1

Introduction

The aim of this paper is to show the field of the comprehension of text and discourse. First of all, I will define some terminology, that is to say, text and discourse and the way in which they will be used. After that, in chapter 3, I will analyze how written and oral input is processed. This passage begins with an explanation that makes us understand why the terms comprehension and recall can be used interchangeably. In section 3.2, which is based on the work of McCabe and Scovel, we will discover what kind of information of written and oral input is stored: propositions or form. At the end of this passage some exceptions will be pointed out, too. In the final section of chapter 3, benefiting from the works of Field, I will show that language is usually processed from left to right, and we will become acquainted with a phenomenon called garden path sentences. The processing of discourse and text will be dealt with in chapter 4. However, first of all, according to Nattinger and DeCarrico, we will have a brief look at the two ways of comprehending oral and written input: bottom-up and topdown processing. Since my topic is The Comprehension of Text and Discourse, I will only concentrate on top-down processing, with chapter 4.1 focusing on discourse only and chapter 4.2 focusing on both, text and discourse. In the first paragraph which deals with discourse, I will examine the crucial role of prosody for understanding continuous speech and for being able to segment a consistent acoustic signal into words. To make this point tangible, I will demonstrate that English is a stress-timed language and why this is important for the segmentation strategy applied in this language. The observations about stress will be supported by the works of Clark and Clark. Moreover, to illustrate the segmentation strategy and the role of prosody, I will draw on experiments run by Cutler & Butterfiled and Shields, Mc Hugh & Martin. Yet, in section 4.1.2 we will see that there are also factors that impede the segmentation strategy which we usually apply. According to Duffield these factors consist of factors concerning the individual pronunciation, inconsistent pronunciation and most importantly coarticulation. Nevertheless, there are more impeding factors. These will be dealt with in section 4.1.3 by regarding a phenomenon called “slips of the ear”, the misplacement of word boundaries, which was proposed by Cutler and Butterflied. Although the segmentation strategy is crucial for comprehending discourse, we have learned in section 4.1.2 f that it does not provide a sufficient explanation for understanding 1

continuous speech correctly. As we will examine in section 4.1.4 there is more to the identification of spoken words – we will consider the importance of context. This paragraph profits from the works of Duffield and Clark & Clark. In the final section of chapter 4.1 the notion of context will be examined in further detail. I will focus on an experiment by Warren and Warren about the phonemic restoration effect which will help us understand the importance of semantic and syntactical context. Further information about context in this passage is supported by an experiment by Warren and Warren and explanations taken from Clark & Clark. After the chapter about discourse we will consider both, text and discourse in chapter 4.2. The first section will be about extralinguistic context: background knowledge and pragmatics. The importance of the former will be supported by an experiment run by Bransford and Johnson, and an analysis of this study by Harley. Thereafter, the passage about the role of pragmatics will be based on the works of Grice, Searle et al., McCabe, and Barsalou. In this section we have learned that comprehension also depends on situation dependent meaning and background knowledge. Thus, to deduce a situation-specific meaning we have to make inferences. In section 4.1.2 we will learn, in accordance with Harley, that there are three types of inferences. Furthermore, working with the results of an experiment by Bransford, Barclay and Franks and an explanation given by Harley, we will analyze if we are aware of the inferences we make. Hereafter, I will demonstrate that inferences are culture-dependent. According to McCabe we have culture-specific schemata which influence the way we experience, interpret, and recall a situation, a fact proven by an experiment run by Invernizzi and Abouzeid. Finally, the last section of this chapter will be about cohesion, a concept necessary to be followed to produce comprehensible text and discourse. The maintenance of cohesion is one of the basic rules of language. Therefore, in this section I will discuss the process of figuring out how a text relates together; a process called resolution proposed by Harley. After that, benefiting from the works of Harley and Fodor, I will define the terms anaphora and cataphora. Focusing on the concept of anaphor, we will learn by what a surface anaphor is characterized and which role the concept of Empty Categories plays for surface anaphors. Subsequently, I will point out the classification of cohesive devices proposed by McCabe, and finally, according to Harley, we will have a look at a resolution strategy called parallel function. I am then closing this paper with an evaluative 2

recalling an information is faster when the information has been understood. people remember the meaning of an information but not its exact form. for that story. 2 Definitions According to the Longman Dictionary of the English Language. discourse preferably refers to speech. Hence. The latter is stored verbatim only for a very short time. These definitions are quite similar. There are various experiments about the recall of text. (McCabe 1998. subjects are presented with a story and afterwards they have to report what they remember of it. the better we store it. when they did not have this title. 277) 3 3. I want to leave the role of context.2 Propositions and Form According to McCabe. out of account. Nevertheless. 1 This story is in the appendix 3 . when examining the way we comprehend language. In one of them. However. for now. Thereafter we paraphrase the form and reduce the content of the information to a gist. the test persons could not fulfil this task. 1 (Harley 1995. Yet. 1973. too. 212 ff) Nevertheless.1 How Do We Process Text and Discourse? Comprehension and Recall The comprehension of an utterance or a text and its recall are interrelated. The terms may be used interchangeably. and the American Heritage Dictionary defines text as “the actual wording of anything written or printed”. the better we comprehend something. and for the purpose of this paper both terms can be used interchangeably. we will look at what is stored and recalled. 450). run by Bransford & Johnson. First of all. which facilitates comprehension. whereas text rather refers to a written passage. a context. discourse is the “formal and orderly expression of ideas in speech or writing” (Longman 1995. When in possession of the title. However.summary and a discussion about the advantages of text and discourse in the comprehension process. Therefore. the subjects had no difficulty in comprehending the plot. the point I want to make is that arguments which are valid for recall apply for comprehension. 3.

