English Phonology and Constraint Based Phonology - Coursework Example

Comparing Theories and Application of Opa within Optimality Theory 2. Compare two theories which claim to deal with opa showing the advantages and disadvantages of each of them. Constraint based phonology is a descriptive grammar based upon how a group of speakers used the language. The rules are formed when all candidates with matching factors produce a grammatical structure. These rules, or constraints, are formed by identifying what “is” and not “what should be”. We have developed this way of modifying language in order to accommodate ways in which the brain can process data. The brain collects all instances of language in use, and categorizes them into groups that behave the same in use. When all instances of a pattern that has been categorized as a group with matching properties, then how these patterns interact with others is also categorized. When all instances of behaviour of any group are matched a constraint exists. Constraint based phonology can be compared to using a neural network on a computer array to compare massive input and then create rules by comparing the input. These rules would be constraints if they apply to all incidents of a particular pattern. That is, they must be applied. Problems can occur when two constraints conflict, and then the results of conflicting constraints are compares in different order until the input matches the output. The process is very much like what is described in Isaac Asimov’s Three Laws of Robotics. In any given case, the three laws are applied and the consequences are noted. The laws are listed in sequence of importance, but they are all considered to be absolute in application with exceptions governed by the other laws. Linguistic constraints are much the same and conflicting constraints are filtered using optimality theory. The basis for this theory is the importance and strength of the constraints of the particular language. (Killing a human is the first and strongest of Asimov’s Laws, and the Second Law states that it applies except where the first law would be violated. The same occurs with the third law, but it is bound to both the first and the second, making it both weaker and less important.) As we learn any given language the brain figures out the constraints of that language and applies them, generally with considerable trial and error first. The more input given the faster the constraints can be identified. This is why there are so many problems as input of the language is acquired. The input must reach a certain mass before the constraints become reliably known. Using this method to learn and use language makes perfect sense since a long list of rules is not required. Categories and constraints are used instead. So the order in which constraints are applied can change according to the results of each. If applying one constraint does not result in a conflict with the other constraints, then it is applied first, usually resulting in a state where the second constraint does not or cannot apply, or when a state which creates an exception has been created. When using the past tense of want an “ed” is added according to the requirement for past tense, resulting in wanted. However, the “t” is dropped in conversation in English. Still, the “ed” is pronounced as separate syllable if the “t” is still there, because if it were, then the “ed” must be pronounced as a separate syllable. The “t” is removed after the constraint for pronouncing the “ed” as a syllable is applied. In a like word “waned” the “ed” is not pronounced as a syllable, because only the “d” was added. And the “a” is long because the “e” was there. The result is that the word remains one syllable. The collection of constraints are universal to all human languages, but they are not universally applied due to the particularity of ranking in individual languages. How constraints are ordered differs among languages and is particular to each, but some global generalizations can be made. The three parts of language use are Input, Process and Output, but these must be made in accordance with the applicable constraints. The Input is simply the data to be used. The Output is what results after the Input is subjected to the constraints and then processed by the operational component, or GEN, to reach the output stage. The application of constraint(s) using optimality is termed EVAL. So in order to process or transform any given input word so that it harmonizes with the other words to be used in context to form a grammatical structure, constraints are applied before GEN generates output. The processes that can be applied are numerous and only limited by constraints. These can be in reference to any factor of the language: grammar, spelling, pronunciation etc. There are two types: markedness and faithfulness. Markedness is related to the language itself and the requirements for grammaticality and pronunciations etc. while faithfulness requires that the output be similar enough to the input to be recognizable by the speakers. The aim is to make any given input work properly as a grammatical component with the other parts of the grammatical structure. So when a new word is added to the lexicon or learned as new vocabulary, it is subjected to the constraints that appear to fit the class to which it has been assigned due to the matching factors. Some good examples in English are some of the new words added for technology. The rolling hand rest that moves the cursor on the screen was called