Linguistics 503: Examination 1

Oct. 22, 1998

Grade distribution 
                                       +
                                       +
  + +     +    +      +      ++      + +  +   +
'         '         '         '         '         '
50        60        70        80        90       100
Instructions: This test is worth 25% of your grade. It is due Tuesday, Oct. 27, at 11:59 pm. Please submit your answers, except for part A, by email if possible. (If you need phonetic symbols, replace them by ASCII characters, explaining which sounds the characters represent.) Also tell me roughly how much time you spent on the exam.

The main reason that this is a take-home exam is to avoid the time pressure that goes with in-class exams. The exam is meant to be completed in about 3 hours. Please don't spend your whole weekend working on it. And please do not consult with others. I would also prefer that you do not consult written sources other than your textbook, the lecture notes, and any notes you took in class. If you do refer to other sources, however, please cite them. If you have questions, feel free to ask me.

A. Phonetic transcription (13%)
Transcribe the words read in class. If you don't know exactly which symbol to use, describe some of the features of the sound in question. You will need to turn in this part of the exam in class on Thursday, Oct. 22.

B. Phonology (25%)
All of the words in this section belong to the same language (Rio de Janeiro Portuguese). ASCII characters are used to represent all of the sounds. Use the same characters in your answers. The meanings of the characters are as follows: All other symbols are those used in class. (Recall that [x] is a voiceless velar fricative, and "'" represents stress on the following syllable.)

  1. The following words contain the phones [t, d, c, j]. (The meanings are left out since they are irrelevant.) For the pairs [t,c] and [d,j], decide whether they are allophones of a single phoneme (i.e., there are two phonemes for the four phones) or separate phonemes (i.e., there are four phonemes in all). Explain your answer. If the pairs consist of allophones of a single phoneme, give a rule which accounts for the distribution of the allophones. The rule should be general enough to apply to both pairs. You can express your rule in formal notation or in English. 
    ta'bladu   
'mwi~tu    
'fawt@     
'tEx@      
@~'dax     
'dram@     
'dOn@      
a'm@~ci    
ci~bri
ko~ci'ne~ci
'noyci
ji'fisiw   
'@~jiS
sa'uji
    [c] and [j] occur only before the high front vowels [i] and [i~] (though we don't have any examples of [ji~]). [t] and [d] occur before other vowels and apparently before some consonants (at least we have [dr]). This [t] and [c] are in complementary distribution, as are [d] and [j], and each pair represents two allophones of a single phoneme. Since [t] and [d] occur in the most environments, they can represent the basic allophone. We then only need rules to change [t] to [c] and [d] to [j] when the come before a high front vowel. The following rule will work for both cases:
    A dental stop becomes an affricate when it precedes a high front vowel. (In this language, [t] and [d] are dental rather than alveolar as in English.)

  2. The following words contain the phones [i, i~, u, u~]. (The meanings are left out since they are irrelevant.) For the pairs [i,i~] and [u,u~], decide whether they are allophones of a single phoneme (i.e., two phonemes for the four phones) or separate phonemes (i.e., four phonemes in all). Explain your answer. If the pairs consist of allophones of a single phoneme, give a rule which accounts for the distribution of the allophones. The rule should be general enough to apply to both pairs. You can express your rule in formal notation or in English. 
    'uzu
'u~Zu
u~
ku
'sujitu
'ku~plisi
'kuSpi
'kilu
ki~'ji~
mi~
i~vi'zivew
'i~petu
'ipiku
    For the pairs [u,u~] and [i,i~] we have the near-minimal pairs [uzu -- u~Zu] and [i~petu -- ipiku]. All four phones can appear word-initially or word-finally, word-internally following or preceding consonants at various places and manners of articulation (though there is only one nasal consonant in the list). Based on the information we have, we must conclude that the environments for these phones are overlapping, if not identical, and that they are allophones of four distinct phonemes.

