Stephen van Vlack
Sookmyung Women`s University
Graduate School of TESOL
Introduction to Linguistics
Fall 2006
Week 7 - Answers
Intro. to Ling. pp. 180-192
1. What are the three main morphological classifications of language? Give at least one example for each classification?
The three main types of languages in respect to morphology are:
-Isolating
-Agglutinating
-Inflectional
The basic idea behind these classifications is that languages seem to be able to be assigned to one of these three groups based not only on the types of morphemes that the language has and allows, but also the type of behavior that these morphemes typically engage in.
Isolating languages do not undergo derivational or inflectional morphological operations. The only type of morphological operation they undergo is compounding. There are only free morphemes in isolating languages. Chinese is typically given as a good example of an isolating language. Agglutinating languages operate derivationally but do not have any inflections. Fixed affix forms are stacked onto bases to make more complicated words. Turkish or Mongolian are typically listed as good examples of agglutinating languages. In both isolating and agglutinating languages the idea is that each morpheme has just one meaning or use and they differ in that agglutinative languages have both free and bound morphemes while isolating languages contain only the former (free). Inflectional languages are languages in which inflection is a very important part of the grammatical marking system. In such languages a wide variety of grammatical features is marked on words in paradigms or systems of grammatical marking. Important to note is that in such systems a single morpheme often has several meanings or functions. Russian is typically thought as a good example of an inflectional language.
This basic approach (the idea the languages can be so classified) used to be very popular in the field of linguistics and was termed typology by those people who practiced it. Typology was first set up and developed by people working in the historical/comparative tradition of linguistics and fell out of favor in mainstream linguistics for a period of time. Interestingly, it is beginning to make a bit of a comeback with new publications relating to typology beginning to emerge.
2. Are these classifications absolute: that is, do languages have to fit their classification perfectly or is there some degree of overlap?
No these classifications are not absolute. One of the reasons that typology did go out of favor was that it was shown to be inadequate in describing languages completely and this is an important issue for us to bear in mind. Languages, even the ones that are widespread and well known and therefore less idiosyncratic in their structures according to a recent theory proposed in McWhorter (2001), did not seem to be able to be fit into any kind of neat classification all the time. Thus, even though we may apply a particular classification to certain languages, such as the classification that English is an inflectional language while Korean is an agglutinating language, neither of these languages really fit exactly into their classifications. English, for example, has elements of all three classifications or types. Korean undoubtedly an agglutinating language in some respects also has aspects of isolating languages. The basic idea is that languages are not neat and clean and therefore cannot be described in such simple terms and manners. All languages are idiosyncratic or have idiosyncratic elements to an extent. This is because languages and here we do not mean just the vocabulary or pronunciation, but systems are constantly changing.
3. How do the concepts of culmination and syncretism work together to allow inflection to be a productive kind of morphological operation?
Culmination/cumulation and syncretism are very important concepts in describing how inflection is able to work as a system. Anybody who has studied a highly inflectional language or has seen the charts of noun, verb, and possibly even adjective declensions and inflections (`declensions` is a term that is used for nominal and adjectival paradigms while `inflection` is often used only for verb paradigms) must quickly realize that such systems are highly complex and also extremely redundant. If you take a look at the charts I gave you for Old English you'll notice that single morphemes (the morpheme -as, for example) is seen to repeat across both masculine and neuter paradigms in Old English nouns. In fact there seems to be an awful lot of repetition and redundancy in the system. The sheer amount of the redundancy leads someone who's learning the system for communicative purposes as well as a linguist who is investigating a highly inflected language to be able to identify inflections in larger patterns and not as individual units. The basic idea is that morphemes such as -as in Old English itself does not have any strong meaning. It's meaning only comes from its placement within the paradigm and its relation to the other morphemes in that same paradigm. The ideas of cumulation and syncretism then are there to help us understand how these paradigms are actually expected to work when it is obvious that there is definitely not any kind of one-to-one relationship between morphemes and their functions as well as morphemes to other morphemes.
