Term Paper, 2008
9 Pages, Grade: 2,4
Comprehending language - either spoken or written - seemingly belongs to the simplest tasks humans are faced with in day to day life. Intelligence and education hardly play a role when it comes to understanding speech. It is an action that can be performed more or less equally by the vast majority of humans, even those with mental disabilities. More than that humans do it effortlessly and unconsciously.
But how do we do it? What is it exactly that happens in our minds when we hear or read a sentence? Many psycholinguists claim that the task of understanding language involves building a grammatical structure in your mind. This process is referred to as syntactic analysis or sentence parsing. It includes assigning a word class to each word, combining them into word groups and then finding syntactic relationships between those word groups. These steps should correspond with rules of grammar and the message intended by the sender.
The procedure of understanding a sentence can be broken down into the following parts:
- auditory and visual word recognition ( phonological/orthographical level)
- lexical and morphological processes ( morphological level)
- parsing ( syntactic level)
- conceptual interpretation ( semantic level)
- referential process ( pragmatic level)
The tasks listed above are, as empirical data suggests, usually undergone more or less simultaneously and exchange information in both top-down and bottom-up directions. What does that mean? The top-down approach assumes that listeners impose their expectations on the message they receive and will get confused should those not be fulfilled. Bottom-up on the other hand means that the addressee will first assemble the information given to him and then put it in the right order. There is some evidence that the top-down approach plays a big role. It becomes quite clear when we take into account that messages can often be misunderstood. As Tom Bever from the Columbia University in New York pointed out, the addressee tends to jump to conclusions when receiving messages. He outlines four basic assumptions that English-speakers make when they try to understand a sentence:
“Every sentence consists of one or more sentoids or sentence-like chunks, and each sentoid normally includes a non-phrase followed by a verb, optionally followed by another nounphrase. ”
The principle underlying that assumption is that the receiver of a message will try to divide it into noun phrase - verb sequences. It is called the canonical sentoid strategy as NP-V-NP is the standard form of English sentences. That the above assumption seems to be true becomes obvious when people are presented with a sentence like “Lloyd kicked the ball kicked it.”. Usually they will try to force an NP-V structure on it interpreting it as “Lloyd kicked the ball (and then) kicked it (again).”. In the rarest cases the actual meaning of “Lloyd (who was) kicked the ball kicked it.” will be grasped.
“In a noun phrase-verb-noun phrase sequence, the first noun is usually the actor and the second the object.”
The corresponding strategy would be to interpret NP-V-NP sequences as actor-action-object unless there is a strong clue not to do so. There have been famous experiments by Dan Slobin showing that sentences not having the actor first take longer to comprehend.
“When a complex sentence is composed of a main clause and one or more subordinate clauses, the main clause usually comes first.”
So the strategy to this assumption would be interpreting the first clause as the main clause if there is no indication to do otherwise. A good example would be the sentence “The elephant squeezed into the telephone booth collapsed.”. Because of the above assumption a reader of that sentence would possibly assume that “The elephant squeezed...” was the beginning of a main clause at-least until coming to the word collapsed.
“Sentences usually make sense.”
The underlying strategy here, i.e. using one's knowledge of the world to pick the best interpretation, seems to be the strongest one. It is what allows us to even understand syntactically incomplete or false utterances. Experiments showed that subject remembered sentences that were ungrammatical but appeared to make sense better than sentences that were grammatically right but semantically dubious.
One of the most famous examples for causes of misinterpretation that disharmonies with multiple of Bever's assumptions is: “The horse raced past the barn fell.” A so-called garden-path sentence. The word raced is seldomly used as past participle which leads to the wrong interpretation of it being a past-tense main verb. Also it is a commonly known fact that horses make good actors of racing- events. The sentence would be easier to grasp if there were more than one horses, making the “raced past the barn” part a modifier to single out one of the horses.
Human languages have a vast number of vocabulary items that can be members of different word- classes. Often this allows a certain sequence of words to be put in different syntactic arrangements. An example that illustrates this is: “The man saw the woman with the binoculars. The prepositional phrase “with the binoculars” can be either attached to the word saw. I.e. “By using the binoculars the man saw the woman” or it can be attached to the word woman. I.e. “The man saw the woman that had the binoculars on her.”
Those ambiguities usually get resolved by the context and/or grammatical constraints. However this is not always the case and the question that therefor arises is, how the human parses deals with such cases. There are three theoretical approaches to that matter: serial parsing, parallel parsing and minimal-commitment parsing.
A parser that is serial will chose one of multiple available options first and, should it encounter a problem arising from that backtrack to an earlier part of the sentence and start over choosing another option this time. As only one syntactic structure is being stored at a time the working memory needed for this method is quite low. However it is quite open to errors and delays caused by frequent backtracking.
A parallel parser would not limit itself to one syntactic structure among different possibilities but rather operate on them all simultaneously. As there is no need to revise former decisions such a parser would be faster than a serial one but undoubtedly need more working memory.
 Friederici, Angela D. 1999. Language Comprehension: A Biological Perspective, 2 nd edition. Berlin, Heidelberg: Springer Verlag: P. 212
 Aitchison, Jean. 2008. The Articulate Mammal: An introduction to psycholinguistics, 5 th edition. London, New York: Routledge: P. 214
 Friederici, Angela D.: P. 212
Term Paper, 15 Pages
Term Paper (Advanced seminar), 26 Pages
Term Paper (Advanced seminar), 15 Pages
Seminar Paper, 31 Pages
Term Paper (Advanced seminar), 27 Pages
Research Paper (postgraduate), 18 Pages
Term Paper (Advanced seminar), 32 Pages
Term Paper, 32 Pages
Term Paper (Advanced seminar), 19 Pages
Term Paper (Advanced seminar), 21 Pages
Term Paper, 13 Pages
Pre-University Paper, 22 Pages
GRIN Publishing, located in Munich, Germany, has specialized since its foundation in 1998 in the publication of academic ebooks and books. The publishing website GRIN.com offer students, graduates and university professors the ideal platform for the presentation of scientific papers, such as research projects, theses, dissertations, and academic essays to a wide audience.
Free Publication of your term paper, essay, interpretation, bachelor's thesis, master's thesis, dissertation or textbook - upload now!