects Tense Aspect Roots I present progressive (-ing) like we us past perfect (have done) draw you (sg) future wear you (pl) you (pl) he him she her they them it Also, find the order in which the morphemes attach. English Gloss MFL phrase I am liking it furynibusatinex You will draw him resœenuvul We like us taloibusa They had drawn you all sutedvenuawt I drew them furmenua You (pl) will draw us sisoenuvul She is drawing him aœenutinex We will have liked it talynibusavulwt You (sg) have worn her resoforotiwt He drew lytenuti I had liked her furofibusaawt She draws them amenuti You (pl) are wearing it sisynorotinex He liked them lytmibusaa They have drawn her sutofenutiwt How would this language construct the following sentences? a. I am wearing it. b. You (pl) will like us. c. He had drawn them. d. She likes you (pl). e. They will wear it. f. I had liked you (pl). 69 Word Formation Processes New words are created all the time. Here are some of the processes that account for the coining of new words. The “<” symbol means “comes from”. Process Examples adding morphology A new word is created by adding morphology to an existing word. undead < un- + dead slayage < slay + -age embiggen < en- + big + -en backformation A new word is created when people take a word that appears to have been created by adding morphology, and try to get at the root by removing the morphology. Happens due to a misanalysis of the origin of the original word. pea < pease – -s resurrect < resurrection – -tion compounding Two words are combined, retaining the entirety of both words. weblog < web + log blending Two words are combined, and some intermediate portions of both words are dropped. smog < smoke + fog Bennifer < Ben + Jennifer clipping A word is shortened. lab < laboratory blog < weblog initialism (or alphabetism) A sequence of initials used as a word (the individual letters are pronounced). ATM < automatic teller machine ID < identification acronym A sequence of initials used as a word (sounded out phonetically rather than saying individual letters). NASA < National Aeronautics and Space Administrations laser < Light Amplification by Stimulated Emission of Radiation generification People come to use a brand name to describe any similar product. Kleenex (meaning any tissue) < Kleenex (meaning Kleenex brand tissue) lexical shift (or zero derivation) Words that used to be one lexical cateogory are used in another, without adding morphology. The phrase “zero derivation” refers to the idea that a null (unpronounced) derivational morpheme is being added to the root to change its lexical category. to Xerox (V) < (a) Xerox (N) to blog/blogging (V) < (a) blog (N) If you like this, you’ll love “Processes of Semantic Change” in the section on Semantics. 70 Syntax Syntax is the study of sentence structure. Lessons Constituency Overview Constituency Tests Example Problems Exercises Lexical Categories Overview Lexical Categories Quick Reference Example Problems Exercises English Phrase Structure & Syntax Trees Overview English Phrase Structure Rules Abstract Phrase Structure Rule Key Points about Phrase Structure Rules Tree Drawing Overview Examples Example Problems Tips Exercises Ambiguous Sentences Overview Example Problems Exercises Complex Sentences Overview Example Problems Exercises Transformations Overview Example Problems Tips Exercises 71 of the tests produce a grammatical result, we can say that many mathematical equations is a constituent in this sentence. 3. Decide if the underlined portion of the following sentence is a constituent: John solved many mathematical equations for Meredith. As above, if John solved many is a constituent, then at least two of the tests need to produce a grammatical result: Substitution: (*) He mathematical equations for Meredith. Q/A: (*)What mathematical equations for Meredith? John solved many. Cleft: (*) It was John solved many that mathematical equations for Meredith. Pseudocleft: (*) The things that many mathematical equations for Meredith were John solved many. All four tests produce ungrammatical results; therefore, John solved many is not a constituent. 