  3. The following words contain the phones [r,x]. Describe the relation between these phones in the language as completely as possible, including any rules that are necessary to account for their distribution. In this case, morphology is relevant, so the meanings are included. (Note: there may be more than one way to do this.) 
    xapa'rig@     'girl'
'ix@          'damn!'
xe'gax        'to sprinkle'
'xiZu         'tough'
'xoStu        'face'
xu'i~         'bad'
ko'xidu       'quick'
sox'tudu      'lucky'
ax@'zax       'to devastate'
'traSci       'piece of junk'
a'rax         'to plow'
'frag@        'crag'
'furu         'hole'
'prEgu        'nail'
'ir@          'anger'
ko'riz@       'runny nose'
pox           'by'
pox ma'ri@    'by Maria'
pox te'rEza   'by Teresa'
por a'lisi@   'by Alicia'
por e'lEn@    'by Elena'
ji'zex        'say'
'awgu         'something'
ji'zer 'awgu  'say something'
'Seg@         'enough'
ji'zex 'Seg@  'say "enough"'
    There is a minimal pair in the list ([ir@ -- ix@]) so we must conclude first that [r] and [x] are allophones of distinct phonemes. But the environments of the two phones seem to be far from identical. In fact the only environment they share is the one between two vowels. [x] can occur word-initially and before consonants; [r] cannot. [r] can occur after consonants; [x] cannot. The morphological evidence in the last ten examples also implies that one or the other of the phonemes has two allophones: [x] and [r], that is, that one of the phones belongs to two phonemes. Thus the words for 'by' and 'say' should each have one underlying representation. There are two possibilities: either these morphemes end in an underlying /r/ or an underlying /x/. If the underlying segment is /x/, then we need a rule to change it to [r] when it precedes a vowel. But [x] can precede a vowel, so there could be no such rule, unless it makes reference to the word boundary. If the underlying segment in the words for 'by' and 'say' is /r/, then the rule would change it to [x] when it precedes anything other than a vowel (a consonant or the end of the utterance). Since [r] seems never to occur before consonants, this alternative is preferable. Therefore there are two phonemes: /x/ and /r/. We know of only one allophone for /x/: [x]. /r/ has two allophones, [r] and [x], the latter whenever it appears before anything other than a vowel. The underlying representations of the words for 'by' and 'say' are /por/ and /dizer/.
C. Phonetics and phonology (25%)

  1. What is unnatural about the following hypothetical languages?
    1. Voiceless stops become voiced when they precede /d/, /g/, /v/, or /n/, but not when they precede /b/, /m/, or /z/.
      The environment for this rule is not a natural class for a language containing these consonants. (Note that the process of voicing stops before voicing consonants is not unnatural. It is straightforward assimilation.)
    2. The consonant phonemes are /f/, /s/, /x/, /l/, and /w/.
      This language has three fricatives but no corresponding stops. In general we expect the more common counterparts for less common sounds.
    3. The consonant phonemes are /p/, /m/, /d/, /s/, /k/, /g/, /y/, /r/.
      There are a number of "holes" in this consonant inventory. There are four places of articulation. For the bilabial position there is a voiceless stop and a nasal, for the alveolar, a voiced stop and a fricative, and for the velar, a voiceless stop and a voiced stop. For the palatal position there is only a glide. The system is highly asymmetric.
    4. The vowel phonemes are /i/, /u/, and /o/.
      We expect the vowels in a language to be distributed fairly evenly through vowel space to maximize their distinctiveness. In this language all of the vowels are cluster in the upper back half of the space. There are no low vowels at all.
    5. In the history of the language all voiceless stops became high vowels.
      Voiceless stops are about as different from vowels as is possible within the pronounceable phones. A change from a voiceless stop to a vowel would involve the change of several features. Voiceless stops might change to voiced stops or voiceless fricatives but not vowels.

