Prelim Answer Attempts

This is a page created in 2004 by students preparing for the prelims.  You are welcome to use it and modify the obvious mistakes.  The only condition is that if you use it and do problems not on it, you must add your answers to the page for future reference.

Selected Topics

Fall 2004 / Spring 2004

Fall 2003 / Spring 2003

Fall 2002 / Spring 2002

Fall 2001 / Spring 2001

Fall 2000 / Spring 2000

Fall 1998

Fall 2004

General / Historical / Phonology / Grammar

General

sample answers: Meylysa

Spring 2004

Historical / Phonology / General / Grammar

Historical - Part I - Part II

Historical - Part I

All problems common to establishing genetic relationship among languages and to subgrouping languages which are known to be genetically related.

- Using sound correspondences justifies your claims

- When finding a cognate set you have to factor in universal tendencies (semantic, phonological (*s>h), etc.), borrowing, chance and drift (parallel development)

- Grammatical and morphological information can complement our grouping judgements

What special complications occur in establishing genetic relationship that do not arise in subgrouping languages and vice versa.

In finding genetic relationships, you have to find a way of choosing which languages to compare, perhaps look at geography.  Or perhaps look one by one at all languages.  In any case, you'll spend a lot of time just deciding which languages are related and which ones aren't.  You can try mass comparison, suggested by Greenberg (1957) which he sites back to 1738, of looking at a few similar words over a large number of different languages, the method does not use recurrent systematic sound change, only phonetic similarities.

Sub-grouping is only successful if genetic affiliation is clearly established beforehand.

In sub-grouping you have to look for shared innovations, not on shared retentions.

However, in sub-grouping, if you find a form in one language which is not attested in any other language, you have to decide whether it was borrowed, or if the form from the ancestor of the other languages was borrowed.

Lexicostatistics has been applied to sub-grouping, it cannot be applied in establishing genetic relationships.

Problems arise in sub-grouping when confronted from dialectal chains (no non-arbitrary way to split languages into dialects)

If the time between language separation is quick, then sub-grouping will be more difficult, since there will be a lack of shared innovations.

Historical - Part II

1) What kind of information can be gleaned by the investigation of semantic change?  Two types.  What is the info and how is it motivated?

Universal tendencies - 1) what are they? a) color term implicational hierarchy (Berlin and Kay 1969), b) plant and animal term implicational hierarchy (Cecil Brown 1977), c) grammaticalization of verbs to become tense or modal markers ("I'm going to go."), d) semantic extension of body part terms (Brown and Witkowski 1981) (pupil of the eye > human or humanlike objects or beings), d) semantic extension of gender words (woman > sexually derogatory words or love and nurturing) 2) what motivates? Universal human biology (perception of color), the ability to use metonymy to extend meanings.

Cultural change - 1) what are they? changes in burial (the word formally used to mean cave now means grave), migration (the word for porcupine fish is used for durians by inland speakers), technology (dial the telephone)...social movements (women's rights movement  induced use of chair for chairman, etc Cameron (1995, 1997)2) what motivates? Cultures are always changing, so people must constantly think of new ways of saying new concepts.  Often times they use words they already have in their vocabulary through the process of semantic extension.

All these semantic changes can be used in comparing languages and looking for cognates.

2) Discuss the concept known as a linguistic area.

Sprachbund or linguistic area - a diffusion of linguistic traits across genetic boundaries.  It's often marked on maps with lines called isoglosses.  Within the line is a certain feature, past the line the feature disappears.

How can it be recognized - You find sound changes, word order, morphemes, syntactic features, etc. which are shared by languages in the same geographic area, but not all of these languages should have this feature if looked at strictly from their genetic relationships.

How can areal features be distinguished from genetic relations? - You first figure out if the languages are genetically through comparison of basic terms, and other relevant cognates.  If there are similarities which exist for languages within the same geographic region, then we must be wary of whether they are areal features or not.  If upon investigation we find the languages in the region are not genetically related to one another, and the change is not expected if looking only at genetic relationship evidence, then you can conclude that they are areal features...Sometimes areal features are more restricted in scope, such as loss of nasalization in languages of the same region (Salishan...), so they are not as hard to differentiate from sound correspondences which reflect real genetic relationships.

How can the sources of areal features be determined?  You look at the protolanguage for all languages with the feature and try to find which language/s had the feature earliest.  If you find that the feature wasn't present in any of the languages' protolanguages, then you would have to assume it was borrowed from outside the language families.

Phonology

Phonology - Part I - Problem

retroflex -> non retroflex / [+retroflex] Vo _

General - Part I - Essay 1 / Essay 2

Essay 1) The notion of 'word' is pervasive, but a good definition is elusive...

I. Difficulties in defining the notion of the word

    A. Morphology

        1. English - words seem to be easy to discern when we look at the spaces inserted between them, spaces which do not exist in languages like Chinese.  However, what Microsoft Word counts as words and what is actually listed in the dictionary is different.  Dictionaries will not include entries for run, running, runs because these words are considered to be the same word.  Inflection is productive and not listed for verbs.  However, words with derivational affixes may or may not be listed (beauty, beautician, space, spacious) since they are not necessarily regular for all parts of speech.  Thus, though no-one would claim that -ian is a word, some might claim that beauty and beautician are two different words.  In addition to these types of affixes, we also have cranberry morphs, such as the cran found in cranberry.  Since berry occurs with blue-, black- and so on, we think of cran as being the morpheme which distinguishes for us what cranberry means.  However, up until recently it only occurred in this one word, so is cran a word?  Most people wouldn't think so.  Are articles words (the and a)?  some are considered clitics (though you can't put stress on clitics).

        2. Chinese - No spaces between words, Words can be one or two characters, most are two characters though two two character words can be combined to form one two character word.  Each character usually means something, even if combined with another character to form a word.  Chinese has two types of dictionary, one for characters and one for character complexes.

        3. Sign Language - many words are composed of meaningful parts, are those parts words?  Ex. Pointing to your hand means "hand", so is this action as a whole "hand", or is it a predicate indicating "hand"?

        4. Agglutinating and fusional languages combine morphemes together.  Fusional languages end up combining morphemes together in a portmanteau fashion.  Polysynthetic languages sometimes create one word which has the same function as a sentence would for an isolating language.  Does this mean all of the morphemes in a polysynthetic word complex are words?  Does this mean Fusional morphemes which encode multiple meanings are words?

Comment from John Kupchik - Agglutinating languages (which are synthetic by definition, but not polysynthetic!) do not create a word that has the function of a sentence in an isolating language, unless it's an extremely short intransitive sentence such as "I went" in which you have pro-drop (for example Japanese "it-ta" (go-PERF),
Finnish "men-i-n (go-PAST-1Sing"), but the key point is you cannot have an S, O and V all in one word in an Agglutinating language. Only polysynthetic and incorporating languages have structures in which a verbal stem combines with nouns, adjectives, and other parts of speech into a blurry "word-sentence" hybrid which can consist of an S (an overt Subject is language specific, I believe), O, V (and other parts of speech), along with a pro-drop encoded Subject (in other words verbs can incorporate their objects and adverbial modifiers, while nouns incorporate modifiers freely). Such constructions are usually triggered by what some linguists call the "gate crusher" (for example, in the paleosiberian language Chukchee), which is a morpheme at the start of
such a construction that has no purpose other than to signal a polysynthetic or incorporating structure is beginning (there is also usually a morpheme that signals the end of such a construction, though I don't remember it having such a fancy name). The difference between polysynthetic and incorporating is polysynthetic only allows 1 verbal stem in a construction while incorporating languages allow 2 (or more?) verbal stems. Incorporating languages are the rarest typologically, and are an extreme (rather berserk) extension of Polysynthesism (which itself is just a more complicated extension of agglutination involving incorporation). All polysynthetic languages also allow
normal agglutinative constructions, because they are inherently agglutinative and this is their underlying morphosyntactic structure. Semantic restrictions limit polysynthetic constructions, for example it is impossible to say "We saw us/me in the mirror" in polysynthetic/incorporating languages because in such constructions the Subject cannot be the same as the Object (this is apparently a universal feature of such languages--they must resort to normal agglutinative syntactic constructions to express such a meaning). Also agglutinating languages do NOT have portmanteau morphemes, I'm not sure but I'd think neither do polysynthetic/incorporating languages (as they are inherently agglutinative). The best tests for "word versus sentence" test in polysynthetic/incorporating languages would be 1)phonological word test (see if the entire structure has one main stress or multiple ones), and 2)foot structure test (see if feet are assigned across the entire structure or separately within constituent morpheme chains in the structure).