4. Decide if the underlined portion of the following sentence is a constituent: John solved many mathematical equations for Meredith. Once again, we perform the constituency tests to decide if for Meredith is a constituent, but these require some thinking: Substitution: John solved many mathematical equations there. (While “there” seems like an odd substitution for “for Meredith”, there is no other word that will fit better. Whether or not this is a valid substitution is up to the individual.) Q/A: Why did John solve many mathematical equations? For Meredith. (While one may be tempted to ask “for whom” in this question, “for” is already used in the answer and cannot be repeated in the question. Whether or not “for Meredith” is a valid answer to “why” is up to the individual.) Cleft: It was for Meredith that John solved many mathematical equations. Pseudocleft: The reason John solved many mathematical equations was for Meredith. (This one is also unclear: is “for Meredith” really a valid “reason”?) Because some of these tests show unclear results, for Meredith may appear as a constituent to some people and not to others, depending on the grammaticality of the above statements. Exercises Decide if the underlined portion of the following sentences is a constituent (show the constituency tests used): 1. Meredith mocked John’s solutions in the science lab. 2. Meredith mocked John’s solutions in the science lab. 3. Meredith mocked John’s solutions in the science lab. 4. Meredith mocked John’s solutions in the science lab. 5. Meredith mocked John’s solutions in the science lab. 6. Meredith mocked John’s solutions in the science lab. 7. Meredith mocked John’s solutions in the science lab. 74 Lexical Categories Objective: Given a sentence, identify any given word’s lexical category. Overview Lexical categories are often referred to as “parts of speech”. Lexical categories are divided into two main groups: function words and content words. Content Words: Generally, content words nouns, adjectives, and verbs – they are words that contain meaning in and of themselves. Function Words: Function words, which include prepositions, determiners, and conjunctions, are used for grammatical structure and do not have inherent meaning. For example, “dancing” is a content word, whereas “with” is a function word. Words can be different parts of speech in different sentences, depending on context. For example, both of the sentences below include the word “dancing”. However, in one sentence, “dancing” is a verb, and in the other, it’s a noun: a) John is dancing with Meredith. (verb) b) John likes dancing. (noun) Similarly, “her” has different lexical categories in the following sentences: a) He likes her. (noun) b) Her books are on the table. (determiner) Generally speaking, one can use the following rules to identify lexical categories. The rules for content words (nouns, verbs, adjectives, and adverbs) are described in the Morphology: Basic Lexical Categories section. These are not hard and fast rules; there are always exceptions. A non- exhaustive list of function words and their categories is in the Lexical Categories Quick Reference (later this section), but it is important to understand that some words can be in any of several lexical categories depending on the sentence; it’s better to look at the role the word is playing in the sentence than what word it specifically is. Nouns/Pronouns, Verbs, Adjectives, Adverbs: See Morphology Section Determiners: The most common determiners are “the”, “a”, and “an”; however, there are others including possessive pronouns (his, her, their) and identifiers like this, that, and those. One simple test: a word is acting as a determiner if it can be replaced by “the” or “a”/“an”. Some examples: • A book is on the table: Both “a” and “the” are determiners. • Her book is on that table: Both “her” and “that” are determiners. • Our book is on his table: Both “our” and “his” are determiners. • He gave his book to her: Only “his” is a determiner in this case because it’s the only word that can be replaced by “the” or “a”/“an”. That is, “He gave the book to her” is an acceptable sentence, but “He gave his book to the” is not, even though “her” was an acceptable determiner a previous example. 75 Modals/Auxiliary Verbs: Auxiliary verbs are also called “helping verbs”. In sentences where there are two verbs immediately next to each other, one of the verbs is a main verb and the other is an auxiliary. For example, “I am eating now” contains two verbs: “am” and “eating”. “Eating” is the main verb; “am” is an auxiliary that helps determine the tense of the sentence. Auxiliary verbs can sometimes be main verbs. For example, in “I am here,” “am” is the main verb. Likewise, in “I have seen a cat,” “have” is an auxiliary, whereas in “I have a cat,” “have” is the main verb. Modal verbs are verbs that can never function as the main verb. There are nine of them in English: can/could, may/might, shall/should, will/would, and must. For example, in the sentence, “I could use a drink,” “could” is the modal verb, and “use” is the main verb. Conjunctions: Conjunctions connect two words or phrases of the same lexical category. For the purposes of this guide, the only conjunctions are “and”, “but”, and “or”. For instance, “and” connects two nouns in the first sentence, two verbs in the