  2. What is wrong with each of the following statements?
    1. Voiceless stop consonants can be identified acoustically by examining the frequencies of the formants that occur during the production of the consonants.
      There are no formats during the production of voiceless consonants. They are normally identified by the transitions in the formants of the vowels on either side of them.
    2. Different vowels are distinguished from each other on the basis of the frequency of the opening and closing of the vocal folds.
      The frequency of the opening and closing of the vocal folds (the fundamental frequency) determines the pitch of the vowel. Different vowels are distinguished from each other by their formant frequencies, which are harmonics of the fundamental frequency.
    3. [k] and [g] differ with respect to the relative timing of the start of voicing and the closure between the tongue body and the velum.
      Voiceless and voiced consonants such as [k] and [g], when they appear at the beginning of a syllable, differ in voice onset time, that is, with respect to the relative timing of the start of voicing and the release of the closure. When they follow a voiced segment, there is normally no break in the voicing for the voiced consonant. When they appear at the end of a syllable, they are distinguished by the relative timing of the closure and the end of voicing.
    4. For infants to learn to hear the distinctions between different linguistic sounds, they first have to learn how to produce those sounds.
      Infants can hear phonetic distinctions before they can produce them.
    5. The errors that infants make pronouncing words often involve the addition of consonants that adults do not produce.
      Infants' errors are much more likely to be deletions than additions. That is, they are simplifying.
    6. Phonological change in the history of languages is always in the direction of simplification.
      Phonological systems can become more elaborate as well as less elaborate. For example, a language can split a single phoneme into two phonemes, as happened when /f/ became /f/ and /v/ in English.
    7. Phonetic change in one phoneme has no effect on the other phonemes in the language.
      A phonetic change in one phoneme may cause it to resemble another phoneme, which may then change to make it more distinctive. We see this sort of process in the Great Vowel Shift in English.
    8. The underlying representation of a morpheme should include the appropriate allophone for each segment in the morpheme.
      The underlying representation for a morpheme should include only phonemes. The appropriate allophone for each of these is derivable from the phonological environment so does not belong in the lexicon.
    9. When there are two or more possible pronunciations for a word, one is always right, the others wrong.
      Two or more pronunciations may be possible because of dialect differences or because of free variation. In neither case is there any way in which one or the other pronunciation is "wrong".
    10. To establish that two phones are separate phonemes, we must find a minimal pair in which those phones make the only difference between the words.
      It is not necessary to find perfect minimal pairs to establish that two phones are separate phonemes. It is enough to find enough near-minimal pairs to convince ourselves that the phonetic environments of the two phones overlap.
D. Linguistics in general (25%)
For each of the following claims, say whether a linguist is likely to have said it. If not, explain why. If so, say what the basis of the claim is and whether the claim is controversial, that is, whether some linguists or other language scientists might oppose it.
  1. The spelling of words can sometimes give us clues about what make up the phonemes of a language.
    Yes. Spelling never does a perfect job of representing phonemes, but the inventors and the users of spelling systems usually base them unconsciously on their knowledge of the phonemes of the language, so they can help us in determining what the phonemes are. This is not controversial.
  2. To investigate the grammar of English, we can ask English speakers whether particular sentences are "good English."
    Probably not. To investigate the grammar of a language, we may ask for grammaticality judgments by native speakers, but we should probably avoid expressions like "good English" because they may lead speakers to base their judgments on prescriptive norms.
  3. Speakers may produce errors for various reasons. For example, they may not have been taught the correct form so may not know it.
    No. Speakers may produce errors for various reasons, but not knowing a particular form is not a possible reason. If they don't know a form, it is not part of their competence (grammar), so failing to produce it would not count as an error.
  4. The dialects of a language are best described in terms of how they deviate from the standard dialect.
    No. Normally the dialects of a language all descended from a common ancestor dialect. We could profitably describe how each deviates from the ancestor, but treating one modern dialect as a standard for the others will not usually provide any insight in describing the dialects.
  5. A Japanese child is able to learn Japanese because the properties of Japanese are encoded in Universal Grammar in the brain of the child at birth.
    No. UG does not include the properties of specific languages but rather the properties that all languages share and the possibilities for particular grammatical types of languages.
  6. Language learning is not really something the child does; it is something that happens to the child placed in an appropriate environment, much as the child's body grows and matures in a predetermined way when provided with appropriate nutrition and environmental stimulation.
    Yes. Some linguists, those who support the innateness hypothesis, would subscribe to the idea that the language facility is a kind of organ like the legs and the liver which does not require special attention because its basic properties are built in. Others would disagree with this position, arguing for a more active role for the child and a greater role for the environment.
  7. Parents can improve the speech of their children by correcting their mistakes.
    Probably not. There is little evidence that correction of mistakes has any effect on a child's grammar (though there have been arguments that correction at the appropriate time can have an effect).
  8. Speech errors tell us nothing about the underlying knowledge that speakers have of their language.
    Yes. Many linguists would support this position. Speech errors reflect problems in production so tell us something about performance but not about the underlying competence, which is what linguists are after. Others, especialy some psycholinguists, would disagree, arguing that constraints on the types of errors that people reflect the way their knowledge of the language is organized.
  9. A native speaker of English who said the following would have made a speech error: "Someone's at the door. See what they want."
    No. A prescriptivist might quarrel with the use of they, a plural pronoun, to refer back to someone, which is singular, but this usage is the norm in conversational English in most dialects, so it would not constitute a speech error.
  10. A native speaker of English who said the following would have made a speech error: "Someone's at the door. See what they wanting."
    Yes. In all of the dialects of English that I'm familiar with, this would constitute a speech error. A speaker who produced this would very likely recognize that they had done something wrong and might be expected to correct it. There is nothing controversial about this.
  11. It is possible for the grammars of two languages to resemble each other in a very general way even if the languages are unrelated and have not been in contact.
    Yes. Two languages can resemble each other in a general way if they belong to the same grammatical type. For example, Japanese and Turkish have many similar properties. There is nothing controversial about this.
  12. The grammars of languages are in the minds of their speakers.
    Yes. Most linguists believe that speakers have grammars in their minds/brains, that is, that they have an explicit representation of the regularities in their language. But linguists differ in terms of how far the descriptive grammars that they come up with are meant to look like these mental grammars.
  13. English is spoken less precisely in the American South than in other parts of the US or in Britain.
    No. Linguists never use value-laden words like "precise" to describe the phonologies of languages.
  14. Language takes place in time, and we cannot understand how language works without taking the detailed temporal aspects of language seriously.
    Yes. Most linguists would probably disagree; they care only about sequential time. But a few linguists, and many phoneticians, do believe that the timing of speech reflects basic properties of language.
  15. The way words group together in language has nothing to do with the way objects and their properties group together in the world.
    Yes. This is one sense in which language is iconic. In most languages words group together in phrases, and these phrases are about objects, properties, or relations in the world. The words in the phrase refer to different aspects of the objects, properties, or relations which also group together in the world. This is probably not controversial, though many linguists would not find it very illuminating (but you couldn't have been expected to know that).
E. Properties of human language (12%)
An animal is discovered (call it a "vazz") which has a communication system with the following properties. Describe the vazzes' "language" as thoroughly as you can, and tell how it is similar to or different from human language.

What follows is most of what I could think of. You would not have been expected to come up with this many points.