Talk:Psycholinguistics/Parsing

From Wikiversity
Jump to: navigation, search

Discussion #1

What you have here is a good start, but introduces a lot of information that can perhaps overwhelm the reader. I found that my understanding of parsing was largely helped by examples showing different kinds of sentences, and not only how they should be parsed, but how a mistake could be made (either in a garden path sentence, or if the reader is less familiar with the language). Parse trees can help provide a visual example of an abstract concept. When you describe thematic roles, give a sentence that demonstrates the various roles that the parts of the sentence can take on; show how this helps with parsing: a verb is unlikely to be an agent, a preposition unlikely to be a goal, and so on. Themes can work together with traditional parsing to try and determine meaning in a sentence.


In the section on Thematic roles, you speak about the “cost” of certain roles. Elaborate upon this. What is the cost? How is cost determined? What sorts of penalties would there be for choosing higher cost meanings, and are there any circumstances where it is actually beneficial to do so? When discussing semantic features, again, give an example of a sentence where the parsing is aided by the semantic features of a particular word. For example in the sentence “The fly's wings were buzzing rapidly,” knowing that a fly (itself possibly ambiguous, although linked with a noun and not a verb because of the possessive 's) has wings makes us more likely to classify wings as a noun, although it too can be a verb (consider “The fielder wings the ball back to the infield”). Also, why would a solid knowledge of grammar be both “critical for parsing to occur”, and “cause errors in parsing”. This seems contradictory. If it is true, expand on why this contradiction is possible.


The section on Serial and Parallel processing needs to be expanded. You give their definitions, but fail to describe whether one or the other is more prevalent (or more backed by research). Again, once the section has been expanded, examples of one sentence being processed both serially and in parallel would help the reader visualise the process.


The Garden Path section does a good job of describing the Garden Path Theory, but could use a little bit of history: where did it come from? What idea, if any, was used before? Why is it being questioned? Again, the section would benefit from examples that show how a bad parse can come out of nowhere after being led in a particular direction. But then, how is this parse corrected? Does the reader create a new parse, or simply overwrite the section that opens the “rabbit hole?” Also, what are the main shortcomings and benefits of Garden Path parsing? You begin to describe constraint satisfaction and briefly mention cross-linguistic studies without elaborating. Do they support or hurt the findings of Garden Path research?


Another potential area to investigate is that of computer parsing. Computers are actually (surprisingly?) able to parse sentences fairly adequately, although they do run into problems with garden path sentences and intentionally ambiguous sentences, moreso than humans. Computer parsing is largely dependent on how we understand human parsing, and tends to use a similar “most likely” system to parse, parsing sentences in the most likely manner. You may not have time to include a section on computer parsing, but you might want to glance briefly at the research if you have the time.


Largely, your article seems to be going in the right direction, albeit it requires a bit of expansion and explanation. Remember that although this page is being designed for a Psych class, it may get viewers who have considerably less experience with the subject. Examples help clarify key points, and can be re-used in new sections to highlight key differences between methods. Definitions of key terms are also useful if the expected audience is less familiar with the subject matter.


One final note on the structure of the article. At certain points, the article tends to either split a section that might be better together (such as talking about other models before talking about the problems with the Garden Path model), or including many ideas in one section that might be better served having separate sections (such as separate sections for thematic roles and semantic features). If it is simply a matter of working around the constraints of wikiversity, I suggest going to wikipedia (which has largely the same format, but more articles), and looking at the formatting of one of the “featured articles.” These articles are the “best of the best” at Wikipedia, and by clicking “edit” on them, you can see how some really neat formatting things get done. Furthermore, if you have any questions or would like clarification, just post on my user page, and I'll try and get back to you as soon as I can. Aericanwizard 22:22, 27 February 2011 (UTC)


Discussion #2


Overall, the author seems to understand the basics in the subject of parsing. The introduction is a simple but clear explanation about what the rest of the article will discuss in depth. It offers a clear definition what parsing is and touches upon the relevance of the article to the readers – it’s something they are doing even as they are reading the article!


The section about “What Causes Parsing?” is a little confusing. Bullet points are a great way to get the main point across, especially when someone is just skimming through the article. However, the information in the bullet points might need to be expanded upon, and some of the ideas are unclear. For example, under “thematic roles” a bullet point says, “- Focus on lexical information, such as frame analysis. They rely on interpreting words semantically in a sentence.” The ideas in this sentence are a quite vague, and need to be more developed.


The comparison between serial and parallel processing is important, but should also be expanded upon. The author asks good questions at the end of this section, but needs to research answers to these questions.


The theories of parsing are very interesting. The author explains garden path sentences well, elaborating on the metaphor behind the name and using it to illustrate how the Garden Path Model works. The article also explains the assumptions of the Garden Path Model and the principles it incorporates, such as late closure and minimal attachment. The effects of the Garden Path Model, such as the misconceptions that are difficult to change after the wrong interpretation of a sentence has already been made, are also clearly explained. The author cites research done by Christianson et al (2001) and explains the implications of the research. It might be helpful to explain more about the research so readers can understand how experiments are performed in parsing research. The author also briefly mentions the “Good Enough” theory and the Constraint Satisfaction Model under different headings. The Dependency Locality Theory and the Competition Model are outlined briefly under “Challenges to the Garden Path Model.” All these theories, though perhaps not as popular as the Garden Path Model, should be discussed in more detail and explained through research related to each theory.


The organization of the article is a little confusing. Under the heading of “Theories of Parsing,” the article lists the Garden Path Model, the Constraint Satisfaction Model, and Cross Linguistic studies. However, cross linguistic studies are not really discussed in this section. In addition, the Competition Model, the Dependency Locality Theory, and the “Good Enough” theory are not listed under the main title, but they are briefly discussed. The section about challenges to the Garden Path Model should not include the explanation of the Competition Model or the Dependency Locality theory. They should only be included in this section to discuss the contradictions between the models. The section for “Challenges to the Garden Path Model,” should be moved up to the section with the rest of the discussion about that model.


Overall, more articles should be included to back up the claims made in this article. General ideas need to be expanded upon through the article. The Garden Path Model, while important, should not be the only theory that is discussed in such detail. The layout and organization of the structure of the article should be more polished to help guide the reader through the article. However, the author has a strong outline of the important ideas that need to be discussed. The points that were explained were clear and concise. The author used language that readers will be able to understand, without making it sound too simple. This will be a strong article when additional research and more complete explanations of some of the topics are added.