    B. Semantics

        1. No one meaning one word = To kill in English is two characters in Chinese, sha si 'try to kill' and 'dead'.  But sha and si both exist as separate one character words, and most linguists would consider this to be a resultative construction, and not a word.

        2. word meanings are interpreted by a culture differently, and may include or not include notions of causality, result, inception, etc.

        3. Idioms have only one meaning, but are composed of many words "kicked the bucket".

        4. I went to the bank and he did so to.  "so" doesn't mean anything but an entire phrase.  Pronouns and deictics are only meaningful from context.  or "Terrorists are working hard to find ways to destroy America and so are we."

    C. Phonology

        1. A lot of phenomenon occur at syllable boundaries and in relation to the stress pattern, doesn't help us define what a word is.  Although, words seem to try to fit into one foot in English, say John F. Kennedy.

    a. Example: voiceless consonant aspiration in onset position (rule is complicated)

    b. in english fake geminates only at morpheme boundaries (English doesn't have true geminates)

II. Overcoming the difficulties

    A. Native intuition - most speakers have a notion of what a word is which is based on prototype effects, what is a typical word, what is an atypical word, etc. 

    B. Morphology - if you have spaces between words use them. Perhaps free morphemes can all be considered words.

    C. Syntax - Different parts of speech tend to have their own characteristics, allowing us to differentiate words from their markers, and categorize them together.

            1. a. Nouns - take determiners, plural markers

                b. Verbs - take tense and aspect

                c. Adj - take modifiers like "very" or "almost"

            2. They can also act as constituents, where you can substitute words of the same grammatical category for one another. 

            3. words act either as Head or Specifier, though clitics act as both

    D. Processing - there have been studies of priming effects that show people may store units such as woman and women as different words, and do not have one lexical entry in their head.  This would mean that in terms of processing, we may store a lot of redundant information.

    E. Phonology - word, morpheme and sentence boundaries often have special phonological phenomenon,

        a. Word test - if you add a suffix, or morpheme (free or bound), and the stress pattern for the new complex is the same as that for the original word, then it's considered a word ( a white house vs the White house, blackbird vs. black bird)

        b. English stops are unreleased at the ends of words

        c. flapping, devoicing, reduplication etc.  at edges of words but not word internally (exception for flapping, I'll see you tomorrow)

        d. pausing between words more likely than pausing between morphemes???  One primary stress per phonological word.  If you're pausing between morphemes you lengthen, whereas if you pause between words, you can have complete silence (I saw the bee....keeper)  foot structure is within the word boundary, not within the morpheme boundary.  feet are assigned based on the word domain.

III. Critical assessment of how useful the notion is

   A. Useful in making dictionaries

    B. Native Speakers have a sense of what a word is

    C. Could be the basic level of vocabulary for us, which means words are easier for people to remember, and the first things that children would learn, classify words together into semantic, grammatical classes allows us to learn them better.

    D. Even though the notion of word is fuzzy, there are certainly many unambiguous examples in isolating languages that give us a starting ground for analysis.

Essay 2) Similarities and differences between first and second language acquisition

Children have no language when they start learning their first language/s (Generative linguists say that learning is perhaps unconscious), while adults do.

    Thus, 1) children are more motivated to learn than adults (there's more at stake if kids don't learn language), 2) adults have interference from L1 (first language).

Children spend more time learning their first language than most second language learners.

    Thus, 1) children have greater exposure and learn to fluency (though they can make their own languages with no exposure, except to social interaction, Nicaraguan sign language), 2) second language learners may or may not have a lot of exposure and may or may not learn to fluency.

Children are expected to not know language, and thus given encouragement and taught explicitly (not in all cultures, some cultures kids don't get taught language specifically, kids may or may not understand corrections from adults, they don't perceive they are making a mistake).  Adults may not be expected to know the language, but other adults will not correct other adults as explicitly as children due to sociological pressures.  Adults may be more afraid to start talking, afraid to make mistakes.

    Thus, children will possibly have a better learning environment for language than adults.

Generative grammarians sometimes claim that L1 is learned by an innate language acquisition device while L2 is less so.  They posit that kids have no negative evidence (with parents telling them not to say certain things).  Whereas adults often get negative evidence in the classroom.

    Thus, children will learn better than adults.

Children sometimes say things they never hear, over generalize tense markers, grammatical category usages, etc.  Adults do as well, though perhaps in different ways depending on L2 interference.

    Thus, children and adults go through similar learning strategies when learning languages, though there may be differences which are hard to prove since L2 must be factored out.

When children learn language they are also building up other cognitive abilities, whereas adults aren't.

Critical period for first and second language (?) learning

Input for children is natural conversation whereas input for adults is often classroom instruction.

Grammar - Part 1 - Essay

1) Discuss the notion of "subject" in ergative languages.

Ergative languages and nominative-accusative languages differ in their grammatical encoding of subject.  In nominative-accusative languages, the subject of a transitive sentence and the subject of an intransitive sentence are both marked in nominative case.  The object of a transitive sentence is marked in accusative case.  In addition, if a language marks nominative case then the language also marks accusative case.  However, if the language marks accusative case, it doesn't necessarily mark nominative case.  An example of this can be seen in Hawai'ian:

Ua   'ai ke   kanaka i    ka  poi.

Perf eat the man     Ac the poi

'The man at the poi.'

In the sentence above, the object of the transitive verb is marked in Accusative case, but the subject is not marked for Nominative case.

An example of an intransitive verb also not being marked for case is given below:

Ua  hele ke kanaka i Maui.

Perf go  the man     to Maui

'The man went to Maui.'

English marks case according to where the subject and object are located in the sentence.  Thus, subjects occur before the verb and objects occur after the verb.  In English, the subjects of transitive and intransitive verbs all occur before the verb, whereas the direct object occurs after the verb.  There are of course exceptions to this ordering, such as passivization, but this requires that the verb be marked by the passive.

In ergative languages, the subject of an intransitive verb and the direct object of a transitive verb are marked by the same case.  This is called the absolutive case.  The subject of a transitive sentence is marked with a different case, called the ergative case.  If a language marks the absolutive case it also marks the ergative case.  However, if the ergative case is marked this doesn't necessarily entail that the language marks the absolutive case (Croft 1990:105; Dixon 1994:11).

An example of an ergative language would be West Greenlandic (Manning 1996:3)

a. Oli-p      neqi          neri-vaa

    Oli-Erg  meat.Abs  eat-3Sg.3Sg

    'Oli eats meat.'

b. Oli sinippoq

   Oli.Abs sleep.3Sg

   'Oli sleeps.'

In the above sentences the subject of a transitive sentence is marked with ergative case, a 'p' suffix.  The subject of a transitive sentence and the object of a transitive sentence have no morphological marking, and are thus in the absolutive case.