the meaning of an information is stored in a network of propositions. (McCabe 1998. we do not remember details. Although the number of propositions and the choice of propositions is individual. but what about the form? Scovel explains that “we tend to remember sentences in a form that is actually simpler than the structure which we originally read or heard. 279) Consequently. but the gist of a story. One occasion where this could be the case. For instance. Thus. There are mnemonists who have a special ability to recall text verbatim. The most significant ones are called macropropositions. Propositions consist of semantic units and their syntactic status in the sentence rather than of grammatical words.this gist of the information does not contain random elements of the original information but consists of propositions. 279 f) Now we have seen what we remember semantically. (Scovel 1998. They consider approximately 80% of the recall to be macropropositions. A proposition is a basic “minimal unit of information. Normally.” (Scovel 1998. 279) 4 . sentences which were perceived in the passive voice are often thought to have been active ones. here the form is crucial for the meaning of the joke. Moreover. Hence. Although the choice of propositions varies individually. the syntactic form is stored if it is important for meaning. Nevertheless. its “visual appearance” is retained.”. in the form of relationships among concepts”. test persons mainly agree on what is to be considered an important or a less important proposition. apart from this. and the rest of the recall to consist of micropropositions. 279) If we wanted to retell this joke after a while. (McCabe 1998. our memory is poor for structure and tends to paraphrase passive sentences into active ones. as we have seen above. there is a different degree of importance of propositions. whereas the less meaningful ones are called micropropositions. Scovel demonstrates this with an example where subjects presented with the source sentences “Isn’t the cat being chased by the dog?” rather recalled its active variant “Isn’t the dog chasing the cat?”. (Scovel 1998. “Why couldn’t the mummy answer the telephone? Because he was all tied up. a proposition comprises a content word and its situation-dependent specific syntactic characteristics. few exceptions where the exact form of a sentence is stored exist. we would certainly recall the answer to have been in the passive voice. (McCabe 1998. is a joke. 66) For instance. To put it simple. (McCabe 1998. 66) Yet. 60 ff) Apart from this simplification rule. the type is not.

In the teacher example. these propositions band together to a network which represents the content of a story. However. In fact.” But what makes us think that a sentence will continue in a certain way and not in another? First of all. in many instances the English language is ambiguous and “leads us up the garden path”. Garden path sentences have a structure which is likely to mislead their reader.3 Processing from Left to Right According to Field. with the teacher being the agent. And the basic unit content is stored in are propositions. Nevertheless. occurs as we listen or read. there are more and less important ones – macro. When a reader reaches a point in a sentence where he hesitates because the sentence does not make any sense the way he read it. which determines the meaning. he is probably facing a garden path sentence. As a matter of fact. it is possible that the teacher may be the recipient of the action as in “The teacher taught by modern methods had great success. This means that we process any spoken or written input at the very moment we perceive it. the most frequent sentence pattern in English is the subject. object pattern and if a sentence is revised it may be misleading. One possible way to dissolve the ambiguity is to read the sentence again and to look for an alternative interpretation of the sentence. our world knowledge also interacts with frequency. In other words. Furthermore. the phenomenon of garden path sentences occurs much more often in reading than in listening because in the latter the listener has the aid of prosody to his disposal. And finally gist (macropropositions) is easier recalled than detail (micropropositions). and do not wait until the end of a sentence or an utterance. from left to right. For instance. Moreover. 3. language is processed from left to right in a linear procedure (at least in Western cultures). verb. This is called syntactical parsing. we assign meaning to the input right away. I want to emphasize that we store content not form. However. when reading the first words of a sentence starting with “The teacher taught” we would probably assume the sentence to state something about an action that is carried out by the teacher. the interpretation of the sentence to be active is probably caused because we assume the sentence to consist of the SVO pattern but also because we simply expect a 5 . The assigning of a grammatical structure.To summarize and conclude the topic of what we recall. the notion of frequency has something to do with it.and micropropositions.

