Stephen van Vlack
Sookmyung Women`s University
Graduate School of TESOL
Introduction to Linguistics
Fall 2006
Week 6 - Questions for discussion
Intro to Ling, pp. 143-179.
1. How do we decide what class a word is?
There are several ways of determining which class a word is. All members of a single class behave in the same or similar ways. The simplest way of determining the category of a word is by looking at the specific behavior of the word. Certain word categories (in English anyway) often take similar endings. That is, word classes differ on how other units (morphemes) can be attached to them, thus only verbs are marked for tense via a past tense morpheme /d/. Members of the same category also behave the same in the way they relate to the words around them. The class of words called nouns, for example, are the only word class to be modified by adjectives in English ([theDP [stupidAP [manNP]]] . Likewise, nouns are often preceded by a noun marker, like the. Taking this idea and generalizing it a bit, we can say that certain categories tend to occupy predictable positions within an utterance. Verbs, for example, tend to occupy second position in English, and the utterance final position in Korean. The last way of determining word class is by looking at meaning. Meaning here is not used in the conventional semantic sense, but a more syntactico-structural one. All members of a particular class serve the same function in an utterance. By looking at the other classes that we know are there we can determine which function is missing and then pin a specific class to that.
Remember as well that class is not something that can be marked on a word in a strict sense. All words with some very interesting exceptions seem to be able to change their class rather freely. This is one way that language can increase its degree of creativity and flexibility while retaining comprehensibility. Thus forget about the idea that certain words belong exclusively to a certain class. For linguists the class of a particular word is indicated by its behavior at the moment not by some rule which stipulates what class it has to be.
2. What is the difference between function (functional categories) and content words (Lexical categories)?
The most often cited difference between lexical and function words are that the former have links to meanings in a real world context while the latter only have meaning in a linguistic/structural context. Lexical words serve a semantic function while function words serve only a grammatical function in the sentence. This is how their difference is perceived on the surface. As we get deeper into the structural elements of language, we will see, however, that the two behave quite differently at a deeper, more basic level.
In fact, some linguists claim that the differences between human languages result from differences in the way function words are allowed to be used in languages. They argue that content words are the exact same in all of the world`s languages. They have the same semantic structure and as a result call for the same syntactic environments and partners. This is a very large claim to make and is one you should think about because it has tremendous implication on how we should teach our students.
3. What is the difference inflection and derivation?
Inflection includes morphological change that does not change the fundamental meaning of the base. It is essentially syntactic relating the certain syntactic requirements. For example, inflection works to mark number on English nouns and TENSE and AGREEMENT on English verbs. Derivation, on the other hand, is a morphological process which results in a meaning change of the base. It is essentially a semantic process which may or may not result in a category change. This might be a meaning change in the general sense, as in the example COVER ---> [UNv[COVERv]. Here we see the category has not changed, but only the meaning. Much of derivational operations in English involve the changing of the categorical status of the lexeme in question, as in the example STRUCTURE --> [STRUCTUREn]ALadj] ---> [STRUCTUREn]ALadj]LYadv]
Inflection is marked on lexical categories (with the possible exception of prepositions?) although it is generally associated with nouns and verbs in Modern Standard English. As a relic of the Old English system, it is also part of the forms of the personal pronouns and here marks case.
4. What are the restrictions on how affixation works?
How affixation works is quite interesting. There seem to a wide range of restrictions on the number and categorical type of affixes that can be used as well as on the order that they can be affixed. The lexeme INSUFFERABLE, for example, is made up if several morphemes [in], [suffer] [able]. We know that SUFFER is the root because it is a free morpheme as opposed to the bound morpheme status of the morpheme [in] and it occurs in the central position. Don`t get confused between the preposition IN and the bound morpheme [in]. They are two totally different entities which just happen to have the same form. We know this because there is no semantic and certainly no historical connection between them. ABLE, on the other hand, is both a morpheme and a lexeme. We know that is can exist on it`s own (The bear was not able to eat the porcupine.) and even function as a base, as in the word ability. Once more, since its meaning as a free morpheme is essentially the same as what we see it doing as an affix, we need to posit that it is the same form used in different ways. Linguists like Selkirk (1984) and Di Sciullo and Williams (1987) have sought to try to come up with a rule-based system to describe derivational operations with limited success pointing to the idiosyncratic nature of at least some derivational operations.