second, and two whole sentences in the third: • John and Meredith like to explore. (John and Meredith are connected by the conjunction) • John flies planes and hates bugs. (“flies planes” and “hates bugs” are verb phrases connected by the conjunction) • John sets them up and Meredith knocks them down (each side of the conjunction comprises an entire sentence) Complementizers: Complementizers are words that introduce an embedded sentence. Some examples: that, because, if, whether, which. For example, the following sentence contains an embedded sentence that is introduced by the complementizer “that”: John is convinced that Meredith is stalking him. “Meredith is stalking him” is a complete, embedded sentence. Note that this use of “that” is different from the previously discussed uses of “that” as a noun or a determiner. In this sentence, it is not possible to substitute another noun in place of “that”, so it does not function as a noun. It is not possible to substitute “the” in place of “that,” so it is not acting as a determiner. It is conceivable to imagine a reading of the sentence where “that” is a determiner, but only if there is more than one Meredith, and the point of the sentences is that “that Meredith” is stalking John, and not “this other Meredith”. However, this reading of the sentence is not the one that first comes to mind, so it is probable that “that” is a complementizer in this sentence. Qualifiers: Qualifiers modify adjectives and adverbs. Common examples include rather, very, and quite. For example, in the following statement, “rather” modifies the adjective “incinerating” and “very” modifies the adverb “quickly”: Meredith’s rather incinerating glare made John catch on very quickly. Quantifiers: Quantifiers are “number” words such as: one, two, fifteen, 300, etc. They are also expressions that refer to number, such as: many, few, some, much, etc. 76 English Phrase Structure & Syntax Trees Objective: Use phrase structure rules to draw syntax trees. Know the parts of a syntax tree and be able to describe and understand them. Overview There are language-specific rules of syntax that govern sentence structure. This section examines one way of looking at English syntax, with an internally consistent set of rules that is not complete (or entirely correct) but that shows, conceptually, how the sentence structure rules are used to create syntactic trees. Keep in mind that these rules are provided as an example of simple rules we might use to describe English; the ones used in your class might vary in terms of sophistication and/or labeling conventions. Recall that in the section on constituency tests, we described units of the sentence which naturally go together, or “constituents.” Now, we’ll call constituents “phrases.” We’ll have a special kind of phrase for every kind of content word: nouns phrases, verb phrases, adjective phrases, and adverb phrases, as well as prepositional phrases. English Phrase Structure Rules CP C IP IP NP I VP NP (Det) (Quant) (AdjP) N {PP, CP} VP V {PP, NP, CP, AdvP} AdjP (Qual) Adj AdvP (Qual) Adv PP P NP The way to read these rules is as follows: P In general, “P” means “phrase” (except in the case of Prepositional Phrases, in which case the first P means, funnily enough, “Prepositional.”) The other letters correspond to lexical categories (the same abbreviations we used in the previous section, “Lexical Categories.”) I means “inflectional”; a sentence is known as an inflectional phrase. Auxiliaries and modals are considered inflectionals, as well as any tense information (more on this later). Arrows The phrase at the left of the arrow is the one “on top”, which contains the elements to the right of the arrow. Parentheses ( ) Optional elements are in ( ), required elements are not. Brackets { } The elements included in { } can occur in any order and are also optional. WARNING: These rules apply to English syntax only. Other languages have their own rules for word and phrase order. 79 Phrase Structure Rules, Abstracted As you may notice from above, there are some regularities in the way phrase structure rules are formed. For one thing, every phrase has at least one non-optional element, which the phrase is named after: a noun phrase always has a noun, a verb phrase always has a verb, etc. That non- optional namesake element is called the head. Like every person, phrase has one head. Beyond that, all the phrase structure rules are basically structured as follows: Some Kind of Phrase (XP) (Some Optional Stuff) X (Some more optional stuff) where X is the head of the X-phrase (whatever X is). We call the stuff before the head the specifier, and the stuff after the head the complement. Thus the general phrase structure rule is XP