It is important to note that many languages are somewhere between strictly nominative-accusative and ergative absolutive.  This is dependent on whether they consider the patient and the subject of intransitive sentences to be topic, or the subject of transitive and intransitive sentences to be the topic.  In addition, resultative constructions will tend to be ergative in nature (Comrie 1981: 113) and imperative constructions will tend to be nominative in nature.

a) What is noun incorporation? Discuss if and how it is different from compounding.

noun incorporation - house-paints, places the object bare noun, or a classifier for the direct object, into the verb complex, forming an intransitive verb.  You cannot incorporate nouns from within a noun phrase *I house-paint yellow.

compounds? house-sit, loudspeaker, blackboard, high school, carpool, compounding may be more common across languages than noun-incorporation

b) Discuss the following data in terms of transitivity.

The prototypical transitive sentence has an agentive subject and a theme direct object which is visibly affected by the action denoted by the verb.  Looking at the sentences below, only a) has a direct object, which automatically makes it more transitive than b) or c).  In addition, when the sentence is made passive, only a) and b) are grammatical.  This shows that in c), Mary is not affected by an agent.  However, in a) and b) she is affected by John.

Degree of volitional agency is also a factor, with a) being less volitional than b), and b) being less volitional than c) (since you can visually see the results).  This would lead us to conclude that c) is the most transitive of the three.

Degree of affectedness is also a factor.  Thus, in a) through c) the direct object isn't very affected, making all three seem more like intransitives.

Individuation of the theme can also be important for establishing degree of transitivity.  Thus, a definite theme may indicate higher transitivity.  In this case, all three sentences have the same direct object, so this test offers no means of comparison.  However, since the noun is a proper noun, it makes the sentences more transitive than if the direct object were birds.

The perfective aspect is more associated with transitivity, where as the imperfective is associated with intransitivity.  All sentences are past tense, thus offering no comparison.  However, past tense probably makes the sentences more transitive than intransitive.

Finally, negation also reduces a verb's transitivity.  The verbs in the following sentences are not negated, and thus are more transitive than their negated counterparts.

a) John heard Mary.

Mary was heard.

b) John listened to Mary.

Mary was listened to.

c) John went to Mary.

*Mary was went to.

Fall 2003

Phonology / Grammar / General / Historical

Phonology - Part 1 - Problem

1. vowel consonant phoneme charts

h = velar n

N = all capitals are palatalized versions of the consonant

F ß q £b - bold signifies that these fricatives are palatalized.

J = voiced ç

vowel phoneme chart

consonant phoneme chart

In the above chart I assume that though c, J, ç and J are usually central, here they are velar.  This does not mean that they don't become more central than their non-palatalized counterparts.  However, for the purposes of phonological categorization, I assume the central to velar region are all one place of articulation for these speakers, here marked velar.

2. underlying forms of morphemes

3. rules a) formally, b) informally C and V, c) prose

Liquid-Pal - a) $    Root [+approximate]                              Root [+cons}                $

                                Place of Articulation----->[+palatal]<----Place of Articulation

b) $ Liquid -> [+palatal] <- C $

c) Liquids share the palatal feature of the consonant to the right of them within a syllable, ex (9), (10) and (11).

Liquid-Cont - a) $    Root [+approximate]                   Root [+cons}     $

                                   Manner----------->[+cont]<----/---Manner

b) $ Liquid -> [+continuant] <-/--C $

c) Liquids take the continuant feature of the consonant to the right of them within a syllable, ex (9), (10) and (11).

V-Feature - a) [+palatal, +cont] -> [+syl, +hi, +front] / [+syl] _ #, b) [+palatal,+cont] -> i / V _ #, c) The [+palatal] and [+cont] features are realized as palatal i, i, when it is preceded by a vowel, ex (2), (3), (12), (20).

VCV-Feature - a) [+palatal, +cont] -> [+cons, +hi, +front] / [+syl] _ [+syl] , b) [+palatal] -> j / V _ V , c) The [+palatal] feature becomes j intervocalically, ex (2), (3), (12), (20).

C-Float-Feature - a) ROOT [+cons]

                                   Node           

                                        ->  [+feature]

b) C -> [+feature]

c) Consonants take floating features located to their right, ex (1), (4)-(11), (13)-(19).

S-to-H - a) [+alveolar, +cont, -voice, +palatal] -> [+glottal] , b) palatalized s -> h  , c) s becomes h when palatalized, ex (13), (15).

4. ordering

1) C-Float-Feature, 2) Liquid-Pal, Liquid-Cont, 3) S-to-H

5. evidence for 4.

C-Float-Feature before Liquid-Pal, Liquid-Cont - C-Float-Feature must occur before Liquid-Pal and Liquid-Cont because the incorporated [+palatal] and [+continuant] features of the word final consonant feed these two rules in (9), (10) and (11).

C-Float-Feature before S-to-H - Otherwise 'ladle' in (13) and 'fish' in (15) would not change.

6. derivations 'church', 'ladle', and 'net'

7. interesting or puzzling data/analysis

Number (4) 'bowl' has the same form in the plural and the diminutive.  This seems to be because the diminutive suffix, precluding the [+palatal] feature, has already been lexicalized into the singular form.  To explain this according to the Obligatory Contour Principle (OCP), to create the diminutive the diminutive suffix is not added to the singular form since this would be adding it twice, thus causing too much repetition.  To explain this in terms of Evolutionary Phonology, after speakers started using the diminutive form as the singular, the old singular form was lost.  The diminutive and the plural forms don't change (the rules creating them for this word are no longer productive, the forms are memorized).  The [+palatal] feature may have been initially borrowed as well, but then lost through analogy with other singular forms.  Or, it may have never been borrowed at all, again due to analogy with other singular forms.

Another fact more puzzling is how s becomes h when palatalized.  Why doesn't it become ç like x?  Perhaps over time palatal s has undergone weakening to become palatal h.

Finally, the fact that the suffix has the feature [+continuant] that can be attached to the liquid consonant in the coda, as opposed to the word final consonant, is also interesting, ex (9), (10) and (11).

General Fall 2003 Part I - Essay

Set One A / B and Set Two C / D

Set One A

Assumptions

    -Assume there is an underlying explanation (or underlying representation) for what your field is investigating.

    -Assume that if you don't follow the theoretical rules, constraints, processes (whatever) you will create         ungrammatical language.

    -Assume that there are universal rules, constraints, processes, etc. for all languages.

    -Assume that these areas are somehow core areas of understanding/investigating our language ability.

What do these assumptions tell us about language?

    -Aren't these assumptions created from what we know about language?

    -Humans have similar language faculties/brain.

    -Language is a system (Saussure says "La langue est a systeme ou toute sur tient" Language is a system where everything holds together.  Say you have an arch, if you remove one stone the whole thing will fall apart.  Chess game - words on their own have no meaning, you only get meaning from context, within a system.)

In what ways are they artificial and thus misleading?

    -Universals are usually tendencies, and not really universals (ex. OCP, Subjacency)

    -You can use a part of a language (phoneme) lost and still have language (old theory according to Saussure was that after we lose a phoneme, we get chaos, then stability...but what were we speaking during the chaos period).  Same for changing from nominative-accusative to ergative-absolutive, losing grammatical classes, etc.

    -Spend too much time on formal theories that don't work, or explain language.  EXAMPLES 1) OT restates universals as constraints, when these constraints were based on what we saw as typological tendencies, thus didn't explain why we have these tendencies.  Evolutionary Phonology explains better by showing how phonetics influences phonology. 2) Generative grammar and emergentism (processing constraints). 

Set One B

What is the motivation for the belief that language users use abstract generalizations?

    -Children acquisition shows that they can subcategorize words into abstract grammatical classes such as nouns, verbs, etc.

    -People store phonemes, that's how they think about the sounds they make.  Archiphonemes, language users will have nasals which assimilate to their environment, representing a big N???