5. What are free and bound morphemes?
The idea of free and bound morphemes is really quite simple. Free morphemes are able to stand on their own. They can function and are stored as complete words (and, of course, we would assume that the mental lexicon would be able to encode this property somehow). Bound morphemes on the other hand, are not able to stand on their own. They cannot function as words proper. They never occur alone. Now, in English, as we shall see next week, we can generally make an association between free morphemes and stems and bound morphemes and affixes. It is, however, not possible to do this in all languages, Spanish, for example.
6. What are heads? Give some examples?
The head is the element that dominates and controls, in this case, a word. (This is a term that was taken from syntax and extended in use to morphology, so we will reencounter this term again in syntax and you will readily be able to understand it.) The head determines the class of the word and the basic core sense of the meaning because it is the controller. In the compound word floppy disk, for example, which is made up of two parts floppy (adj) and disk (n.), disk is the head and floppy somehow modifies this head. We know this because the word is a noun and the floppiness of the disk is a secondary feature to the fact that it is a disk of some sort.
7. What is suppletion and how does it work?
Suppletion is a specific type of lexically triggered allomorphy in which the resulting forms are not necessarily based on the phonological structure of the base form. They are the forms we all love to hate, like went. Partial suppletion is when only one part of the lexeme is changed, as a result of a morphological operation, in an unproductive way. Ablaut nouns and verbs in English; foot-->feet, shall-->should are good examples of this. As the system that once made these forms possible has since passed from the language, they simply need to be memorized, so that both forms can be and need to be stored in the lexicon.
8. What is the Separation Hypothesis?
The Separation Hypothesis claims that affixes do not have a fixed meaning like lexical words. This seems to be true in that the exact meaning of the affix often seems to be dependant on the word that it is being affixed to. Based on this simple claim the separation hypothesis claims that morphological operations must be conceptually separated from the morphemes that perform the operations themselves. The linguist Robert Beard came up with this theory to explain certain irregular morphological operations like ablaut and umlaut in English in which one particular morphological operation like past tense formation requires more than one thing to change.
(stand ---> stood)
In such cases we can not simply say that any one element is the past-tense marker. Following this, the separation hypothesis says that is no one-to-one correspondence between morphemes and their intended meaning (function). This can be seen even more strikingly when we look at the highly complex inflectional systems of Old English.
Looking deeper, we see, however, that in general lexemes seem to have the same problem in relation to meaning and an overall situational context and not just the strict linguistic context. Look at the examples below and see what you think about the meanings of the words in bold.
(1) Giant Jihosiphat was hiding in the tiny garbage can.
(2) Giant Jihosiphat was hiding in the crowd.
(3) The ballerina walked around the fire.
(4) The cowgirl stomped around the fire.
The word hiding in sentences 1 obviously means something different. More clear is the example with around. In 3, around means `to make a full circle` while in 4, it means `in the general vicinity of`. Strange, huh? Well we will get to talk about all this next week.
More comments
Chapter 9 - Word classes
The most obvious thing that one has to do with words is to divide them into different classes. This seems to be one of the most basic things that children acquiring their first language have to do and therefore seems like a good place to start our discussion of words.
All words in a language are able to be broken down into different classes. There are three ways that this is done. First, in order to determine the class of a word we have to look at its meaning (semantics). Then we have to look at how the word is used in its grammatical context, i.e. what is the environment the word is used in (syntax). Lastly, we have to look at the possible forms the word can take (morphology). All native speakers do this constantly as they process language.
So, all languages have different classes of words, but the categories are not always the same across languages. The main distinction often made is between content and functional categories. The content category includes lexical categories such as noun, verb, adjective, adverb, and prepositions. Each of these categories differs in its meaning, syntax and morphology. Nouns, for example, often refer to concrete objects and abstract concepts. They occur as arguments (either in subject or complement position) of the verb. They are often preceded by a noun marker (a, the, some, etc.). Also they are generally able to take plural endings in -s or -es. This is how we distinguish nouns and all the other content words can be distinguished in the same way. Content words are the main parts of language and they are all similar in that they all have a clear an identifiable meaning. The different classes of content words are referred to as lexical categories.