Possible meanings: the building collapsed. When encountering difficulties in comprehending a sentence we are usually facing a garden path sentence which comes into being due to syntactical peculiarities or a violation of our world knowledge. the enterprise went bankrupt (Field 2003. their knowledge 2 “The pencil bought…”. Yet. 4 Top-down Processing When we listen to a person we are conversing with or when we read a text. Hence. I will focus on top-down processing which helps the reader to identify and disambiguate words and sentences by making use of the perceiver’s world knowledge. and by enabling the perceiver to draw inferences. 103). 124126) In conclusion. for example. we process the information in two ways which complement each other: the bottom-up processing and the top-down processing. Consequently. since the topic of this paper is the comprehension of whole texts and discourse as opposed to single sounds. which is deduced from our world knowledge. Then this “incoming data […] is analyzed. Top-down processing makes it possible to understand fast.” (Nattinger and DeCarrico 2001. single sounds and letters are comprehended by bottom-up processing.teacher to teach and not to be taught. begins with the simulation of the sensory receptors. by generating expectations. 124-126) 3 E.: The bank collapsed. categorized. is. sentences are processed linearly from left to right. and interpreted on the basis of information in the data. One basic guideline. Listeners and Readers apply their world knowledge.g. And not all ambiguous sentences are necessarily garden path sentences. which is always the first step in the comprehension process. by making upcoming information predictable. and even to skip some of the steps proposed by the bottom-up processing. (Field 2003. to comprehend disrupted text and discourse. Pencil is an inanimate noun and therefore assumed to be the recipient of an action as in “The pencil bought by my mother is beautiful”. that animate nouns are rather agents and that inanimate nouns are rather recipients2. Bottom-up processing. 125) 6 .3 (Field 2003. To conclude the topic of garden path sentences it needs to be emphasised that the ambiguity of meaning results from the syntactical structure as opposed to other cases of ambiguity where a sentence has several meanings because of the presence of homonyms. the application of world knowledge encourages specific interpretations.

Yet. their understanding of the context. rhythm is the main factor to rely on. due to prosody. And depending on the clarity of the pronunciation. Secondly. (Nattinger and DeCarrico 2001. we cannot rely on context solely to identify the lexical entry referring to the phonetic segment. disambiguating in discourse is easier than doing the same in text. 7 . it may be difficult to differentiate between the two options. Because otherwise we would not be able to tell whether a stretch of speech like <thecasesarelight> is supposed to mean <the cases are light> (/ðɘkeɪsɪzɑ:lʌɪt/) or <the case is alight> (/ðɘkeɪsɪzɘ’lʌɪt/). On the one hand. After /keɪsɪz/ both interpretations are possible because they produce a valid context: /ɑ:lʌɪt/ represents the verb be. The pronunciation of either sentence only varies in one sound. matching phonetic segments to words and considering context only. and the adjective light and /ɘ’lʌɪt/ stands for the adjective alight. we do not match a perceived phonetic segment to a lexical entry in our memory. does not enable us to understand continuous speech. in this paper we will only consider a small part of the extralinguistic factors.2 we have acquired some knowledge about the dissolving of ambiguity. Yet. we first need to understand what rhythm is characterized by. and they also need the extralinguistic context to understand discourse and text correctly.1. Both options are possible: “The cases(N) are(V) light(Cs)” and “The case(S) is(V) alight(Cs)”. we have the help of prosody to our disposal to comprehend discourse. Yet.1 In Discourse 4. To understand this I first want to show how speech segmentation does not work. Hence. However. Like one might think. This factor obviously complicates the comprehension process of discourse.1 Prosody and Segmentation Strategy In the processing section 3. Now one might think that. 103) 4. Cutler and Butterfield explain that on the other hand there are no consistent pauses between spoken words in continuous speech like in written text where there are gaps between every single word. As we have seen.about certain rules of the language. 2nd person. To demonstrate this we may use the same stretch of speech as above. one cannot say that discourse is easier to comprehend than text. So how are we able to know when one phonetic segment ends and when the next one begins? The answer is that we automatically apply a segmentation strategy which is based on rhythm. the schwa sound /ɘ/ and the long vowel /ɑ:/.

listeners focus on them.2 Segmentation Strategy: Impeding Factors 8 . 215 f) This statement is based on an experiment run by Shields. we still know what to fill in the phonemic gap. The outcome was that test persons detected the [b] in the nonsense word faster when it occurred in the stressed syllable BENkik than when it was in the unstressed syllable benKIK. 215 f) As demonstrated in a study by Cutler and Butterfield. when pronouncing English words with his French accent. That is because stressed syllables carry the most phonetic information. A speaker of French. In the case of English this rhythm is a stress-timed one. consisting of one stressed syllable followed by two unstressed and a stressed one. 185) A stress-timed language. From this we can deduce that an initial stressed syllable most likely refers to a lexical word and since lexical words carry the meaning and not the grammatical ones. Furthermore. (Clark and Clark 1977. Mc Hugh and Martin (1974) where subjects had to listen for a [b] in this stretch of speech: “You will have to curtail any morning sightseeing plans. 215) In other words. (Cutler and Butterfield 2003. it takes an English speaker the same amount of time to articulate the word lickety-split. the rhythm enables us to anticipate what is coming up. This is done automatically to retain the distinctive rhythm of English by speeding up the unstressed syllables in lickety-split and by slowing down the stressed syllables in black horse.Every language has a distinctive rhythm and listeners are very sensitive to disruptions in it. as the plane to benkik leaves at noon”. (Clark and Clark 1977. a strong syllable initiates a lexical word with a chance of 75% and a weak syllable initiates a grammatical word with a chance of 2:3. therefore. Consequently. would take longer to pronounce lickety-split than an English speaker because French is a syllable-timed language where it takes the same amount of time to articulate any syllable no matter if it is stressed or unstressed. even if we do not understand all of what has been said. is one where “speakers […] speak with accented syllables at roughly equal temporal intervals. in continuous speech listeners listen for stress. the listening for stressed syllables seems to be a good segmentation strategy. as to say the words black horse which only consist of 2 stressed syllables. Therefore. in discourse. In English “listeners organize their perception around the stressed syllables”. as stated in Clark and Clark. (Cutler and Butterfield 2003.1. 185 ff) 4. due to noises or an unclear pronunciation.” (Clark and Clark 1977.