    -the ability to categorize meaning into classes, such as colors, kinship terms/relations, causation, transitivity, thematic roles

    -words are abstract categories, "tree" represents all trees, even though they don't all look alike

    -easier for us to learn and process if we categorize, generalize

    -Models such as GB, OT, Construction Grammar, MP, Underspecification Theory, Formal Logic, are all models which place abstract generalizations on language. 

Alternative

    -Evolutionary Phonology, Natural Phonology, Emergentism, Connectionism.

Set Two C

3 or 5 design constraints on a message you would send to aliens.

-Variety of languages, Send several kinds of texts, morse code, video tapes of human interaction, spoken and signed, recordings of friendly dogs, birds, whales (Star Trek 4)

-Make the sentences simple transitive and intransitive sentences with concrete agents and patients.

-Make the semantics relevant to enclosed pictures perhaps. 

-Make it pragmatically relevent, after a hello, then introduce ourselves, tell them we received the message and we don't understand it.  Tell them we will continue trying to understand but we have enclosed our own languages from our planet for them to translate.  End with an acknowledgment of the receipt of the message and a friendly goodbye.

Charles Hockett 9 Design Features for human language

1) Language is a mode of communication

2) Semanticity - the signal has some meaning

3) Pragmatic function - all systems of communications serve some useful purpose

Some communication exhibit

4) interchangability - the ability of individuals to both send and receive messages

5) cultural transmission - some aspect of communication has to be learned through communicative interaction with other users of the system

6) arbitrariness 

7) discreteness - complex messages are build out of smaller parts

8) displacement- communication of things that are not present in space or time

9) productivity

Set Two D

ANIMAL COMMUNICATION

Phonetics - animals communicate with their bodies, with humans using their hands or vocal tracts.  In addition to humans, apes, dogs, chinchillas, quails, and many other animals use movements of their bodies and vocal tracts to communicate.  Ants secrete some kind of chemical, some female moths secrete a chemical when they want to mate, bees have special dances (three: round, sickle, tailwagging) which vary with region.  These dances convey distance, quality and amount of food to be gathered.  Crabs and spiders have simple communication systems (snapping claws) where they show other members they are from the same group.   Dogs urinate.

Phonology - studies have shown that chinchillas and quails can learn to distinguish phonemic distinctions, however, most animals don't seem to combine the sounds or motions which they make into more complex sound sequences.  Birds distinguish pitch.  Bees distinguish movement.

Morphology - bonobos, orangatans, chimpanzees: Gua - raised with child 30's (Kellog), Vicky - raised with child 50's (Hayes) Washoe (Gardner), Sarah - plastic chips language board '66 (Premac), Kanzi bonobo chimp (Savage Rumbaugh) Nim Chimpsky -failed (Herbert Terace), Koko '72 (Patterson) gorillas can be taught words and simple sentences with concrete objects and subjects. No inflectional morphology, combinatorial morphemes.  Limited vocabulary (number of calls, dances).

Syntax - no productivity, no discreteness, a bird call doesn't discriminate between the order of the different pitches it hears.  Chimps were taught words and simple sentences only.  Bees only have three dances.

Semantics - animals only talk about emotions, finding mates, building nests and food, they don't talk about the past or the future, primates talk about danger from the air and from the ground, always concrete things, always present in the environment. Robins only hear high and low pitch and don't attend to the order of the signals, they just pay attention to the intensity of them which mean how much they want to defend what they have and make a family in a certain area. Monkeys convey simple messages, also just reactions to the environment.  Great apes have limited vocal tract, fewer than six calls.  Dogs urinate to mark territory.

Pragmatics - Animals seem to follow the same H. P. Grice's pragmatic maxims of 1) quantity - being informative, 2) relevance - be relevant, 3) manner - be orderly, brief and concise, 4) quality - be honest.

Animal Communication Project

Historical

Historical - Part I - Essay

Long Essay - Phonetics important to the historical linguist

Phonetic knowledge used to explain sound change - Sound correspondences are used to justify how we designate cognates, they are also used between cognates to establish evidence of genetic relations, subgrouping, to reconstruct protoforms.  The reason phonetic evidence is used, as opposed to semantic and syntactic, is that it has a greater variety of markers for which we can use in comparing languages.  For example, syntax only has a few basic word orders, SVO, SOV, VSO, etc.  These orders change often within a language's history, thus leading us to have difficulty using word order as a decisive means of establishing genetic relationships.  Phonetics on the other hand encompasses more than just three basic elements (S,V,O) and their combinations.  It includes patterns formed by various consonants, vowels, tone, suprasegmentals etc.  Thus, the possibilities for change are much greater, leading to a more unique outcome for each language.

When using phonetic data, it is important to differentiate common (natural?) vs. uncommon (unnatural?) sound change, for instance, you may have two languages with an *s>h change, however, this may not be valid when sub-grouping since it is a common sound change, thus it could have happened that both sound changes developed independently.  If the sound change is uncommon, then it is more useful for establishing relationships.  For instance, if we have a cognate undergo metathesis in selected languages, this would be stronger evidence for assuming that these languages have a common ancestor.  In addition, when constructing sound correspondences, we will not construct sequences of sound change when there is no phonetic motivation.  For instance, if we find a correspondence between r and k, we will assume that either there are many immediate steps (perhaps r > t > tj > k), or that the correspondence can be explained through other means (perhaps borrowing).

Phonology important to the historical linguist

drag chains?

Morphology important to the historical linguist

shared or borrowed morphological traits such as classifiers, noun complexes, verbal constructions, tense markers, etc. all can be used in classifying languages.  Morphology is a better marker than syntax.

Syntax important to the historical linguist

Since languages tend toward the grammatical pattern of the dominant language, this isn't as useful an indicator of historical relationship.  However, if a language undergoes an unusual syntactic change, then it may be important in classifying its daughter languages together.  Perhaps an odd post-position construction, or an odd typological verbal construction, etc. 

Semantics important to the historical linguist

In order to identify cognates we need to use both phonetics and semantics.  A cognate is a word of sufficiently similar phonetic form to another word and shares the same semantic origin with that word.  For instance, if we have the word apple in English, and the word afle in some other language, however, their meanings are different, they are not considered cognates.  They are only considered cognates  if their form and meanings are likely to be of similar origin.

In addition to finding cognates, semantic change is also of interest to the historical linguist since it sometimes follows predictable paths.  For instance, there are certain universal tendencies that can be used to establish cognate sets.  Also, cultural changes can be reconstructed through language change, and also used towards establishing cognate sets.

Another important function of semantics is when linguists establish which words are borrowed or not.  For instance, if words of a specific semantic class are all related to a seemingly distant relative, this is probably the result of borrowing.  An example of this would be the extensive use of French words in English for terms associated with government.  Basic vocabulary is also established on the basis of semantics.  For instance, words that are less likely to change are those relating to basic semantic concepts that most cultures already have, such as colors, kinship terms and food.

Lexical innovations are also good indicators of subgrouping.

Typology important important to the historical linguist

Isolating, agglutinating, inflectional languages.  In general agglutinating languages usually develop to form inflectional, then isolating languages.  Thus, if the protolanguage of the family under consideration is an inflectional language, and a possible daughter language is an agglutinating language, then it may be less likely that the language is really a daughter language.  Of course, this is dependent on other factors as well.

Sociolinguistics important important to the historical linguist

Labov in his Martha's Vineyard study and his New York study showed that sociological factors are very important to understanding language change.  Weinreich, Labov and Herzog 1968 spent a great deal of effort to call attention to the fact that language diversity exists in the present tense.  In addition, this language diversity is present an any period of time, and is the source of sound change.  Their main point is that the sound changes already exist in the community (and hence haven't changed), what's at issue is which of the sound changes are considered standard, or acceptable as the written form.