Function words are different. Function words do not have any identifiable meaning, rather they serve to make the overall meaning of the utterance understandable. Function words are the glue that holds all the major chunks of the sentence together. The function category includes such word types as, determiners, demonstratives, auxiliary verbs, pronouns, and coordinating conjunctions. Function words are important in language because they play a role in helping to determine which lexical category a given content words belongs to. They are different from lexical categories in that they have no meaning and have a minimal number of different forms. That is, they are a closed class.
The whole deal with content classes is different and more complex, especially when dealing with verbs and nouns. Words that appear in a different number of forms based on its grammatical role are said to inflect. Inflection is an import part of understanding the behavior of verbs and nouns in English. In older forms of English there was much more inflection than there is today (even adjectives inflected), but a certain amount of inflection is still there. English verbs inflect to show Agreement with the subject, but only in the third person singular. Tense is also an inflectional category for English verbs. English nouns today only inflect for the category of Number today. The important thing to remember is that inflection is a regular type of morphological operation. In standard generative morphology people do not have to encode all the inflection we find in nouns and verbs in the mental lexicon unless there are irregularities present, such as the verb go changing to went in the past tense. The mental lexicon only contains the idiosyncratic (or exceptional behavior) features of the lexemes in question in the generative view.
As mentioned above, Old English had a much more complex system of inflections than Modern English. Verbs were conjugated for categories like person and number and there were different conjugational classes, such as strong and weak. Noun inflection (declensions) were also much more complex in Old English, again with several different classes of nouns occurring based on their declension. Nouns were marked case and number for In Old English even adjectives were declined just like nouns.
Chapter 10: Building words
Morphology is the study of the building blocks of meaning in language (word formation). This definition necessarily differentiates itself from phonology. Phonemes lack meaning while morphemes are the smallest units of meaning in language. Each morpheme has a single meaning. There are two types of morphemes; free and bound. Free morphemes are morphemes that are able to stand alone. So, for example, in the word `snouts` snout is the free morpheme and -s is a bound morpheme. The word snout is able to exist by itself in an utterance while -s cannot. Bound morphemes cannot stand alone. They only survive attached to other morphemes.
Most words used in the English language (and Korean as well) are really complex collections of different morphemes fused together. Take for example the word porcupine (Latin: pig-spine), which is an animal that lives in the forests of the Americas and Africa. It is composed of two morphemes, `porcu` and `pine`. When we look at this word the first thing we have to figure out is, which part is the root. The root is the most basic morpheme in the word. We determine which is the root by looking at the meaning of the word. What is a porcupine? Is it a type of pig or a type of needle or spine? Certainly it is a type of pig. Porcu must, therefore, be the root.
One of the main ways in which words are created is by tacking morphemes onto the root. We can either prefix them onto the front (to the left) of the root or suffix them onto the end (to the right) of the root. This process is called affixation. Affixation is only one of many morphological processes. Some others include; compounding, conversion, morpheme-internal changes, and suppletion.
Let`s take a deeper look at affixation. In affixation, there are normal operations which must be employed. For example, the suffix -able is only able to be suffixed onto verbs. Other lexical categories cannot receive this suffix. So affixes have limited distribution. Also, a given affix will have the same effect on all the words it is attached to. So all the verbs which have -able suffixed to them will turn into adjectives. The use of these restrictions means that affixation must occur in steps. The word un-happi-ness was not formed with just one operation. The two morphemes were added to the root in succession. We know this because of the restrictions mentioned above. The prefix -un only attaches itself to adjectives (un+happy is OK. un+happiness is not OK). This means that the un- had to attach itself onto the root before the suffix -ness and we have the order. In this way, we say that complex words are arranged in hierarchical order.
We can use this hierarchical order to explain ambiguity in certain complex words. The word `unlockable` has two meanings; able to be unlocked and unable to be locked. The difference in the meaning can be explained by using tree diagrams which show the hierarchical order of operations.