Hence. we are not only facing the difficulties proposed by coarticulation. Consequently.1.to sound like an /m/. The first fact that needs further consideration is that every individual pronounces words in a unique way. the pronunciation of sounds varies according to their momentary location. depending on the register and momentary masking conditions. (Duffield. The traditional ones assume that the lexical preprocessing also applies to the coarticulation problem. The bilabial place of articulation of the <p> in -put influences the <n> in in. no matter how they are pronounced and by whom. but also the problem of placing wrong word boundaries. in turn. This signifies that a sound is influenced by a following or a previous one. For instance. we are able to identify a word spoken by two different individuals as one and the same. The third factor which counts among the segmenting problem is the difficulty caused by coarticulation.of the word input is pronounced identically to the prefix im. To demonstrate this we will look at an example of Duffield: the prefix in. a person pronounces words differently when speaking to friends than when lecturing in university. coarticulation is the most troublesome factor of the segmenting problem. Nevertheless. This is done automatically to facilitate the pronunciation because it is easier to pronounce the sequence /mp/ than /np/ since the former consists of two bilabial sounds and the latter of sounds with a different place of articulation: an alveolar and a bilabial sound. psycholinguistic researches disagree on the way a correct identification of words is made possible. becomes more like the source sound. comprehension is achieved. and factors like suffering from a sore throat also affect the way one speaks.3 Misplacement of Word Boundaries Yet. 40 ff) 4. there must be a preprocessing system at work which enables us to separate the suprasegmental features of the phonetic segment from a word. Obviously. because of the fact that we are able to identify words. Nevertheless. whereas others believe that all of the phonetic variations are stored directly like allophones of single sounds. This means that sometimes is comes 9 . Secondly. there is an inconstancy in pronunciation even if the very same speaker says one word in different situations. however.The process of segmenting a stretch of speech is complicated by three factors which influence the acoustic signal we are to comprehend. and this target sound.of the word impact.

prosody and pragmatics. After the pre-processing of the word. to infer the communicative intention of the speaker correctly. This results from the fact that analogy is a polysyllable initiated by an unstressed syllable and followed by a stressed syllable. semantics. As we have seen in section 3. when hearing the word analogy one may think that he has heard the words an allergy. the processing of phonetic material begins before the listener can possibly be able to decipher the communicative intention of the sender because he only has an insufficient amount of phonetic material to his disposal. it is still possible to understand an utterance without difficulty.4 Word Identification: The Importance of Context At the end of section 3. Then the selection function is employed where the one word is picked that matches the perceived signal the most. Nevertheless. one needs more information apart from the phonetic information: syntax. The phenomenon illustrated above describes the most frequent “slip of the ear”. The last step is the integration function where “the 4 Insertion of a boundary before a weak syllable (effective  effect of) Deletion of a boundary before a strong syllable (is he really  Israeli) Deletion of a boundary before a weak syllable (my george is  my gorgeous) 10 . 185 ff) 4.4 (Cutler and Butterfield 2003. For instance. there is the access stage where some portion of the acoustic signal activates possible lexical entries. there are four main types of boundary misplacements. earliest in order. Hence. There still is a considerable chance that a stressed syllable may be the onset of a grammatical word and that a lexical word is initiated by a weak syllable.to “slips of the ear”. there are more factors we rely on to comprehend correctly. According to Duffield. one might think that the first syllable correlates with the grammatical word an and that the following strong syllable initiates the lexical word allergy. In addition to the segmentation strategy. The identification of a word in continuous speech occurs in three steps. an occasion in which the receiver falsely inserts a word boundary making two words out of one. or incorrectly deletes one making one word out of two. which makes us anticipate an upcoming sound automatically.1 not all strong syllables initiate lexical words and not all unstressed syllables initiate grammatical words.1 it was stated that even if not all phonetic material is perceived. In fact.1.