Spec X Comp Remember that even this seemingly basic rule only applies to English—Japanese, for example, puts the specifier after the head. Key Points about Phrase Structure Rules • The order of phrases is important, and is language specific. In English, a sentence must be composed of a noun phrase, an I (inflectional), and a verb phrase, in that order. • Some elements could occur in many different phrases (an NP could be in an IP, a VP, or a PP, for example), while others can only occur with certain phrases. For instance, determiners can only occur in noun phrases, and adverb phrases can only occur in verb phrases (which makes sense, since adverbs modify verbs). • Recursion: Any rule/structure which could, theoretically, go on forever is recursive. For instance, a NP can take PP (optionally) and the PP requires a NP. The NP within the PP can then take another PP, and so on. This creates a structure such as: “the box with the ribbon with the gold trim with the white lacing…” etc indefinitely. • Conjuctions: There are additional rules for combining elements using conjunctions. So, there are additional rules for NP, VP, PP, and AdjP/AdvP as follows: NP NP conj NP “John and Meredith” VP VP conj VP “loved pie and hated citrus” PP PP conj PP “with milk and with sugar” AdjP AdjP conj AdjP “pretty and nice” AdvP AdvP conj AdvP “quickly and efficiently” As noted above, the specific rules from your class may differ; the important thing is that you follow the rules that you develop (whatever those rules are) so that your structures are consistent. Tree Drawing Overview We will use the phrase structure rules above to draw tree diagrams which show the structure of sentences in a pretty visual diagram. Some vocabulary for tree drawing: Node: Any place on the tree which is labeled (e.g. N, NP, AdjP). The point of tree-drawing is to show how the nodes attach to each other. 80 Branch: A line on the tree showing relationships/attachments between nodes. Mother: The node directly above/on top of another node (only when directly connected with a branch) is its parent. Daughter: The node directly below/under another node (only when directly connected with a branch) is its daughter. Sister: Nodes which have the same mother are sisters. Immediately dominates: A node immediately dominates any node which it contains/is on top of, as long as it is connected by one branch. Dominates: A node dominates any node which it contains/is on top of, as long as it is connected by some series of branches (it need not be a direct, one-branch connection, but you may not include branches which lead to a higher level). Examples Here’s a sample tree which demonstrates the vocabulary we’ve talked about so far. “Example Problems” (next page) gives more detail on how to draw these trees for yourself. IP NP1 VP I +past NP2 N V D N John kissed the scientist. You can see how the phrase structure rules relate to the tree: IP NP I VP The IP contains an NP, I, and VP NP (Det) (Quant) (AdjP) N {PP, CP} NP1 contains its only mandatory member, N. NP2 also contains an optional determiner. VP V {PP, NP, CP, AdvP} The VP contains a V and an NP, acceptable in the rule • All of the labeled areas where branches can connect (IP, NP1, I, VP, N, V, NP2, D, N) are nodes. • IP is the mother of NP1, I, and VP. IP is not considered the mother of N (either one), V, D, or NP2 because there is no direct connection between them. • V and NP2 are daughters of VP. Can you find some more mother-daughter relationships? • NP1, I, and VP are sisters. Can you find some more sister relationships? • VP immediately dominates V and NP2. IP dominates both VP and V (actually, IP dominates everything), but only immediately dominates VP. • NP1 is on a higher level than, say, D, it does not dominate D. There is no way to trace down branches from NP1 to D without going up a level first. 81 Once this is complete, the tree is the same as above: IP NP VP I +present NP N V D N John likes the scientist. 3. John likes Meredith and turkey sandwiches This is a little more complicated, as it contains a conjunction. First, we label the words: N V N Conj Adj N John likes Meredith and turkey sandwiches Note that “turkey” in this case is an Adjective because it describes the kind of sandwich that John likes. As before, we now find the subject N (John) and the V (likes) and draw them higher than the other phrases: NP VP NP NP AP N V N Conj Adj N John likes Meredith and turkey sandwiches Notice that the adjective phrase modifies the noun that follows it (what kind of sandwiches? turkey sandwiches!). So, the AdjP connects to the NP “sandwiches”. Also, now we can see that the conjunction combines two noun phrases (“Meredith” and “turkey sandwiches”) so we can combine them into