[A [V un[V lock]]able] and [A un[A [V lock]able]]
At this point we have to introduce the term lexeme. A lexeme is an abstract notion used to describe a basic word. The word tables is composed of a lexeme TABLE and an inflection -s. The addition of the infection -s creates another word form; tables. So the two words table and tables are two word forms of the same lexeme TABLE. This once again revolves around meaning. Different lexemes must have different meanings (they must refer to different concepts), such as TABLE and CHAIR, but word forms while they have different forms, retain the basic meaning (concept) of the lexeme.
Once we have this idea clear we can now form useful definitions for inflectional and derivational morphology. Inflectional morphology is made up of morphological operations which involve changes in word forms but not in lexemes. Derivational morphology, on the other hand, is made up of morphological operations which create new or different lexemes from the old ones. There are, and must be, meaning changes in derivational morphology. But the difference do not end here. In inflectional morphology the bare form of the noun (the part to which the inflections are added) is called the stem. The stem is different from the root in that the root is the most basic form of a lexeme. Take, for example, the word `killer`. KILL is the root but killer- is the stem. This has implications in ordering. It is not possible to create the word killser. This is because there seems to be a rule in language (possibly a universal rule) that inflection occurs further from the root than derivation. Inflection occurs on the outside.
In addition to the idea of lexeme and word form, we also need to add one more distinction. To avoid confusion among surface level similarities in forms, especially verbs, the authors say we need to have a category called morphosyntactic word. Single word forms and single lexemes can have two morphosyntactic forms as in the example, walking (progressive) and walking (verbal-noun). This merely allows us to process the meaning difference in two forms which obviously contain the same lexeme and word form. It`s a nice linguistic convention.
All this stuff is somehow encoded in our brain in a place we call the mental lexicon. I think the authors do a good job of describing this on page170 of the text (Radford et al. 1999).
Allomorphy in morphology works much the same way as allophony in phonology. Morphs are the surface forms of the morphemes; the way that they are actually produced in speech. The morpheme, on the other hand, is the kind of linguistic underlying representation of the concept. Allomorphs are variants of one morpheme. Take for example plural endings in English. The morpheme might be -s but this morpheme comes in different forms -s, -z, and -Iz as in the following words; cats, dogs, horses. These forms are allomorphs of the plural morpheme. Once more, they are examples of phonologically conditioned alteration. Looking at some other plural forms like fish, cacti, and geese, we see that these forms are also examples of allomorphy, but this is lexically conditioned through such processes as partial suppletion (goose-geese)and nullification (fish-fish).
This seems nice but it is a bit messy. Just look at English plural creation. Yes, we can phonologically derive some of the plural forms as phonologically derived allomorphs, but there seem to be some many exceptions which are idiosyncratic that something must be done to straighten this out. The proposal is the Separation Hypothesis, which is quite simple. All it says is that form must be separated from meaning. We create morphological categories like PLURAL and PAST TENSE and we add the morphological category to the lexeme to create new forms. That`s it.
CogLing, Unit 5: Constructions, pp. 70-96
9. How are frames combined?
The process of how frames are combined relates to the underlying meaning composition (not strictly syntactic composition). These meaning constructions often call for a certain number and type of elements to be involved (an in a specific order) in a particular structure. These elements being necessary as demanded by the meaning frame are mapped onto the original meaning structure, but they themselves open their own meaning frames which have their own requirements. In this way mapping continues to occur until all the requirements of each meaning frame are met.
This seems simple enough and all models of language make similar claims with one major distinction. In the cognitive linguistics model it is meaning and meaning constructions which drive the system and cause constructions to be created. In other models, most notably the Chomskyan model, but also computation models like HPSG and even dependency grammar, there is a strong distinction made between syntax and semantics resulting in the belief that constructions are generated through a strictly syntactic process based on categorical information, subcategorization frames and theta-roles. Lexicalists like David Singleton have pointed out that lexical items, not syntactic/computational systems as is prevalent in the Chomskyan model, choose their partners in a phenomenon called colligation. Singleton (2000), however, still falls short of claiming meaning as the driving force in such operations as we might expect from one working on lexical aspects of language. This shows the ingrained thought in linguistics that there is a barrier between syntactic and semantic operations. It is important to remember that such a distinction not only does not exist in the cognitive linguistic model but it is the integration of different elements which drives the entire system.