a mouse, because of its shape. Therefore when a verb was needed to express the action of rolling the mouse so that the cursor crosses over a digital object on the screen, it was called “mousing over” and the compound verb “to mouse over was created. The verb “to mouse” is a regular verb and follows the rules for most other verbs. The noun also is regular, and therefore, the plural of mouse is mouses. One of the key factors is applying optimality in the comparison of constraints. This is where the brain works like a neural network in computing. It sifts possibilities through sets of constraints in order of frequency of use in the language. If a constraint never happens, it cannot be applied and another is applied if present. If the second constraint eliminates the condition that prevented the application of the first without creating another that does prevent application of the first, then it will be applied in turn. This is the process of candidate chaining. The chaining can be done in order of perceived importance or in the reverse order. In each of these cases, the filtering is opposite. Candidate chaining is the basic process of applying constraints in order of perceived importance, or ranking, until an order of application is found that eliminates all important conflicts, that is conflicts in constraints that have no exceptions. The process of filtering using a candidate chain is applied in two ways. Candidates are ranked according to the constraints which they complete or violate. Constraint ranking is particular to each individual language, and the constraints are applied in order of their rank if possible without creating a conflict with a second ranking constraint. Ranking is done using three rules (McCarthy, 2004). First, the constraints to be ranked must conflict in assessment of possible output forms. Second, one of them must be the actual output form (winner), and it must dominate the loser. Last, there cannot be a third constrain applied that could be the reason that the winner works. So C1 is applied unless it cannot be. In such a case, if a C2 is present, it is applied to consider if C1 can then be applied. If C2 does not create a state where C1 can be applied then other strategies are used. For example the dropping of a final “e” of a noun in order to add “ing” to form a gerund, and the doubling of a consonant before adding “ing” to make a gerund in order to indicate the root word and govern pronunciation. A constraint here is that a noun with a final “w” has the “e” dropped in order to add “ing” without creating an unpronounceable result. However, if a third constraint is applied that causes C1 to become applicable then the ranking fails and must be reconstructed. Graphs can be used to illustrate either comparisons or violations. If one is not visually based or mathematically astute then word descriptions can be used to grant points and faults to each and mark their seriousness. This can again be compared using Asimov’s Laws of Robotics. In the movie Artificial Intelligence, made from Asimov’s book, the robot must kill his master, because if he does nothing then his master, much or all of humanity and himself will be destroyed. If he destroys himself the same thing ensues. The only way to save the rest of humanity is to kill his master and save himself. Some people process sound better than they do visuals, and for them this method works better than charts. McCarthy (2006) shows these charts very well, but he finishes with a caveat that there are times when one is superior to the other, such as when using the violations chart it does not show how it functions within the particular language, and so is not opaque after all. When using comparative charts, they do not determine which set of a given candidate set would win under differently ranked constraints, though they are better at showing the likely winners under different constraints. One further problem in ranking is the propensity for constraints simply stated to be assumed dominant over those with more complex explanations. As English is a language derived from many others, so it has many different sets and combinations of constraints for linguistic structures taken from other languages. This means that many times the more viable constraint is ignored for its complexity. In looking at the processing we encounter the idea of opacity, that is the how much can be figured out about the processes and the consequences of those processes by looking at the surface structure of that language, or how opaque the surface structure is and how much of the deep structure can be deduced through it. Certain processes in optimality ordering apply to measure the opacity of any constraint or set of constraints applied to input during GEN processing. Kiparsky identified and described opacity as follows: A phonological rule P, , is opaque if any of the following surface structures exists: instance of A in the environment; instance of B created by P in an environment other than ; instance of B not derived from P that occur in the context . Basically it means that a rule of constraint is opaque, that is it shows a fundamental deep structure of the language if it makes the structure obvious by eliminating any other possibility. The application of some processes tend to create opacity until an exception is added to input. Then opacity no longer applies. The major problem this can cause is that the lack of previously perceived opacity may not be noticed. There are different ways in which rules can be applied in order to