is situated four words after the sound that is overlayed by a cough. The sentence was: “It was found that the *eel was on the axle/shoe/orange/table”. After listening to the sentence “The state governors met with their respective legi*latures convening in the capital city”5. the subjects stated that there were no sounds missing from the stretch of speech they had just been presented with. Thus. is missing in a given stretch of speech because it is overlayed with a noise. In a further study Warren & Warren demonstrated that “phonemic restorations appear to be effected over quite some distance and from constraints at all levels of language. heel. is needed to identify the word correctly. we do not even perceive it as imperfect. Obviously. 211 f) 4. Hence. From this study it becomes obvious that context determines the meaning. a longer stretch of speech. 212 ff) 5 The asterisk * stands for a cough 11 . and meal respectively. only the last word was different every time. which determines the damaged phonemic material.categorical. 214) In this experiment subjects listened to four variants of a spoken sentence. syntactic and semantic properties of the selected lexical entry [are abstracted] and [integrated] […] into the developing syntactic and conceptual representation of the utterance. Although the phonetic material is incomplete. But remarkably.”. Nevertheless.1. context. if partly overlayed by a cough. (Clark and Clark 1977. it only works due to semantic and syntactic constraints. Depending on the last word the subjects were presented with. in fact. they interpreted “*eel” to signify wheel. the constraining element. This ability to restore phonemes means that we are capable of naming a sound which. 38 ff) Clark and Clark have put this process more simple. 43. peel. we first identify a word by its acoustic signal. According to them. we test our hypothesis syntactically and semantically. (Clark and Clark 1977. who discovered a phenomenon called phonemic restoration effect.5 Discourse: Semantic and Syntactical Context To emphasize the importance of context I will describe an experiment run by Warren & Warren (1970). the restoration occurs automatically. for example. our ability to restore phonemes works extremely well. Secondly we elaborate a hypothesis about the meaning of the word and thirdly.”. (Clark and Clark 1977. (Duffield.

” (Bransford and Johnson 1973) 12 . it is better to do fewer things at once than too many. Any utterance or written paragraph is comprehended and recalled easier if surrounded by context. Harley states that context provides a frame for understanding. It also facilitates the process of disambiguation and it helps us to understand the non-literal meanings in language. and the whole cycle will then have to be repeated.1 Context To not repeat myself the following passages are true for discourse and text whereas prosody is only important for continuous speech. In the short run this might not seem important. Of course. Soon.1. After the procedure is completed. Eventually they will be used once more.1 we are only able to properly recall an information which was understood and comprehended. these constraining elements are also valid for the comprehension of written texts. Furthermore. regarding the phonemic restoration effect I considered it convenient to analyze discourse separately. I now want to discuss the experiment of Bransford and Johnson (1973) in detail. 4. otherwise you are pretty well set. In Bransford and Johnson’s study. subjects were presented with a story. one arranges the material into different groups again.2 In Discourse and Text 4. However. in this chapter the term context is more concerned with background knowledge and pragmatics than with syntax and semantics. It is important not to overdo things. However. If you have to go somewhere else due to lack of facilities. The procedure is actually quite simple. A mistake can be expensive as well. Then they can be put into their appropriate places. to avoid misunderstandings.2. however it will become just another facet of life. First you arrange things into two different groups. Coming back to the role of context which was skipped in 3. That is. They were divided into two groups but only one group new the title of the following confusing story. but then one can never tell. Nevertheless. that is the next step. As stated in the beginning of 3. context here refers to the extralinguistic factors which help us to make sense of what we perceive. It is difficult to foresee any end to the necessity for this task in the immediate future. one pile may be sufficient depending on how much there is to do. that is part of life. At first the whole procedure will seem complicated.In this passage we have only looked at semantic and syntactic context in discourse. That is. but complications can easily arise.

when it comes to understanding the meaning of text and discourse which goes beyond its literal meaning. Consequently. context does not only help to comprehend and recall better. This means. a sentence like “You’re a fine friend” does not become ironical on its own. the “title-group” recalled an average of 5. whereas the “no title-group” only remembered 2. to understand his intention. the sentence is meant ironical. I will not discuss the different models of language processing. According to Grice the participants of a conversation rely on the conversational maxims of quality. in this experiment it even decreased down to 2. In fact. 7 One instance where this is the case is irony where the sender’s intention is put in the nonliteral. 13 . (Grice 1975) Obviously. it does not improve the ability to recall. they make a ‘conversational implicature’6 to comprehend correctly.”. depending on its context. And this background knowledge needs to be activated before the exposure to the story. The same applies for the speech act theory which states that there are direct speech acts and indirect ones. 7 Since it is not important for this paper how a multiple meaning sentence is processed. and if they seem to be violated. quantity and relation. For instance. (McCabe 1998. McCabe names a few instances where one and the same sentences can be understood literally or figuratively.The subjects who possessed the information that the story was called the “clothes washing” story had no problem to follow the plot. If provided with the context after reading or hearing the story.7 ideas. In the former the sender uses “common syntactic forms to encode the common linguistic function for which they are specially designed. that we are able to understand the literal meaning of an information as well as the figurative one. the figurative form of the sentence. If not. but sometimes to comprehend at all. 285. This intention may be deduced from the non-literal meaning of the oral input. Only when used in a context where the addressed person has done something to upset the sender. it might as well be meant literally. (Harley 1995. manner. In numbers. it is the context that makes us understand a given information one way or another. indirect speech acts are the ones where the sender’s 6 A conversational implicature is made to understand the content of an utterance that is suggested by the sender. However. One needs to be communicatively competent to maintain a conversation or to read between the lines. Searle & Kiefer & Bierwisch 1980) As opposed to this. 213 f) However.8 ideas. they were also able to recall more ideas of the story than the group who did not know what the story was about.8 ideas out of 18. In this experiment it was the context in form of background knowledge which facilitated the recall.