Deborah Cameron (1995, 1997) also shows that prescriptive grammar, and social movements and institutions are responsible for the bulk of language change, contrary to popular belief.  Thus, the women's movement brought about many changes in the use of words such as Miss, chairman, etc.

Understanding how taboo works in southern Papua New Guinea and Northern Australian tribes, we find that sound change can be quite abrupt, with the death of a community member bringing about a ban on speaking that person's name.

Child acquisition important important to the historical linguist

Children learn language and pass it on to their children.  Without them there wouldn't be any change.  Jakobson (1968) discusses how children first enter a cooing stage, then move on to babbling, then to a one word stage, a two word stage, then sentences.  The phonological segments and features that the child can produce or is likely to produce causes universals such as words for mother and father being very similar throughout the world, independent of cultural background.

Instrumental phonetics important to the historical linguist

Studies on phonetics in the laboratory have shown that people can often misperceive or misarticulate certain sequences of words.  Some of these tendencies are universal and some are found on the basis of L1 interference.  These studies help historical linguists understand what types of sound changes are possible and likely to occur for what reasons.

Casual Speech important to the historical linguist

Casual speech shows how sound change can occur.  In casual speech certain segments may be deleted or weakened, which could lead to explanations for sound change.

Interlanguage important to the historical linguist

Interlanguage?

Short Essay 1) - Homeland -

the area of greatest primary branches, linguistic diversity within a language family, is probably the homeland.  This is because the original dialect is located there, it is the principle of shortest move, where people will only move as far as they have to.  Thus, it is more likely that out of a family of 10 languages, one would migrate out and nine will stay behind, rather than 9 moving out and one staying behind.  For instance, using the Melanesia and Taiwan debate, Melanesia is only has languages descended from one primary branch of PAN, while Taiwan has many (9?).  Is it really likely that the origin of the language is in Melanesia, and that 9 out of 10 major families immigrated to Taiwan and no where else, while the one family that was left behind gradually migrated throughout the pacific?

Vortor and Sachen Principle - if you reconstruct a word to the protolanguage, then that the thing which that word stands for must have been present during the time the original language users lived in their homeland.  So, if you reconstruct *rice, then the homeland must have been in a region where they grow rice.

Finally, you look at archeological and historical evidence.  What are the likely patterns of migration?  Do the tribes have similar means of building tools or constructing burial sites, etc. 

2) phonological reconstruction

Sometimes you run into sound correspondences that you just can't explain.  So you may want to reconstruct a phoneme for a certain sound of unexplained origin, this phoneme would be illustrate k

Be economical, don't posit more sound changes than necessary to explain the data.

phoneme inventory - voiceless stops before voiced stops.  That you have maximally distinctive segments that tend toward symmetry.  Thus if you have a t, you will probably have a d.  If you have only two vowels and have an "a", the other vowel will probably be an "i" or a "u" but not another low vowel.  if you're going to have a nasal, m is the most stable nasal there is among the world's languages.

Spring 2003

Grammar / General / Historical / Phonology

General Essays Part 1 and Part 2

General - Part I - Essay

Sociolinguistics - data collected includes sociological markers such as age, sex, language region, socio-economic class, educational background, the types of linguistic data they look at include phonological variation (Labov in his New York (The Social Stratification of English in New York City 1966) and Martha's Vineyard 1963 studies), vocabulary used, discourse strategies (women and men), etc.  They analyze it by running statistical analysis to see if the sociological markers coincide with variations in linguistic behavior.  This method is due to historical precedent of William Labov's studies which he probably based on precedents in the field of sociology.  It is different from other traditional fields in linguistics, such as syntax and phonology, in that it looks at a large quantity of speakers and factors.  Many studies need to use statistical analysis due to the nature of the data involved, a quantitative method.  However, there are studies on sociolinguistics which rely on historical fact digging (Cameron's Verbal Behavior) and observation (Tannen's classroom observations of the participation of women and men).

Phonology and Syntax - influenced more from traditional methods developed from Chomsky, they assume that since all languages have the same properties, one informant should be representative of everyone, a qualitative method.  In addition, the main goal of these two disciplines is to map out the universal grammar that is in the heads of all speakers.  Thus, emphasis is placed on creating formal models which can predict grammatical linguistic behavior.  The data looked at starts from the simplest parts (phonemes for phonology, words for syntax) and then builds up to more complex units.  Sub-branches within both fields do look at quantitative data, examples are phonetic studies applied to phonological theory in Evolutionary Phonology, and information on processing within our brains used to constrain syntactic rules, found in Emergentism.

Historical linguistics - confined by the history of the subject, mainly looks at phonological correspondences which have proved more useful for historical reconstruction, to reconstruct they need to find enough data (often written) on various different daughter languages, they look mostly at basic vocabulary, so you don't need sentence patterns, you just need words and the sounds that make up the words.  The subject matter also confines how they analyze, sometimes there are no more living speakers, sometimes they compare existing dialects where they use dictionaries and word lists compiled from interviews of hopefully more than one speaker.  Lexicostatistics is a quantitative method employed to find relations among language families.  The comparative method is also quantitative in that it looks at a large number of languages, comparing the shared sound changes and sound correspondences.

Child Language Acquisition - subject matter collected restricted to what you are studying, child language, so you collect lots of data, from a historical perspective

Phonetic Analysis - data collected according to biological factors, women vs. men vs. children, nationality, usually collect lots of data, run statistical analysis, quantitative in nature

General - Essay - Part II

What does the existence of Universal tendencies tell us about language?

Universal tendencies in language show that certain phenomenon occur at above chance in the world's languages.  The interesting question is why these universal tendencies exist in the first place.  I will look at research in phonology, syntax and cognitive linguistics on universal tendencies and the explanations that have evolved to explain them.

In phonology, there are several universal tendencies, one of which is the Obligatory Contour Principle.  This principle states that you cannot have identical segments adjacent to one another in the underlying form.  However, according to Juliette Blevins in her draft of Evolutionary Phonology (which I'm not supposed to quote without her permission), the phenomena attributed to the OCP can be explained through a careful study of the historical development of individual languages.  For example, the OCP is often used to explain tonal phonology in African languages.  However, one exception to the OCP is the word hokpo, which has a high tone on the first V and a HL contour tone on the second V.  According to the OCP and mapping conventions, this pattern should not be attested.

Finite phonemic inventory which includes vowels, consonants and syllables (prosody) - tells us language is a combinatorial system, where we start with smaller parts to construct larger structures, we have a finite phonemic inventory but create an infinite number of words (potentially).

Syntax - Most languages have nouns and verbs.  All languages have closed and open classes of lexical items.  Most languages have some kind of tense or aspect, or markers for time.  All languages have questions, negation, (as far as we know).  - tells us that one of the functions of language is communication, about events and entities, about when things occurred or to what extent, and whether it did happen or not.  Most languages have Subject before Verb.  Chomsky has make claims that recursion is unique and universal to humans.

Acquisition - Jakobson (1968) showed there are universal tendencies for "mama" and "papa", first bilabials occur for children because children are reproducing suckling motions which they need to drink their mothers milk, He actually hypothesized that "mama" first comes to mean "food", then "papa" comes to mean father, and "mama" then changes to mean "mama", sequential progression of sounds that children make when they are babbling: bilabials and velars are easy for them to make first, in addition, dentals are easier than alveolar dentals.  Children reduce consonantal clusters.  cooing, babbling, one-word stage, two-word stage, short sentences and by age 3 or 4 children speak more or less with the same patterns as adults. - tells us that humans share certain phonological patterns based on development and the biology of the vocal apparatus, also shows that children have a tendency to coo, babble and so on, and that they work toward learning language.