a new NP. As before, the subject NP and the VP go to IP: IP NP VP I NP +present NP NP AP N V N Conj Adj N John likes Meredith and turkey sandwiches 84 Note that in the case of conjunctions is the only case, according to our rules, that a XP phrase can dominate itself (i.e. normally, NP is not allowed to contain an NP). Now, for the final tree, we connect the object NP to the VP and we are done: IP NP VP I NP +present NP NP AP N V N Conj Adj N John likes Meredith and turkey sandwiches 4. The military commander likes the pretty scientist from Canada As before, we label the categories: D Adj N V D Adj N P N The military commander likes the pretty scientist from Canada The new element here is the prepositional phrase; the key with these is to remember that a prepositional phrase always takes the noun phrase immediately to the right of the preposition. We find the subject N (commander) and the V (likes), and draw the extensions as before: NP VP NP PP AP AP NP D Adj N V D Adj N P N The military commander likes the pretty scientist from Canada We know that the main NP and the VP will go to IP; we know that “the pretty scientist” is an object of the verb “likes”, so will attach to the VP. However, the PP has two places it could go – either to the NP next to it, or to the VP a little further away. The question is, what does “from Canada” modify – the scientist, or the liking? It’s pretty clear from this example, that “from Canada” is talking about the scientist, so it will connect to the NP: 85 IP NP I VP +present NP PP AP AP NP D Adj N V D Adj N P N The military commander likes the pretty scientist from Canada Tips • Find the subject N and the main V, and immediately place them as the head of their respective NP and VP. These will go directly to IP. You can draw them higher than the other elements. • Generally, Adjectives will modify something, so the associated AP needs to be drawn lower than the phrase containing whatever it is they modify (usually an NP). • Follow the rules – the rules are very limiting, and can help you if you’re not sure which words to put into the same phrase (e.g. in “the pretty scientist”, “the” is part of the NP the head of which is “scientist” – if you’re not sure about this, the rules will tell you because they don’t allow determiners in adjective phrases). • The entire tree has to be connected; the only node that doesn’t have a mother is the main IP. • Everything else has one (and only one) mother. A node can have (but doesn’t have to have) many daughters, but can only have one mother. Exercises Using the structure rules, draw the following trees: 1. Meredith loves chocolate. 2. Meredith loves chocolate truffles. 3. Meredith loves chocolate truffles with real milk chocolate. 4. Meredith loves cookies and milk. 86 Meaning 1: John sold the device to the scientist who is currently in Canada: IP NP VP I PP [past] NP PP NP NP N V D N P D N P N John sold the device to the scientist in Canada. Note that “to the scientist” attaches directly to the verb phrase; it is not part of the other noun phrase “the device”. This is because “to the scientist” refers to the selling; it doesn’t say anything about the device. In another sentence, as in: “John sold the device with the shiny light in Canada”, “with the shiny light” would be describing the device and would therefore attach to “the device” NP. Meaning 2: John sold the device to the scientist, and the sale took place in Canada: IP NP VP I PP PP [past] NP NP NP N V D N P D N P N John sold the device to the scientist in Canada. Note that the PP in question (in Canada) attaches directly to the VP. 89 3. Meredith drinks coffee with extra milk and sugar. Here, the ambiguity is of a slightly different sort: the question is whether Meredith drinks coffee with extra “milk and sugar” (that is, if there’s extra milk AND extra sugar) or if Meredith drinks coffee with “extra milk” and sugar. So, the question is in whether or not the conjunction dominates the word “extra”. Meaning 1: Meredith drinks coffee with extra milk and extra sugar. IP NP VP I [pres] NP PP NP AdjP NP NP N V N P Adj N Conj N Meredith drinks coffee with extra milk and sugar. Note that with this reading, the adjective “extra” is modifying both milk and sugar, since it attaches at the level of the conjunction. Meaning 2: Meredith drinks coffee with extra milk and a normal amount of sugar. IP NP VP I [pres] NP PP NP NP NP AdjP N V N P Adj N Conj N Meredith drinks coffee with extra milk and sugar. Note that here, the adjective phrase “extra” only modifies “milk”, because it attaches to “milk” before the conjunction unites the milk NP and the sugar NP. 90 Exercises Determine the meanings in these ambiguous sentences and provide the meanings with corresponding trees. 