From the point of view of pragmatics. or clause to which another word (especially a following relative pronoun) refers back.intention is conveyed in the non-literal meaning of the message. Thus. the bridging inference. a person who has no understanding of Christian Western cultures would probably not know what the sentence is all about. there are three types of inferences: the logical inference. The first one results from the direct meaning of the words. and the elaborative inference. If we consider the sentence “They carved the bird for Thanksgiving”. it is very probable that person A wants person B to close the window. Yet. For instance.2 Inferences To comprehend the situation-dependent meaning of an information. Nevertheless. (McCabe 1998. (McCabe 1998. (Barsalou 1998. 108 f) 4. Barsalou states that sometimes only the implied meaning is understood and processed.8 The bridging inference is made to figure out to which word a specific cohesive device refers. in a situation where two people are in a room and the window is open. but does not want to say so directly. For instance. For example. in “Tom has never been married. 285) For instance. We infer constantly and automatically. the sentence “Can you pass the salt?” has the surface form of a question but the meaning: request. 280) In accordance with Harley. person A tries to transmit his intention through the implied meaning of the sentence.2. in the indirect speech act “It is cold in here”.” the pronominal reference “He” refers back to its antecedent9 “Tom”. However. phrase. These “inferences are deductions or guesses based on evidence in the text or derived from a person’s preexisting knowledge. Yet. not the literal one. the utterance is rather a request than a mere statement. Finally. there is the elaborative inference which follows from the world knowledge of the receiver. depending on the context and salience of an utterance.”.” (Oxford Dictionary) 14 . He is a bachelor. we do not have to make an effort to draw inferences. they form a salient pattern where the implied meaning is activated automatically and the literal meaning which asks for the physical ability is bypassed. since this particular surface form and meaning frequently co-occur. After a short time we are not even aware 8 9 The female form of the word is „bachelorette“ (AE) or „single woman“ (BE) Definition of antecedent: “an earlier word. we do not encounter any difficulty to infer that this bird is a turkey. the receiver draws inferences. a possible logical inference we can draw from the sentence “He is a bachelor” is that this person is male because the word bachelor is restricted to men.

was all about a moral. This is the case because we know from our world knowledge that a fish which swims beneath the turtles (“them”) that are on the log. (Harley 1995. They rather recited details about the literal plot and the setting. interpretations. McCabe explains that schemata are mental structures acquired through many experiences […]. people from different cultures remember different features of a text.of the inferences we made any more.g. This happened because they have no schemata for moral because the concept of moral does not exist in their culture. who where very aware of the fact that this story was designed to convey a moral. the American children. In their experiment Ponam (New Guinea) and American children had to retell a European story. the inference one makes may be influenced by the wording of a question. the Ponam children did not grasp that this story. Bransford. but missed the point of the story. [They] guide people by setting up their expectations for what usually will happen and helping them interpret what does happen and remember what in fact did happen […]. and recall. (McCabe. also swims beneath the log (“it”). retold the plot focusing on the resolution of the story. They have demonstrated that. Barclay.10 Curiously. “The Boy Who Cried Wolf”. Therefore. since schemata influence our expectations. In contrast.3 10 Cohesion The story is in the appendix 15 . (McCabe 1998. the subjects were not able to differentiate the inference from the original.”. it is difficult to distinguish an assertion from an implication (e.2. in advertising) and on the other hand inferences are guided by our culture-specific schemata. When asked whether they were presented with the original sentence “Three turtles rested on a floating log and a fish swam beneath them” or with the valid inference “Three turtles rested on a floating log and a fish swam beneath it”. and Franks (1972) demonstrated in an experiment that subjects are not able to tell an original information from a valid inference after a brief moment. 281) That schemata are different in every culture is proven by a study run by Invernizzi and Abouzeid (1995). On the one hand. 281 f) 4. 216 f) As we have seen in the “Thanksgiving-example” not all people draw the same inferences from one and the same source.

anaphora occurs much more often than cataphora. is a “non-overt constituent” of a sentence. (McCabe 1998. 4.2. substitution. if we see 11 An example for anaphora: President George Bush has claimed he was told by God to invade Iraq. (Fodor 1995. 277). the action that is carried out by Boris is realized by an Empty Category. lexical cohesion/reiteration. 1995. In fact. The former describes the use of cohesive devices to refer back to prior text or discourse. 4). we would not be able to understand the sentence because there is an obligatory sentence constituent missing: the verb. One of these rules is the maintenance of cohesion. a sentence constituent. According to Harley. However. the anaphor is based on the syntax of this sentence. 220 f) Fodor explains that an Empty Category. there are surface anaphors and deep anaphors. 12 Since anaphora is much more frequent. the soldier (= source word) left the barracks. but their meaning must be deduced by the receiver. (Harley. 1995. The President (= antecedent) made the assertion during his (= anaphora) first meeting with Palestinian leaders in June 2003. 107) This means that these constituents are neither realized orthographically nor phonologically. I will focus on this concept. an EC.”. ellipsis. In other words cohesive “devices […] glue a group of sentences together. Thus. This is a “semantic concept that refers to relations of meaning that exist within the text and that define it as a text”. In accordance with Harley. and rules of discourse. and conjunctions. but only the second one “And Boris a bat. (Harley.2. in the example “Tom caught a rat and Boris a bat”. 221) There are two different types of cohesion: anaphora and cataphora. context. (Fodor 1995. 12 An example for cataphora: After he (=cataphora) had received his orders. For instance. 277 f) I will come back to these categories later. a cohesive device which refers to upcoming text or discourse. 11 However. The distinguishing criterion is that a surface anaphor depends on the syntactic form of a sentence only. the use of all of these devices is related to the drawing of bridging inferences. 107 f) If we had not seen the first part of this sentence.”. this process of “deciphering” the antecedent is called resolution. 16 .”.In the passage about inferences. And “anaphoric resolution is nothing other than a […] backward [bridging] inference that we must do to maintain a coherent representation of the text. These cohesive devices fall into five main groups: reference. (Haliday & Hasan 1976. (McCabe 1998. Since we need these cohesive devices to make a text comprehensible. it was said that these deductions are based on prior knowledge. we also need to understand which antecedent it belongs to.