Iconicity - icons - stands for exactly what is physically in the world (visual, auditory, etc.), index - extended from the actual thing (flame for "cook" or seeing smoke means "fire", symbol - culturally determined such as a red light which in our culture means stop, no direct iconic connection. languages are iconic because all languages developed from a semiotic (signs) system which tries to represent the outside world, developing more or less abstact over time. -tells that people try to be iconic when making language iconic, though for economical reasons, they will be trimmed down to symbolic communication.

Cognitive Linguistics - universal tendency to categorize, categories have prototypical effects, there are basic level categories

Language is used for communication, based on our brains and biology which makes it similar and different from other animal species.  One similarity is the need to express present needs with other members of the species.  Another similarity is the use symbols to express these needs.  Differences include the ability to talk about the past, future and potential events.  Our ability to subcategorize into binary sequences, leading perhaps to why we have recursiveness and analogy, explaining noun verb oppositions, semantic extensions,  etc.  Humans a social species who transmit language as a cultural artifact, though, if not taught language, can create one among themselves.

Fall 2002

Spring 2002 

Historical / Phonology / Grammar / General

General - Part I

Is the separation between diachronic and synchronic linguistics useful?

EXTRA: It was Saussure who first introduced the term diachronic, to make a distinction between diachronic and synchronic, from papers and lectures taught in the late 1800's and early 1900's.  His book - Cours de linguistique generale - 1913 and 1915.

Pick A, B or C

A. historical and child language acquision - don't reconstruct from universal words such as "mama" and "papa" since they are universally the same worldwide due to baby acquisition of phonology.  Languages are transmitted vertically and horizontally, vertically meaning from older to younger and horizontally meaning among peers.  Thus, to understand language change, you need to take into consideration these two directions.  Children learn from adults, but also create their own forms of speech and accent with their peers.  In addition, culturally obsolete categories and distinctions (gender and other types of semantic classes) are lost upon the following generation.

Children learn language by first learning basic categories, which show prototypical effects.  Thus, they are more likely to retain the central members of semantic categories, but may or may not retain peripheral members, or may create their own class of peripheral members.  These peripheral members are the result of semantic extension, which arise from analogy.  An example of this that may be used in historical reconstruction would be considering a word meaning "month" and a word from another language meaning "moon" to be cognates from which we can reconstruct.

B. historical and psycholinguistics - The McGurk effect has been proven experimentally, it shows that visual cues influence what we hear.  Thus, if a subject hears a recording for a "g", but sees a video of someone saying "m", they will hear "d" and so on.  Thus, visual clues are important in language acquisition, and will effect how languages change.  One possible consequence of the McGurk effect would be the tendency for bilabials to be more stable than other segments with weaker visual clues (m is a more stable segment historically speaking).  Perhaps, segments with similar visual clues will be more likely to merge, or switch places.

Perception experiments mentioned in Blevins' Evolutionary Phonology show that segments often are pre-articulated in the previous segment.  This sometimes leads to metathesis, since the first segment of a pair has the phonetic qualities of the first and second segments, thus sometimes leading to a switching of segments.  This would explain why sometimes there is metathesis occurring.  There are also perception experiments which show that geminates are resistant to change because they are often longer and louder, making them less likely to weaken.  In addition, children in general must perceive the language of others in order to learn it.  Because transmission is less than perfect, we can get phenomenon like word final devoicing, where those perceiving the language don't hear the voicing distinctions at the end of the word.  "t" can change to "s" in front of "i" for perceptual reasons, listeners of the language hear an "s" and opposed to a "t". 

In production, though we may want to produce a "ti" sound we may produce an "si" sound instead.  Thus leading to transmission of "si" as opposed to "ti".  In production we may voice intervocalic consonants, leading them to be interpreted as underlyingly voiceless consonants by the younger generation.  Aspirated consonants are often longer than other types of consonants, thus leading the younger generation to decide whether the salience feature is the air burst, loudness, or length, or a combination.  One phenomenon common in production is lengthening at boundaries, thus, word boundaries may sound longer, leading to the development of geminates and long vowels at word or morpheme boundaries.

C. historical and sociolinguistics - Wang "Competing changes as a cause of residue" 1969 lexical diffusion- change doesn't have to be regular or incremental, it can occur in one field, which spreads to some extent, but can be stopped or changed by other changes which occur in the language.  Labov - '63, '66 sound changes co-exist in time, some become prevalent than others within specific social groups depending on societal factors.  The more isolated the group, the more different its development will diverge.  Means of separating groups from one another include prestige, educational background, socio-economic status, social markers of gender, race, age, etc.  This explains the diversity of languages in the present time in terms of phonology, syntax, etc.  Language contact will change language through borrowing (within specific semantic classes with phonology very similar to the language from which it was borrowed from, such as word for governmental agencies in borrowed from French into English during the Norman Invasion), developing areal features, which historical linguists must attend to when making their classifications.  Social change in general, changes in technology, changes social roles, will explain semantic changes in words like "chairman" becoming "chair".  New terms are coined for new concepts "that's so 80s", describing new salient events and so on.

Historical linguists should look at words less influenced by social change, such as kinship terms and basic terms for food, etc.  Taboo words, sometimes sound change can be explained to sound change, in cultures of Southern New Guinea and Northern Australia (lardil?) if you die, your name cannot be repeated and the sounds can no longer be used (for a certain period of time), thus new sounds must be introduced into usage after someone's death, explaining how the community would have a sudden shift in sounds used.

Fall 2001

Phonology / Grammar / General / Historical

General -Part I - Part II

General - Part I

Linguistic theories generally focus on how speakers assemble words and phrases and how speakers pronounce them by applying rules.  So how do people understand language?

When making a transcription, I have different techniques depending on how fast or slowly the speaker is talking.  For instance, if the speaker is speaking really slow, perhaps because they know I'm writing a transcription, then I write down each word as I hear it.  If the speaker is speaking quickly, then I listen to the first, second and possible third sentence as I begin typing the first sentence.  Then, after writing as much of what I heard as possible, I would either rewind the tape, or I would ask the speaker to repeat themselves.  This type of transcription, however, is not how I understand language, since it involves me having to spend time and energy on spelling and writing down all of the segments in a sentence.

When I understand language, I don't think I spend near as much equal effort per word.  For instance, idioms and sayings are all stored in my head as one meaning, so when heard, I already know the meaning, often times before it is even finished being said.  An example of this type of idiom would be, "I let the cat out of the bag."  Other words that I don't spend much cognitive effort on would be tense markers, especially since often times they are redundant once you get into a long conversation about the past.  In addition, determiners seem to signal the presence of nouns, but I often don't even require this signal since I pretty much know when the noun will appear, and so on.  In addition, the role of determiners in adding new topics into conversation is often times redundant once a new topic is talked about for an extent of time.

In addition to the kind of word, the part of the word heard is also important to understanding.  In psycholinguistic research they have found that the first segment of the word is the most important when trying to recognize the word.  Then, the second is second most important and so on.  If you don't say the first part of the word, you probably won't understand it at all.  Thus, when we understand words, we start from what we first hear and use this to narrow down the possible candidates in our heads.

From studies on slips of the tongue and slips of the ear, psycholinguists have also come to find that people make mistakes by substituting, deleting or adding segments, syllables, whole words, etc.  An example would be children who must say the pledge of allegiance, and instead of saying "indivisible" they say instead "invisible".  This type of error is deletion which occurs because the children find the closest word that they actually do have in their vocabulary.  From studies on speech errors, psycholinguists come to conclude that there are various stages which we must attend to when understanding or producing speech.  First, we access the meaning we wish to express.  Then we access it's phonological structure.  In any case, no matter what mistakes we make, we never make words or segments which violate the phonotactics of our language.  For example, in the slip "fraudian slip" we might say "shraudian flip", where the s becomes an sh in front of the r.