Cumulation is a simple concept which states that a single inflectional ending can represent more than one bit of grammatical information. So, if we look at the Old English data, again using the simple example of -as, we see that -as is a marker of the both masculine and neuter declensions, for both the nominative and accusative cases, as well as being a marker for plurality. We therefore say that the morphine -as cumulates all that different information at the same time.
Syncretism on the other hand states that a single morpheme may fit into more than one inflectional situation. Looking again at the morpheme -as, we can see that it appears in two different places in the masculine declension as well as two different places in the neuter declension. The fact that -as appears in several places and with different meanings is explained by the idea of syncretism and syncretism can only be understood within the paradigm itself.
4. What is the difference between word-based and stem-based inflection and what kind of repercussions does this difference have on lexical theory?
The difference is simple. In word-based inflection the element upon which the inflections are affixed is a recognizable word which is able to function on its own. In stem-based inflection, on the other hand, the element upon which the inflections are affixed is not a word that can stand on its own.
(1) Word-based = dog-gy, light-house
(2) Stem-based = boysen-berry, forti-tude
chaek-sang, mori-bit
hak-saeng, nam-ja
This difference has had a profound affect on how linguists have come to view lexical entries. Traditionally the mental lexicon was viewed as a simple list of the lexemes that a particular language has/uses in addition to containing some rudimentary information about meaning and grammar in relation to those words. Here we run into a bit of a problem with this oversimplified model of the lexicon. Looking at some of these stem based languages, which linguists often call fusional languages, we realize that often verbs in these languages cannot be listed as lexemes in the lexicon because these stems are not identifiable words. They never occur as free morphemes.
Psychologically this has a lot of meaning. Certainly from a productive point of view we can imagine how a speaker of any one of these languages would be able to put together the right form. In such languages there would be a lexical entry, which is made up of this stem. According to standard theory, the basic rules of inflection would then allow a speaker to produce the right form. That's not an insurmountable problem if enough different phonological rules are proposed. The more damning problem is how would people actually recognize units that they hear if what is actually in the lexicon is a form which is never actually encountered. This is one of the main problems. Another interesting concept revolves around a simple the idea that if we allow stems into the mental lexicon then why don't we simply allow morphemes themselves into the mental lexicon. At this point once we allow this we have changed entirely our definition of what the lexicon contains. In the beginning we claimed that the lexicon is generally viewed as a word list and now we're saying that it might possibly be a list of stems/morphemes as opposed to words.
We don't have the time or the know-how to go into this in-depth right now has it would be a tremendous diversion from what we need to do but it is something that we will come back to in more depth in about two weeks so please keep this in your mind.
5. What are some of the main types of morphological operations?
The simplest way of understanding this is to simply arrange these different morphological operations in a very basic chart.
A. Concatenative
1.Affixation
a. Derivational
b. Inflectional
2. Circumfixation
3. Infixation
B. Non-Concatenative
1. Reduplication
2. Ablaut
a. Umlaut
3. Consonant Mutation
C. Compounding
6. How do proportional analogy and paradigm leveling work to change morphology over time?
Both of these types of change alter the accepted forms of words to make them morphologically more regular or simpler over time. That is, they both work to eliminate irregularities from the language. The former works through analogy to a related parallel system which is more regular and therefore more productive while the former works on elements within a single system. Simple!
The simplest example to illustrate the effects of proportional analogy is plural marking in English. Again, one look at the charts I gave you from Old English should tell you that languages change and sometimes a lot in a relatively short time. The -s plural marking was not always the standard marker for plural in English. It became the standard only over a long time and great struggle. Once, however, it did become the statistically most used form of marking the plural the forces of proportional analogy kicked in. Over time, the dominance of the -s plural marking has caused nouns which used other plural marking ro shift to the -s marking.
(3) Cactii -----> cactuses, feet -----> feets
In recent history we can still see the effects proportional analogy on forms like cactus and even feet is beginning to feel the irresistible pull of the -s plural. Here we can see how proportional analogy is a competition model showing how one more powerful system displaces a smaller, possibly older one. Interesting though is the question of how one system comes to be more powerful.