in the sentence “Tom likes Boris so he gave him a present. Here. In the 17 . the resolution becomes slightly more difficult. In the second paragraph of this passage I talked about resolution. the EC assumes the properties and meaning of the antecedent it relates to. That is. there is substitution. and in “Tom sold Boris his bike because he needed it”. in “Tom sold Boris his bike because he hated it”. the second anaphor to the second antecedent. the mentioning of an office to refer to its holder: “George Walker Bush” (= antecedent) is an American politician who served as the 43rd President of the United States (= substitution). also called reiteration. different words refer to the same thing. if this order is violated. too.” Finally. we rely on a strategy called parallel function. So when it comes to comprehension in consideration of cohesion. Among them there are for example additives. but to make out the cohesive device itself as well. adversatives. To discuss this process further I want to demonstrate that there are occasions when it is difficult to know what the sender refers to. because of our world knowledge. Normally. Consequently. 277 f). there is the concept of reference where one word refers to another part of the sentence.”. Nevertheless. In addition there is lexical cohesion. However. Coming back to McCabe’s classification of cohesive devices. These words may be personal pronouns. Put simply. For instance. For example.the two parts together we have no difficulty in understanding the sentence because the verb (“caught”) is realized in the first part and specifies the meaning of the EC in the second part. demonstratives. we automatically prefer to match the first anaphor to the first antecedent. if there is more than one anaphor. the EC obviously belongs to the concept called ellipsis which is the leaving of gaps that are filled in automatically. there are conjunctions to make a text cohesive. definite articles or comparatives. we see that there are certain impediments. This can also be seen in Footnote 11 where “President George Bush” (= antecedent) is referred to with “The President” (= reiteration). “he” refers back to “Tom” and “him” refers back to “Boris”. (McCabe 1998. Harley explains. a concept where clear synonyms are used or the same word is repeated in different sentences. we are able to solve cases like these. we have two possible antecedents but only one pronominal reference. Moreover. For an example see footnote 10 where the demonstrative pronoun “his” refers back to its antecedent “The President”. and so on. causals or temporals. not only to make out the antecedent of a cohesive device. Furthermore.

in discourse. Yet. I have discussed the following topics: prosody. and in the latter Boris buys the bike because he needs it. We have looked at garden path sentences. Finally.former Tom sells his bike because he hates it. we have to interpret a situation correctly to know what someone wants us to do or to act according to the circumstances. the problems caused by the segmentation strategy we apply. which is a main resource to draw on in discourse. in continuous speech cohesive devices are probably applied less than in text. there is the extralinguistic world which helps us to disambiguate certain lacks of clarity or violations of the parallel function. is impeded by variations in the acoustic signal and lacks of the segmentation strategy. I only want to briefly summarize the points that enhance our ability to comprehend. context in form of background knowledge and pragmatics. the immediate semantic and syntactical context does not only assist the process of comprehension. Concerning background knowledge and pragmatics. and we do not remember details. semantic and syntactical context. 18 . However. The process of understanding correctly and inferring the intention a sender has tried to transmit is fairly difficult. inferences. They assist us in the comprehension process. Precisely. Regarding the part about top-down processing. For every factor that facilitates comprehension there is one factor that hinders comprehension. this applies for inferences and cohesion. oral and written information is processed linearly from left to right. but also generates ambiguity and garden path sentences. we have seen the cohesive devices we apply to make text and discourse comprehensible. but the gist of a story. but sometimes we draw wrong deductions or we cannot make out the right antecedent. and we have learned how we make out an antecedent. 221 f) To sum this passage up. and cohesion. Besides. information is stored in the form of propositions. 5 Conclusion In this paper I have analyzed how language is processed and I have also pointed out the role top-down processing plays for the comprehension of text and discourse. (Harley 1995. in ‘face-to-face interaction’. That is. Prosody. and at the complicated task to infer the intended meaning of an indirect speech act. too.

I suspect that prosody is the most helpful resource to draw on for discourse. comprehension is almost always achieved. However. Reference List American Heritage Dictionary <http://dictionary. 2011).reference. It is exactly this absence of intonation that makes the understanding of written text difficult.Nevertheless. apart from some exceptions and misunderstandings. whereas in text I consider it most advantageous that one has the possibility to go back in the text to double-check the information one has just read. there are facilitating factors for both.com/browse/text> (accessed July 20. Yet. one cannot say whether it is easier to understand text or discourse. As I have summarized above. and it is also the absence of a double-checking device that complicates the comprehension of continuous speech. 19 .