make changes so that other rules previously not applicable can be applied. These are called Feeding and CounterFeeding. Bleeding and CounterBleeding do just the opposite: they make changes so that previously applicable rules can no longer be applied. If the rule order is reversed first, then it is CounterFeeding or CounterBleeding. Feeding and Bleeding and CounterFeeding and CounterBleeding are the processes which, when applied, create states where conflicting constraints can be harmonized. These are results of applying constraints that make changes that apply to other constraints. Candidate chains incorporate derivational features into an Optimality Theory analysis. This is seen as a way to show how opacity applies in phonological generalizations. Opaque relationships between rules in derivational theory are the previously discussed CounterFeeding and CounterBleeding. That is, they are examined in reverse order for what changes they can make to allow for application of other constraints present. However, the surface generalisation, or surface opacity, can only be understood if intermediate forms are available. Sometimes by applying constraints with no intermediate form the results are similar enough to suggest derivation, which obscures the deep structures completely. Another problematic area in optimality and the ranking of constraints is in pronunciation. There are so many contributory languages to English and not all imported words have been properly Anglicised. So the constraints of pronunciation just do not apply at all times. There are probably nearly as many exceptions to the constraints on pronunciation as there are constraints that apply. Hors oeuvres is a shining example of a French word imported centuries ago that does not comply with any rules of English pronunciation. In fact, it also does not follow the grammar rules of nouns either. This problem also occurs in Mandarin Chinese, due to the character based nature of the language. Imported words from English are often very far from their origins and they also have strange characters to represent them with no connection to the meaning. However WonTon was originally Guan Dong too. Of course that is another English word that breaks all rules. It is a countable and uncountable noun that is always used in it plural from but can be transformed at will from singular to plural. “I ordered won ton and the three lonely won ton on my plate were not tasty at all. Opacity is a difficult factor to properly gauge even given all the research. Derived languages and blended languages, such as English and Yiddish, are problematic for nailing down any true method for measuring opacity. The original rules may carry into the new language or be left behind and there is no certainty that modifications will be made to make the constraints harmonize. Divergent dialects represent another case where opacity may appear more than it is, since this language sound like the original but may have incorporated large portions of other rules, as did Mexican Spanish. The spelling of foreign words is a matter of accident ruled by how they entered the language. Rules may have changed, especially if there were intermediate languages, such as with Yiddish which incorporated some German and Russian into Hebrew. Idioms may become new words, creating even more confusion. Rules may or may not be harmonized before the word enters the public lexicon. Another reason opacity is problematic is that the deep structure of a language is not the same for every speaker, and each person using the language has the power to change it. Public figures, writers and journalists used to be the most powerful contributors to English, but with the advent of the World Wide Web, contributions to English, especially, have been influenced by millions. All living languages change daily and the changes are usually tracked on an annual basis as research and publishing companies try to stay with the current usage. However, it is not really possible to keep up. Therefore, the basic phonological structure of the English language is undergoing constant change. Of course, this makes optimality theory possible the best method for creating a descriptive grammar and analysing the language, because it also changes as fast as the input changes. The constraints may be universal, but English probably has more different combinations and possibilities of rankings and exceptions than any other language due to its complex history, and that is what fosters its enormous flexibility. In Chinese many words in English have to be translated into long phrases, because there are no words that can be used to translate the complex content of many English words. One example is the kill switch on movable mud pump platforms for oil rigs. That two word phrase requires a whole string of about 10 words to replace it. Mandarin is simply structured grammatically, and that very simplicity eliminates flexibility. So how does optimality apply in those cases? The language has thousands of characters, but many fewer words than English which as only 26 letters. One question did occur during this research, and it has not been answered: can OT be applied to the Chinese written language? References Kiparsky, Paul (1973). "Abstractness, opacity and global rules (Part 2 of Phonological representations)". In Fujimura, Osamu. Three Dimensions of Linguistic Theory. Tokyo Institute for Advanced Studies of Language. pp. 57–86. McCarthy, J., 2004. Optimality theory in phonology. 1st ed. Malden, MA, USA: Blackwell Pub. Read More