When parsing words into sentences, we tend to parse as quickly and as easily as possible (minimal attachment theory).  This can be shown in studies on sentences such as "The cotton clothing is made from grows in Mississippi."  What this sentence means is "The cotton, from which clothing is made from, grows in Mississippi."  This is because the simplest structure we know is SVO order, thus we expect cotton to modify clothing, etc.  This effect is often called the garden-path effect, since you are getting lost in the garden when you chose the wrong path to take.  Another common example of a garden path sentence is "The horse raced past the barn fell.", which is supposed to mean "The horse, (which was) raced past the barn, fell."

In terms of recognizing the semantics of words, psycholinguistic experiments have found what they call recency effects.  If a sentence includes a word that is related to a different word found later in the sentence, it will prime the second word, making it easier to understand.  An example would be comparing "His aorta is damaged." with "The surgeon cut into the right aorta."  Thus, semantic items which are related to one another (found in the same semantic network), will prime each other, and allow us to better understand other members of the category that may appear in discourse. 

In addition to word and phrase recognition, I find that often times I know what the other person is going to say.  For instance, if you say "Good Morning!" the other person will respond in the same fashion.  If  you say "How are you?" the response will be something such as "alright", "pretty good" or some other relevant response.  Thus, there are certain pragmatic considerations which must be taken into account in discourse that make conversation easy to understand.  We know people aren't going to say things that are out of the discourse frame, which may be anything from school, to home, to work, to our social lives.  Obviously, depending on the person, the discourse frames conversed in will change, but we typically know which frame we will be conversing in when we see the person we will be talking to.  An example of this would be if I said "So how is the prelim preparation coming?" and someone responded "What prelim exam?".  This response, if given to me by my classmate who has been preparing with me for weeks, would cause me to pause and process longer, unless interpreted to be a joke.  However, if I asked a little child the same question, then I would expect such an answer and would not think twice about the response.

EXTRA: Shadowing experiments - subjects must repeat what they just heard, they can do so with language but not with nonsense words or a foreign language they've never learned.  This shows that we have some way of parsing the speech stream, even though it doesn't have enough acoustic evidence in itself to ascertain where word boundaries are.  Some mistakes of perception in these experiments include hearing "I pledge allegiance" heard as "I play jelly dance" and "kill germs where they grow" is heard as "kill germs with eggroll".

Ambiguous sentences such as "The cop saw the spy with the binoculars." and "He took the right turn at the intersection." are different from garden-paths since both interpretations are possible, depending on context.  Another attribute of garden-path sentences is that we don't realize our mistake until the sentence doesn't make sense to us, and then it takes us a long time to even figure out what the original sentence meant.

General - Part II

1) animal communication follow the link to Fall 2003 General Set 2 Problem D

2) Why do we not always speak literally?

    - euphemism (I'm on my period.  It's that time of the month.  He passed away.) taboo - In Northern Australian and Papua New Guinea, they have a custom where if someone dies, you can't say their name, syllables which are in the name in .Kabana (Papua New Guinea)

    - abstract concepts - time, social status, importance, energy (I'm in love. I'm out of time. I jump to conclusions.  Next week. The heart of the problem. The head of the household)

    - feeling (I'm exhausted.  He's high.  She let me down.)

    - to be more expressive (otherwise language would be so boring!) / also artistically expressive (She's hot.  He's a babe/pig.  I'm running out of money.  The water is running.  Bored to tears.) / (Life evanescent as the dew.)

    - innovations - ship names were used for space shuttles, space ship

What is non-literal (idiomatic, metaphorical) language useful for?

This language is useful for the above reasons.  Some concepts cannot be expressed literally.  To create new language.

What do you have to assume about your audience when using such language?

You have to assume that your audience shares the same cultural assumptions (time is horizontal in English and it's vertical in Chinese).   It's good to be up, and bad to be down (perhaps a universal?). 

You can talk about L2 and L1 differences, how in your L2 you don't get humor, etc.

Fall 2000

General / Historical / Phonology / Grammar

Historical

Historical - Part I - Essay

Phonology

Phonology - Part I - Essays

1. and 4. Why is there a need for morphological alternations to be included in phonological descriptions?

impossible, indirect, incapable, irregular, illegal - can be explained be an underlying archiphoneme N, which assimilates to its surrounding, however, this assimilation is no longer productive in the language today.  An affix which was adopted later in the language (?) and which is more productive is -un, unlikely, unable, the uncola.  It does not undergo assimilation before liquids.  Thus, there are distinctions between the phonological behavior of different morphemes, dependent on what time the entered into the lexicon.  In order to explain the difference between -in and -un, an historical reason must be made apparent.

Natural Phonology - processes and rules

processes - in, applies from your L1 to your L2 during, occurs across word boundaries, English alveolar nasal -> ten boys -> temboys, (exception flapping), palatalization -> whatchu talkin' about, "white shoes" and "why chose" shows the componential nature affricates (NOTE)

rules - un, does not apply from L2 to L1, not necessarily because of articulatory difficulty, lack phonetic causality(?), the application depends on morpheme identities (the deletion of t in soften and hasten only occurs before the en chaotive suffix), they always create a phonemic distinction, they apply only in words, not across word boundaries

morphological rule exceptions cause they're lexically based - trisyllabic laxing -divine, divinity, serene, serenity BUT obese, obesity. ivory (underived form also doesn't undergo trisyllabic laxing)

soft, soften, haste, hasten but softer, and hasty

elastic, elasticity, electric, electricity, opaque, opacity BUT adhoc, adhocity? yuck, yuckity (if you think the word is of latinate origin, or realize that the word is similar to other k softening words, you will apply it to new words)

Japanese: oo+kaze ("strong/violent wind") -> * oogaze

oo+sora ->oozora ("the sky" - just a more emphatic, slightly poetic form emphasizing the enormity of the sky, sorta like we say Montana is "big sky country").

/oo-/ means "big, great, strong" (in Mandarin, the character is read "da" or "dai" (both falling tones)), /sora/ means "sky"(in Mandarin this character is read "kong" (high level or falling tone)), and /kaze/ means "wind" (in Mandarin
this character is read "feng" (high level tone))

 Note from John Kupchik: in regards to the Japanese data, I should note/clarify that it shows a morphophonological process called "sequencial voicing" ('rendaku' in japanese). Basically when you compound two words the second one's voiceless onset voices (lenition!!). There are restrictions: 1)it only happens with native Japanese words (no Chinese loans, or European loans), 2) if there is a voiced obstruent in the root of the second word the process is usually blocked (that's perhaps why "ookaze" doesn't turn into "oogaze", the "z" blocks the process), 3)it is also based somewhat on pitch accent (fairly complicated, I won't get into details on this). It's said to be non-productive, though native speakers tend to apply it to unknown nonsense words by analogy, as long as they follow the above conditions (I tested this in Japan), which is rather curious...

A better example of this in relation to lexical strata (and the General essays where this would be relevant!!) would be a form like [mi+tooshi="mitooshi" ('visibility')] compared to [mi+tokoro="midokoro" ('a good point';'highlight')]. The "mi" is the same morpheme in both (it has a meaning of "see") but sequencial voicing only occurs with the second compound, not the first, even though the first compound SHOULD in theory undergo this process, if
it really was a purely phonological process. And in fact "tooshi" DOES undergo sequencial voicing with some other morphemes, for example [yo+tooshi="yodooshi" ('all through the night')]. This is clearly due to the lexical/morphological nature of the morphemes in question, it cannot be explained any other way. KEY:
"tooshi" means 'letting pass; showing in; through'; "tokoro" means 'place'; "yo" means 'night'.

president, presidency, regent, regency, agent, agency (but, if you add a -ty, to mean "like a president", it works, which shows that the change is not based on difficulty of pronunciation.)