Paradigm leveling, instead of, focusing on two systems of the same thing, works to simplify existing complex systems like noun or adjective declensions. The basic observation is that declensions seem to become more similar over time. In the handout I gave you from Old English again we can see how not only are the masculine and neuter noun declensions slowly converging, but we can also see how the declensions themselves are simplifying. This is especially apparent in the feminine noun declension where many of the inflectional endings are becoming the same.
7. How do back formation and folk etymology work to affect morphology?
Back formation is an interesting phenomenon that applies to borrowed forms. In back formation a borrowed form which is morphologically complex will, through analogy to pre-existing morphological structures in the language, be broken down into morphemes and as a result a new root will be created which did not originally exist.
(4) work + er / work -and- burglar / burgle
(5) operat + ion / operate -and- orientation / orientate
The idea is that the morpheme -er in English is highly productive as an agent marker. It is freely added onto verb roots to create agent nouns. When the noun burglar was borrowed from French the coincidental correspondence in the shape between the English morpheme -er and the final syllable of the word burglar lead people to believe that they were the same thing. Based on this analogy, English created a verb form burgle from burglar. (French does not, in fact, have the morpheme -er. Thus the -ar ending on burglar has nothing to do with English -er.)
Folk etymology, on the other hand, basically arises because of a misanalysis of a word`s morphological structure. Basically people misunderstand a rare morpheme for a more common one and if this occurs enough the morpheme will actually change.
(6) sam+blind ----> sand blind
garden snake -----> garter snake
sickle-cell anemia -----> sick-as-hell anemia
Alzeimer`s disease -----> old timer`s disease
nepotism -----> nephew-tism
In the above examples we can see how folk etymology works. In Old English the word samblind was a morphologically complex word form consisting of the parts sam `half` and blind `blind`. Over time, through the process of folk etymology the form of this concept changed to sand blind based on the belief that it might have been sand in the eyes which caused the blindness. A garden snake changes over time to be called a garter snake presumably because it is not always found in gardens and the stripes on its sides resemble in some way the garters women use to hold up stockings or more likely men used to use to hold up socks. As we can see from those examples discussed above and the other examples supplied, folk etymology occurs as a result of phonological correspondences which happen to coincide with semantic correspondences. It just so happens that Alzeimer`s not only sounds like old but also afflicts old people. For folk etymology to occur both requirements must be in place.
Week 7 -Discussion of morphology
In looking at morphological operations in different languages we want to get some idea of how language, in a universal sense, actually works. The basic idea is that all languages function using the same basic system. By looking at some of these different languages we can make more enlightened postulations about the nature of that system and therefore teach our students better.
Central to the idea of language teaching are several linguistic concepts including word and sentence. In looking at several languages as we did in class, we find that the concept word is indeed very dodgy. There are basically two types of languages, according to the two books we read, in relation to morphology: analytic and synthetic. Chinese is usually cited as the prime example of an analytic language. Most languages in the world (including Korean and English) are synthetic, but to different degrees. The brief study that we did in class of several synthetic languages should have raised the questions: what is a word really and what might a sentence be. Indeed, we saw in a wide range of different languages that often one word can be an entire phrase (we saw this in Turkish) or one word can even be an entire sentence (we saw this in Swahili), or there can be a mix (we saw this in Quiche).
Based on this, the concept of word, as we generally use it and teach it, seems to be more complex than we originally thought. Many languages, including languages similar to English like Spanish and Italian (at least for verbs), do not have words in the conventional sense. Taking this into consideration, we can even assume that English, as well, does not necessarily store only word units either in the lexicon. No language does. We can assume this based on the fact that any linguistic system in the world must operate based on the same principles as all other linguistic systems. (This is the fundamental rallying call of UG.) If we see that there are differences between two languages, for example Korean and English, we have to create an underlying system that allows for both those surface-level systems to exist, the idea being that they both come from the same underlying system. If this were not so, then it would be impossible for English speakers to acquire Korean and visa-versa. Easy right?? OK, let`s get a little more concrete here.