& Butterfield. Hillsdale: Lawrence Erlbaum Associates. N. Visual Information Processing. New York: Harcourt Brace Jovanovich. Harley. M.shef. 2nd Ed. Field. Ed. 1992. A. 1-14. 2011). Journal of Narrative and Life History. S. 1972.ac. & Franks. Ed. Hove: Erlbaum. Ed. A. 193–209. L. & Barclay. “Sentences Combined: Text and Discourse”. Cognitive Psychology: An Overview for Cognitive Scientists. Gleason J. T. Clark. & Hasan. Taylor & Francis.uk/papers/eolls6. Duffield. Cole. 275-300.L. 2003. Psycholinguistics. J. P. J.Barsalou. “Consideration of Some Problems of Comprehension”. M. Invernizzi. <http://www. 20 . “One story map may not fit all: A crosscultural analysis of children's written story retellings”. 213. 1976.ngduffield. 1995. Fodor.staff. N.P. H. Psycholinguistics. & Morgan. Bransford. R. “Sentence memory: A constructive vs. J. “Locating Word Boundaries”. Fort Worth: Harcourt Brace College Publ. The Psychology of Language: From Data to Theory. 185-190. Longman Dictionary of the English Language. J. & Liberman. Cuttler. 1998. Speech Acts. Cohesion in English. M. Field. & Clark. A. J.K. 1995. Psycholinguistics. D.pdf> (accessed July 25. A Resource Book for Students. R. Bransford. M. 1995. & Bernstein Ratner. (5). An Invitation to Cognitive Science. 1975.W. J. McCabe. M. 1995. H. “Logic and Conversation”. 1977. Cambridge: MIT Press.G. London: Longman Group. Cognitive Psychology. Chase. J. Gleitman. & Abouzeid. W. K. E. 5. “Comprehending Sentence Structure”.G. 2003. London: Longman UK Group Limited. Language. Psychology and Language: An Introduction to Psycholinguistics. Ed. A Resource Book for Students.C.D. Ed.B. 41-58. L. London: Routledge. J. J. Psycholinguistics. New York: Academic Press. interpretive approach”. R. 262-263. & Johnson. Halliday. 1973. London: Routledge. New York: Academic Press. 400. 3. Longman. Grice. A.

J. Psycholinguistics. A. J. Speech Act Theory and Pragmatics. 2001. Journal of Experimental Psychology. Oxford: Oxford University Press. “Auditory Illusions and Confusions”. J. S. Reidel Publishing Company. & Kiefer.R.com/definition/antecedent> (accessed July 21. 1974. 30-37.P. J. 2011) Scovel. & Warren. & McHugh.G. M. J. & Bierwisch. Shields. F. 250-255. M. Oxford: Oxford University Press.Nattinger. R. 223. Appendix • Footnote 1: “The clothes washing story”: 21 . 102. Dortrecht: D. Scientific American. & Martin. Warren. “Reaction time to phoneme targets as a function of rhythmic cues in continuous speech”. Searle. Lexical Phrases and Language Teaching. 1998. T. 1980. & DeCarrico. Oxford Dictionary <http://oxforddictionaries. R.L.

one pile may be sufficient depending on how much there is to do. but the house caught on fire. 282) 22 . A mistake can be expensive as well. 212 ff) • Footnote 10: “The Boy Who Cried Wolf”: o Ponam (Papua. You should stay home with your grandfather. (McCabe.“The procedure is actually quite simple. but nobody came. However. ”Kalai. It is difficult to foresee any end to the necessity for this task in the immediate future. After the procedure is completed. but it was too late. First you arrange things into two different groups. It is important not to overdo things. Soon. If you have to go somewhere else due to lack of facilities. Kalai said. Everything had burnt down to the ground. one arranges the material into different groups again. but complications can easily arise. Kalai’s mother always carried him everywhere. When they arrived they saw nothing o American version: Kalai was running up and down the beach yelling “Fire. Of course. you are to small to go out fishing in the sea. Suddenly they saw the flames and the smoke and they came. fire. that is part of life. that is the next step. One day Kalai’s mother and father went out fishing: Kalai’s mother said.” Kalai was lonely on the beach. That is. He got his red laplap and ran down to the beach and waved his laplap to his family and said. The next day came.”How could I get my family home?” He sat down and decided to get his family home. They came home. however it will become just another facet of life. Nobody came. In the short run this might not seem important. New Guinea) version: Once upon a time Kalai and his family they lived on an island. The next day the same thing happened.” His brother saw his laplap and went home.” Everybody came home. He ran up and down the beach. At first the whole procedure will seem complicated. otherwise you are pretty well set. it is better to do fewer things at once than too many. but then one can never tell. fire.” (Harley 1995. Kalai kept waving the flag. Eventually they will be used once more. and his brother told him if he kept telling lies that nobody will come when you call for help. Then they can be put into their appropriate places. and the whole cycle will then have to be repeated. “Fire.