FLAPPING occurs to alveolar [-cont] consonants when it's the single consonant between two vowels, the second of which is unstressed.

2. Why are phonological processes described in terms of phonetic features?

Jackobson used to use acoustic features, Chomsky and Halle introduced more more articulatory features.  Simpler representations (English plural formation is almost impossible to articulate without features, z becomes s after t, k, etc), can show phonetic distance, can be used to form natural classes, they actually tell you something about the physiology of your vocal tract, whether it be perceptual or production

3. What kinds of phenomena have prompted linguists to propose that phonological rules apply in a language-particular order? Could there be other explanations for these phenomena?

Morphophonological rules, if looked at from a synchronic framework, then you have to apply them before processes, since they are the results of historical change, which occurs over different times, creating different orders.

Simultaneous - if all rules are applied simultaneously, then an original feeding order will become a counter-feeding order, a counter-feeding order will become a feeding order, only if all the rules can be applied again.  This leads us to think that counter-feeding would be more likely to become feeding order than vice versa.  This seems to be the case, that counter-feeding often becomes feeding.

Counter feeding -

coronal stops become flapped when they are syllable final and released

nasalization - sonorant becomes a nasal before a nasal (obligatory in a syllable, optional in a stress group)

ordinarily, nasalization doesn't apply to the output of flapping bottom, batting, it will though if

feeding 1) A -> B applies before 2) B -> C veke and vege (finnish)

V del        vek    veg

FinDev    vek    vek

counter-feeding

bleeding

counter-bleeding

Grammar

Translations

1. Vai zala zaale kas ir plavaa ir laba? Is green grass that's in the meadow good?

2. Kur ir veetule kas ir maajaa? Where is the letter that's in the house?

3. Zalaa maaja ir plavaa. The green house is in the meadow.

Translations

1. A green letter is in the house. Zala veetule ir maajaa.

2. Mother has a green meadow. Matei ir zala plava.

3. The good garden is green. Labsaa (or Labaa) ir zala.

Spring 2000

Historical - Part I - Essay

Short Essay

1) "Irregular" sound changes, that is those which cannot be explained by reference to the phonetic environment alone, have been explained in various ways.  Choose 3 different schools and describe the explanation that they typically give.

Neo-grammarians - unconditioned change doesn't exist (Hermann Paul 1891), every sound change is completely regular, sound change imperceptible and incremental, takes a very long time to be implemented.  Change is physiologically motivated.  It is constrained by the organization of our articulatory and perceptual apparatus in order to enhance ease of articulation or perception.

Lexical Diffusionists - Wang (1969) believes that change is regular, and occurs abruptly, then it spreads gradually throughout the lexicon.  Thus, if irregularity occurs it is because the change hasn't spread completely, either because it stopped or because it ran into another sound change which blocked its completion.

Dialectologists - dialect chains, due to lexical diffusion, change is spread slowly and might stop abruptly for the same reasons given in Wang (1969), or for different reasons such as the dialect becoming isolated from the source of the change (migration, etc.)

American Structrulists - Regularity Hypothesis (Hockett 1968), same as neogrammarians in that sound change is gradual and starts with the individual.  The individual speakers aims at making a "t" and is more or less accurate depending on his abilities to perceive and articulate.  Since there is a lot of redundancy in language, the phoneme he does produce is understandable.  they have a strong distrust of anything that is not observable (not generative),  

Fall 1998

Grammar Part I--Essay

1) I only know of one grammatical framework that made the assumption that grammatical structures are binary and that is Government and Binding Theory.  This is also the only theory that I've learned anyway, I don't know if the Minimalist Program, Construction Grammar, Role and Reference Grammar or Lexical Semantics assumes binarity of some type.  In any case, what other theory could they possible be referring to in 1998?  If anyone knows any other theories they might be referring to please let me know.

Anyway, using binary branching you could draw a tree something like this.

For this discussion we can ignore IP since it isn't relevant since a non-binary branching framework could incorporate it if need be.  The reason theorists like binary branching is because they think it's simpler, though I'm not sure in what way they think it's simpler.  As far as I know, simpler is referring to the fact that at every node you only have to worry about drawing two branches.  Maybe that means simpler for the theorist drawing trees.  For processing I don't know of any experiments proving that drawing binary trees somehow is cognitively real, I could just be ignorant though.  As for being aesthetically simpler I don't think so since you have to write all those V' and N', which makes the whole tree look more cluttered to me. 

The good point to this binary drawing is that it allows us to differentiate the argument of the verb from adjuncts.  The first NP to attach to V creates V', thus signifying that it is an argument.  The unfortunate unsolved problem however is that there is no way in the above tree to distinguish the indirect object from any adjuncts we may add to the sentence, and this doesn't seem to be very helpful in explicating grammatical relations.  It seems to me one solution to this problem would be to have adjuncts marked with V'' instead.  I don't know if that's been suggested already, I'm kind of rusty on all this grammar stuff.

Another good point is that the each XP specifier can also be shown, since it is the last element to combine with X', thus creating XP.  Thus, Det is the specifier for NP, rather than some adjunct.  In the sentence above, the first NP and VP have no specifier.

A tree with multiple branches is given below.

In this tree, the VP has two arguments, thus attaches to both NPs and the V at one level.  Thus, you don't have the differentiating objects vs. adjunct problem found in binary branching.  Both types of trees seem to differentiate direct object from indirect object by distance from the verb.  Since there are more options with multiple branching, Specifiers and objects can both be differentiated from adjuncts.  The drawback to multiple branching, as far as I can tell, is the simplicity argument.  This, of course, hasn't been proven yet so I'd say in the end whatever Chomsky thought was best (binary) was best.

¡@

2) Give examples of two different kinds of modifiers (adjuncts) and discuss the syntactic properties that distinguish them.

(a) I went to work in Chicago.

In the sentence above, the adjunct is "in Chicago".  It is a prepositional phrase (PP) giving location, beginning with the preposition "in".  It is modifying "work", which acts as a noun according distributional characteristics (compare to "I went to the park.") Thus, "in Chicago", is acting as a modifier to "work", within the PP.  Some may or may not consider the PP  "to work" to also be an adjunct.  If they feel that "went" has an argument of location then "to work" is an argument.  If they don't think that "went" requires this argument, or they don't think that "to work" indicates location, then "to work" is also an adjunct.  This sentence is interesting because it shows how "work" seems to be a verb, but when using distributional criteria, it seems instead to act as a noun. 

(b) I saw three blind mice.

In (b) above, "three" and "blind" are both adjuncts to the noun "mice".  The first is a quantifier indicating number, it modifies the noun phrase, "blind mice".  The second adjunct is "blind" which is an adjective indicating a property.  It modifies the N "mice".  These adjuncts differ from the adjunct in (a) in that they are located to the left of what they modify.

¡@

3) Different word categories are differentiated by their distributional properties in sentences.  For example, verbs in English can take aspect markers and tense markers.  Nouns take determiners and adjectives co-occur with degree words.  They are also distinguished sometimes by their semantics.  Verbs describe events, nouns describe entities and adjectives describe properties.

(a) We work more because we are highly motivated.

In the sentence above, "more" is an adverb.  This is a general term for all words that can modify verbs, adjectives or other adverbs.  In this case, the semantics of the adverb is describing the extent to which we work.  The class adverbs is rather diverse, it includes words with various semantic functions.  If we want to be specific, we might call more an extent adverb.

(b) The car is more expensive, but it might be worth it.

In sentence (b) above, "more" is an adverb showing extent.  This time it is modifying an adjective.

Selected Topics

Phonology

Natural Phonology