In class we saw how many languages simply do not have words as we might normally define the (free morphemes). They do not have units that can function on their own. Rather, they create meaning by combining small chunks of meaning through morphological processes. OK, no words!! How can we equate this with English which seems to have units that can function on their own. Well, we need to leap into the heads of people.
The human brain is often seen as being strictly divided into several sections, each of which has its own particular job and function. This is called the Modularity Hypothesis and is an important factor in Chomsky`s view of UG. According to the Modularity Hypothesis, the language function of the brain is not only housed separately, but works separately and in a fundamentally different way than all other brain functions. The language component contains everything we know about our language; that is, the raw language data. It is broken up into two main components; UG (which includes (actually is) the LAD (ha)) and the mental lexicon.
The mental lexicon, is a collection of all language specific data; that is, the bits that are not part of or run contrary to UG (like irregular operations). The central core of the mental lexicon is comprised of lexical units called lexemes. Words are not part of the mental lexicon for reasons that should be obvious by now. There is no psychological reality to the notion of words, but only to the units that make up words. This means that words in English like infatuation and dogs are not stored in the mental lexicon in those forms. They would have to be stored according to their base form, what happens to be the lexemes; infatuate and dog. Now, in the genitive model only language-specific, idiosyncratic information is housed in the mental lexicon. This means that we have to come up with a way to derive infatuation and dogs from infatuate and dog. We can start with dog because it is easier. Because this lexeme is made plural through a regular operation, then it does not need to be encoded at al in the mental lexicon. Derivational morphology, as we should all know by now, is different. Much of it is idiosyncratic, at least the part about which words can receive which affixes, so this information must be encoded in the mental lexicon. As a result, it is easy to imagine that the entry for the lexeme infatuate necessarily includes the fact that infatuate is able to take the lexeme -tion.
(As you might have figured out by now, there is special notation for words and lexemes when we use them in a text such as this. When we use words as examples then they should be italicized. Lexemes should always be in capital letters (caps) and if possible small caps. This kind of notation simply makes it easier to show the distinction between words and lexemes when the need arises.)
Following this we can readily see that the mental lexicon is composed of morphemes (and only some in the standard theory ) and not words. Lexemes are essentially morphemes and idiosyncratic lexical units, like went, for example as the irregular past tense of go. OK! Big deal!! Well, this rather simple seeming difference gives us food for thought as to how we should teach vocabulary. Giving a lot or rules and/or meaningless word lists does not necessarily seem like the best approach in lieu of what we now know about the mental lexicon. The native speakers of a language do not have rules and such for the formation of words. (Think about articles and how they are used and perceived by native speakers. I think we can posit the radical theory that all nouns are stored with different articles already attached to them) All this is encoded in the mental lexicon via the entries for individual lexemes. The way to try to get our students to encode things in the same way is to provide them with meaningful, natural input. By exposing them to lexical items as they are really used is the only way to fire the right mechanisms in the brain which will activate the mental lexicon.
Generally, one difference between the systems of the L1 and the systems of the L2 in the brains of consecutive dominant bilinguals (and here I mean bilinguals whose control of the L2 is far below that of their L1) is that the L1 is reliant on the mechanisms of the mental lexicon and UG while the L2 is generally reliant on memorized material, like words and lists because they are coded and therefore housed separately . This memorized material is housed outside the regular language loop of the brain even though it relates to language. Therefore, in many cases the L2 is actually used with limited use or connection to the actual mechanisms in the brain designed for language. For that reason, the actual use of the L2 is strained, faulty and even impossible.