1. John shot the marine with the gun. 2. Meredith likes chocolate truffles and cake. 3. John burned the picture of Elizabeth with her ex-boyfriend. (three possible meanings here, one requires treating “with” as a conjunction) 91 The English language allows for an infinite construction of this type (another example of recursion!) It’s always possible to embed the sentence in another sentence. It’s important to think of every CP as just another constituent in order to not become confused about the attachments of the various phrases. 4. John believes Meredith likes him. Sometimes you’ll have an embedded sentence which is seemingly introduced by no complementizer. No, the complementizer is not optional—it is the head of the CP which contains the embedded sentence, after all. In this case, we assume the C position is still there in our mental model of the sentence, but the corresponding word (e.g. “that”) is dropped in fast speech. We insert a null symbol (Ø) to show that something belongs there, but that it is not pronounced. Otherwise, this is identical to sentence #1. IP NP VP CP I [pres] IP NP VP I [pres] NP N V C N V N John believes Ø Meredith likes him. Exercises: Construct the trees for the following complex sentences: 1. John insists that Meredith brought the cake without citrus to the party. 2. The announcement that Meredith went to the party raised some eyebrows. 3. John played the game because Meredith insisted on his participation. 4. John thinks Meredith sang the song about love for him. 5. John heard that Meredith ate the cake yesterday. (This sentence is ambiguous; give both meanings/trees) 94 Transformations Objective: Draw trees for yes/no and wh-word questions. Overview Think about how we phrase questions in English (e.g. “What has John done?”). The question form is related to the declarative statement form, and you can form a declarative version of any sentence (“John has done what”). We assume that the declarative version is the basic version you start with in your brain (the deep or underlying structure), and the version you actually say—the question form—is the result of a transformation, moving words around in a specific, rule-based way. This section covers the transformations that occur in English sentences to turn them from declarative statements to question statements. Now that we’re talking about questions, we’ll start drawing the main IP of all our sentences as embedded within an overarching CP. The transformations rely upon having those extra nodes to work with. (The examples will explain further.) We’ll also start including a non- specific Specifier (Spec) position in a CP. Recall that “specifier” just means “stuff before the head” (for English, at least). The two transformations we cover are Inversion (movement of the I to the C position of the main CP, as in yes/no questions such as, “Can John love Meredith?” from “John can love Meredith”) and Wh- Movement (movement of a wh-word, such as who, what, where, when, to the Specifier position of the main CP, as in “What has John done?” from “John has done what”). Example Problems Draw the underlying and surface structure of the following sentences: 1. Can John love Meredith? First, we find the underlying sentence (also known as “deep structure”). You could think about this as the sentence that provides an answer to the question. In this case, it is “John can love Meredith.” We draw the tree for this sentence as usual. The only difference is that we have to put in the empty CP at the top to indicate where the moved elements will end up. Underlying structure: CP Spec C IP NP I VP NP N mod V N John can love Meredith 95 The surface structure (or the version of the question that you actually pronounce) is very, very similar, so the easiest thing to do is to re-draw the entire tree from above and then make the changes with a little erasing. The changes are as follows: - Figure out what word moves, and where it moves to. The only term that moves here is “can,” and it moves to the beginning of the sentence. We put in the C position in the main CP. So, write “Can” under C, and erase it from where it used to be (under mod). - Replace the moved term with a trace (t). In this case, write t where “can” used to be, under mod. - Attach the trace to the word’s new location with an arrow, and label the arrow to describe the kind of movement. Here, draw an arrow from the t to “Can.” Since we’re doing an inversion, we write “inversion.” The resulting tree looks like this: Surface structure: CP Spec C IP NP I VP NP N mod V N Canx John tx love Meredith inversion 2. What has John done? Underlying structure: John has done what. “What” is generally a NP. CP Spec C IP NP I VP NP N mod V John has done what 96