Gilbert K. Krulee

THE INFLUENCE OF SYNTACTIC ERRORS ON SENTENCE RECOGNITION.

In this experiment, errors have been introduced into sentences in a systematic way in order to study their effect on S performance. More specifically, the experimental variables include two basic syntactic structures; three alternative positions in sentences for the introduction of an error in place of the subject noun, the main verb, or the object noun; two parts of speech, adjective and adverb, for the introduction of an error; and presentation with and without knowledge of the presence or absence of an error in the presentation. In general, the effect of an error is to increase average viewing time and knowledge of what to expect has no effect on S performance. Errors in place of the main verb are more disruptive than are errors in either the subject or object position. Adverbs as errors have approximately the same effect as do adjectives, although there is a complicated interaction between type of sentence, form of error, and position of error. There is a difference in viewing time for the two types of sentences. However, the difference is particularly apparent for sentences with errors. A discussion of certain implications is included.

Word Processing and Effects of Variability in Type Fonts

Current models for reading words often assume the existence of features or shape characteristics for distinguishing one letter from another. However, the form of these features need not be invariant, particularly as one considers variations in type font or in handwriting. In this paper, we introduce the possibility of a two-level process with one level helping the viewer to identify the style or font being used at a given time. This first process leads to the activation of a decision tree that is specialized with respect to a particular set of characteristics while a second process makes use of this decision tree to identify the letter or word being presented. The results of an experiment designed to explore empirical implications of this two-level approach are described. Results are consistent with the model although additional experimentation is needed to explore more fully the implications for the processing of letters and words.

Speech perception as a multilevel processing system

This is the report of an experiment that emphasized how variability in voice characteristics will affect a listeners ability to process sentences. Consideration is also given to the description of a possible model for the processing of continuous speech. As part of the experiment, use was made of a number of speakers, each of whom recorded the same set of sentences. The recorded sentences were used in order to form a pair of experimental tapes. On one tape, sentences were arranged randomly such that the listener could predict neither the content of a sentence nor the voice of the speaker who had produced it. On a second tape, sentences were arranged in blocks by speakers such that there was uncertainty as to content but not with respect to voice. Four groups of subjects listened to these tapes, with each tape being processed under a low and high condition of background noise. In interpreting the data, emphasis was placed on the idea that the speech signal contains a variety of characteristics or features and that these features are processed by three interacting subsystems. One is a prosodic system, responsible for the segmentation of continuous speech into sentences, phrases, and words. It attempts to establish a context within which a second system, responsible for the processing of words and syllables, can operate. However, this pair of systems is speaker-dependent in that it makes use of features that need to be adjusted on the basis of an assessment of voice characteristics. Thus, the model also provides for the inclusion of a third system, responsible for the assessment of voice characteristics. In effect, this third subsystem is responsible for “training” the two primary systems in order to normalize for differences in individual voice characteristics. We had predicted that performance would be uniformly more accurate in the blocked condition than in the unblocked condition since one type of uncertainty would have been eliminated. The results confirm this hypothesis only in part since there is an interaction with variations in backgroud noise. However, a plausible explanation is offered for these findings and for a variety of related results.

Organizing factors in the comprehension and recall of connected discourse

This experiment is based on the theory that, in reading, one uses cues in order to activate an organizing system as an aid in reading the material. The variables introduced into the experiment include (1) the form of organization (meaningful, scrambled, and pseudo) and (2) the choice of material (descriptive, abstract, and technical). Having read a paragraph, subjects read a list of words including some from the paragraph and some not and were asked to check those words that they remembered as belonging to the paragraph. Accuracy is highest when subjects have been most successful in activating an appropriate organizing system.

ORGANIZING FACTORS AND IMMEDIATE MEMORY SPAN.

In a series of four experiments, the processes used in decoding visual inputs were studied, with particular attention given to the influence of organizing factors in immediate memory span. In one experiment, the effect of variations in the number of alternatives per item of input was studied. In a second, the effect of different grouping procedures was evaluated. In a third, variations in internal organization of the input strings were introduced in order to show that recoding is more likely when naturally pronounceable units are available. The final experiment focused on the effect of prior knowledge of the length of the stimulus strings. The results confirmed the difficulty of applying informational measures to immediate memory span and the importance of other variables that influence the ability of Ss to recode complex inputs into more manageable units.