CogLing, Unit 6: Mental Spaces, pp. 97-115
8. What are mental space and what do they do?
Mental spaces are basically locations in the brain which are set up so that we can form utterances. This is an idea that cognitive linguists come up with which gives them space to be able to structure the thoughts and connections between the language and what is presumed to be in the brain itself. Maybe the simplest way of understanding this is through analogy. We can think of mental spaces as being kind of like a canvas that a painter starts out with. The canvases blankets and empty space which the painter is somehow going to fill with ideas. Mental spaces can also be likened to the tree structures that syntacticians draw much to the annoyance of their students. The easiest way of thinking of these is as being a kind of a template upon which language is built and which shows the connection between the language itself and the items in the utterances and the structure and contents of the brain.
In its most basic sense language is structured based on projections or mappings. The idea behind a mapping is that concepts in the brain are somehow mapped onto each other. Simply put, in mapping across spaces, an association is formed between two concepts which were for some reason not related previously. We can identify three basic types of mappings the first of these is protection mappings. Then we have pragmatic function mappings. And finally we have what are called schema mappings. The idea behind mappings is that there are obviously several different types. Each different type works slightly differently because mappings are usually not 100 percent. Certain elements of a host item will be mapped onto a new item. In protection mappings this is exactly what happens, only certain elements of the host item are mapped. In a pragmatic function mappings the role of a certain entity is exchange for the entity itself. We can never identify items by their functions worries some other salient action that they perform and which can be used to identify them. The schema mapping we take things which are related in the schema and mapped them together.
The structure of a mental space can be fairly simple and rather flat or might actually be quite complex. Here are a couple of examples.
(7) I feel tired today.
(8) I should beat you like a rug.
(7) above shows a fairly flat mapping area. The meaning structure is fairly straightforward. It Doesn't mean that it's entirely flat because the adjunct today is forming a certain type proposition which is limiting the construction, but there are no strong mappings here. Example (8) is much more complex. In this example we are likening a person to a rug, which is obviously not true. In some sense, however, the person is taking on some characteristics of a rug, albeit in a very hypothetical form. In this case we have some sort of mapping going on between 'you', 'rug' and information we have about rugs in the brain. We'd also argue that we have some sort of mapping involving the modal 'should'. Looking at this we can really see how mental spaces work. Mental spaces provided framework wherein we can insert the necessary links and connections which structure are utterances from a meaning point of view. At the same time the mappings that we put onto the mental space also serve to define the mental space. They extended it shape it.Mental spaces each reflect a different type of reality in some way, shape or form. They can relate to time, space, person, number and possibly other concepts. It is interesting to note that even in traditional grammar these are concepts which are seen as being part of grammatical systems, so already from the start we have some sort of justification for the idea of mental spaces. The basic idea is that as a speaker pulls word units out of the mental lexicon based on the conceptualizations of these words and their configurations it will open up mental spaces. All concepts that enter the speakers head are either put into a preexisting mental space or open up a new space. It is while they are laid out in the speaker`s head that certain parts of different spaces are mapped together.
Mental spaces are theoretical spaces in the mind that people create when they speak or think. As language or thought unfolds these spaces are built and expanded upon in idiosyncratic but highly constrained ways. In a real sense, mental spaces allow people to track the flow of their thought as well as keep track of ideas which were presented throughout the individual discourse session. In a real way every new idea which enters a thought our conversation were any additional aspect of the already existing idea will in fact open a new mental space. What is important is that all these mental spaces are connected in different ways: basically they are mapped onto each other. This is essentially how mental spaces work. By creating a mental space for every aspect of meaning in our thoughts and language and connecting them to form a chain of meaning which both the speaker and listener (at least in a discourse situation) have access to, we are able to fully understand what the other is saying and plan further discourse. Both interlocutors have access because, theoretically at least, both people are creating the same kind of mental spaces as they listen to each other and react to the ideas that the other is expanding and which are being created in their head. It is only the point of view which may differ.
According to the theory expounded by Fauconnier (1994), grammatical roles and construction elements play an important role in building and thereafter expanding mental spaces. In short, it is grammatical devices which play a pivotal role in building spaces. Some of the ones Fauconnier (1997) mentions include:
-Space builders (......to the market)
-Names and descriptions (Bob, that damn pig)
-Tense and moods (was, would have been)
-Presuppositional constructions (these are invisible in speech and come based on the semantic encoding of certain lexical items and their surrounding context: Bob has three siblings presupposes that Bob also has/had parents, etc.)