10. How is meaning constructed in the cognitive view?
We already discussed last week how there does not seems to be the same strict division in cognitive linguistics between what are termed semantics and pragmatics as there is in the Chomskyan model. It is one of the cornerstones of cognitive linguistics that meaning, including semantic meaning, comes from the world indirectly. General cognition, which necessarily includes cultural knowledge and norms, acts as a buffer between the real world and the world we create with language. As we have gone over all these points now quite a bit all this should be quite clear. In this chapter, Lee (2001) introduces a couple of new ideas which relate to meaning (in fact it should be clear that in the cognitive linguistic model all things eventually boil down to meaning as meaning is what drives the entire linguistic system).
Meaning comes from units of different sizes (constructions/phrases/words, frames, and idealized cultural models (ICMs)) which are joined and effect each other through a process called mapping. It should be noted the extreme importance that the construction claim brings to the idea of the mental lexicon. Since the 1980s lexicalists/lexicologists have been claiming that the lexicon consists of units of various sizes which behave in more of less the same way. This view conflicts with the view in generative linguistics that the lexicon is composed of words only: That is, lexemes are free morphemes, nothing more and nothing less. The construction view means that there is no theoretical distinction between constructions and words or even productive particles. From a lexical point of view this adds fuel to the fire of belief that the lexicon is the core of linguistic production and not just a tool used by a separate computational system. More about this later. In an idea developed by Fillmore et al. (1988) and later reworked and extended by Goldberg (1995) which has come to called construction grammar, structures themselves are stored in the lexicon and just like all other lexical entries contain meaning. As a result, constructions come with their own radial categories (frames) which enable and require them to chose the specific elements which will flesh them out in the linguistic world. The elements pulled into the construction have the their own sets of frames and which in turn will pull up certain meanings from the world depending on how they and where are mapped onto other elements including ICMs. It is through mapping and certain elements that meaning is achieved.
11. How does the evidence provided by the three different construction types discussed in the chapter support the claim that the construction of linguistic utterances is based on general properties of cognition rather than a specialized and separate grammatical competence?
The simple phrase `subject to more specific conditions` is the key to all this. While it is clear that there are semantic (in a generative sense) restrictions and effects on linguistic operations a careful observation of such operations always seems to come up with certain inexplicable phenomena. Let`s put it another way, if all rules have exceptions, as they appear to do, then maybe there is something operating above and beyond the rule itself which is running the system. These `more specific conditions` are shown to come from restrictions in our cognitive systems. Our cognitive system, in turn comes from the world itself and our continuous interactions with it. It seems then that much of the inexplicable phenomenon researchers are bound to encounter in their analysis of language can be explained if some higher mediating system is used. For Chomsky this system is the innate computational system of UG, but for cognitive linguists it is the general cognitive system, derived from real world knowledge. Think about these simple examples which cannot be explained effectively in the generative model.
(5) ?The elephant (a real pachyderm) danced in the corner. (Highly unlikely but probable)
(6) *The drunk elephant (a real pachyderm) sang and danced on the table. (Impossible)
In order to understand both of these utterances, but especially (6), we need to think about the world, the behavior and size of elephants, and rooms, like the fact that a corner is a pretty small place. This type of knowledge is used to create these utterances and the speaker is betting on the belief that the listener shares the same knowledge. If the speaker does not how big elephant are or that they cannot really dance well and certainly cant`t be drunk or sing, then they will misunderstand the speaker`s intention. Certainly the speaker must have meant what we find in (7) and (8).
(7) The elephant (a real pachyderm/a person in an elephant costume) danced (jumped up and down a bit) in the corner.
(8) The drunk elephant (a person in an elephant costume/a cartoon elephant) sang and danced on the table.
In order to come up with any meaningful interpretation of these utterances the listener has to think about the real world and even imaginary worlds. The job of listeners is, after all, to try to find the interpretation of an utterances which was intended by the speaker. This requires some sort of mediating knowledge above and beyond simple fixed world meaning. This process of integrating cognitive structures, in effect, has a profound effect on the creation of specific constructions and the elements used within them. It is what allows langa8ieg to be at the same time creative and comprehensible.