Scanning Processes and Sentence Recognition

This report summarizes an experiment designed to study the disruptive effects on immediate recall of interfering with the ability to scan ahead. Sentences were printed on paper tape which was pulled through a window, the size of which could be controlled. There were two exposure widths, three rates of tape movement, and three types of material which varied in length: equational and transitive sentences as well as random strings of words. Errors increase as the exposure width is decreased, and the magnitude of the disruptive effect is most marked for the longer sentences and the faster rate of presentation. A possible explanation is advanced emphasizing the likelihood that Ss store information in a sensory register and scan that register in order to segment sentences into phrases, phrases into words, and words into smaller functional units. As one limits exposure width, one interferes with the transfer of information into the sensory register and with the processing of information after it is stored in the register.

Organizing factors in remembering and comprehending: A developmental analysis

This report describes an experiment that emphasizes: (a) the influence of organizing factors in remembering and comprehending and (b) the development of those skills that are related to the ability to organize. The analysis of the data was based on a theory of comprehension that emphasized a pair of interrelated processes, one providing for the construction of a schema or plan and the other for the semantic integration of items into the evolving plan. As predicted, adults perform more accurately with the paragraphs vs. the scrambled paragraphs and with the word lists as compared to the scrambled word lists. Accuracy with unrelated words is generally greater than with related words, a finding that is consistent with a reconstructive theory of remembering. The pattern of results for the 5th grade Ss is quite different: they perform more accurately with the scrambled paragraphs than with the paragraphs and somewhat more accurately with scrambled word lists rather than with word lists. These results were interpreted as evidence for a problem of control for the 5th grade Ss. We have assumed that the ability to process certain forms of information depends upon the ability to select and maintain an appropriate focus of attention. The results for these younger Ss suggest that their ability to file information into a schema as a filing system is compromised by the need to change or modify the filing system.

Semantic Anomalies and Sentence Recognition

This report summarizes the results of an experiment in which Ss read for immediate recall sentences which either were meaningful or were constructed to be anomalous in particular ways. Three types of sentences were used: equational, transitive, and intransitive. There were also three types of anomalies involving an anomaly either in the subject phrase of the sentence, in the predicate phrase, or in the relationship between the subject and predicate phrase. From the results, equational sentences can be processed most easily. For all types of sentences, the presence of an anomaly has a disruptive effect in that time to process a sentence is increased. However, the effect of a particular type of anomaly appears to be a function of the type of sentence in which it occurs. A possible explanation for these findings, emphasizing the use and verification of the selectional restrictions associated with each sentence, was advanced.

Resolution of ambiguity in Chinese and its application to machine translation

In this paper, we present a typology of ambiguity in Chinese, which includes morphological, lexical, syntactic, semantic, and contextual ambiguities. Examples are shown for each type of ambiguity and sometimes for subtypes. Ambiguity resolution strategies used in the ALICE machine translation system are presented in various levels of detail. A disambiguation model, called Four-Step, is proposed for resolving syntactic ambiguities involving serial verb construction and predication. As the name suggests, the model comprises four steps-well-formedness checking, preference for argument readings, precondition checking, and late closure. For resolving semantic ambiguity, we propose a new formalism, called Semantic Functional Grammar (SFG), to deal with the resolution problem. SFG integrates the concept of Semantic Grammar into Lexical-Functional Grammar (LFG) such that the functional structure (f-structures) include semantic functions in addition to grammatical functions. For dealing with lexical and contextual ambiguities, we briefly describe the mechanisms used in the ALICE system. As for morphological ambiguity, the resolution is a problem of word-boundary decision (segmentation) and is beyond the scope of this research. The mechanisms presented in the paper have been successfully applied to the translation of Chinese news headlines in the ALICE system.

Two‐level representations for natural language

Abstract For a number of years, transformational grammars have been extra‐ordinarily successful as models of natural language. Nevertheless, there has emerged a body of criticism based partly on formal as well as on practical considerations. As to the formal criticism, it emphasizes mathematical results which suggest that reliance on a transformational component might well be unnecessary. On the other hand, the more practical line of argument emphasizes the need for improved models of language from the standpoint both of performance as well as computational adequacy. The emphasis in this paper is directed toward the importance of two‐level grammars as a formal alternative. Two‐level grammars have an advantage in that each level in the grammar can be written in what appears to be a context free form. Even so, two‐level grammars have a generative capacity that is equivalent to that of a transformational grammar. After reviewing some mathematical preliminaries, a number of example grammars are introduced for...