-Trans-spatial operators (copula be or stative verbs like become)
-Identification of elements (This deals with the properties of one lexical item being extended into a new mental space where that same lexical item is also found: If the lexical item John is in the BASE and the MEANING space then there has to be a link which carries the properties of John from one space to another.)
Looking at the above list in this important to mention that it is that the very difficult to make a distinction between what we normally think of as formal grammar and lexis or vocabulary. It would seem that a lot of grammar is actually handled by the vocabulary itself and not by certain hidden or universal grammatical principles. In this view of language grammatical relations are created and handled more or less through lexical items. This is a very large change from what we have encountered thus far in our study of standard linguistics. What is extremely interesting for us in this course is that it also fits in fairly neatly with ideas of the brain and particularly memory systems and learning. There is something very attractive about the idea of lexical items themselves forming or determining structure and not for structure sake but really with a meaning basis.
Another very simple way of looking at this is to think about word association tests. When people take word association tests what they are really doing is opening the mental spaces and forming a mapping chain from one mental space, which would be the basis to further mental spaces. So, if I give you the stimulus word `water` and you begin to form associations by saying things like `fish, drink, swim, cool`, etc. Depending on the nature of the relationship between the lexical items you may well be opening new mental spaces every time you mention a new word. This would of course depend on whether the relationship between the words is merely semantic or if it actually comes from memory. Looking at the above associations most of the connections are not really based on semantic principles therefore we would expect a word association mapping like this to really involve a lot of different mental spaces.
9. What are the advantages of using mental spaces theory in models of language?
The main advantage of a theory of mental spaces is that it ties grammar to both lexis and meaning. In doing so it is able to explain a wide variety of linguistic phenomenon with one relatively simple concept. Not only are elements of grammar like the use of specific tenses easily explained by mental spaces theory, but language change and creativity (and remember in the cognitive model of language these two are basically the same thing) are also elegantly accounted for. In addition, certain elements of semantics which have proven highly problematic or inexplicable in the generative model like counterfactuals like in (9) below.
(9) If wishes were money then I would be rich.
Since wishes clearly are not money and there is no reality where they are the truth conditional semantics of the generative school has trouble with dealing with such utterances. Here are a few more simpler ones.
(10) My father was dead yesterday, but he`s fine today.
(11) My brother is no longer my brother.
(12) Has God arrived at the party yet?
Such examples are common and reflect the creative power of language in that people regularly utter sentences which are, strictly interpreted, impossible. In a mental spaces model, however, such problematic examples are easily explained.
10. How can mental spaces be used to elegantly explain certain grammatical phenomena? Give an example.
Fauconnier (1997) postulates, ̋Mental spaces operate to build and link mental spaces.̋ And he defined them as, ̋Mental spaces (Fauconnier, 1994) are partial structures that proliferate when we think and talk, allowing a fine-grained partitioning of our discourse and knowledge structures.̋ Fauconnier (1997) gives an example as follows:
In saying Liz thinks Richard is wonderful, we build a space for Liz` reported beliefs, with minimal explicit structure corresponding to Richard`s being wonderful. In saying Last year, Richard was wonderful, we build a space for `last year,` and in saying Liz thinks that last year Richard was wonderful, we build a space for last year embedded in a belief space, itself embedded in a base space.
(Fauconnier, 1997: 11)
Based on this and the discussion above it should not be too hard to come up with an example.
(13) This tree grows like a weed.
The structure `this tree` opens up its own mental space, especially as it contains the space-builder `this`. Likewise, `like a weed` opens up another mental space. Both these mental spaces are contained within the basic or foundation space (Fauconnier, 1997) which also contains the verb `grow`. Within this main space we map elements from the `weed` space to the `tree` space. It is from this mapping of elements from one space to another and through the main space that the meaning of the utterance is created.