Publications

Abstract

Incremental sentence generation imposes special constraints on the representation of the grammar and the design of the formulator (the module which is responsible for constructing the syntactic and morphological structure). In the model of natural speech production presented here, a formalism called Segment Grammar is used for the representation of linguistic knowledge. We give a definition of this formalism and present a formulator design which relies on it. Next, we present an object- oriented implementation of Segment Grammar. Finally, we compare Segment Grammar with other formalisms.

Abstract

A quantitative model of human categorization behavior is proposed, which can be applied to 4-alternative forced-choice categorization data involving two binary classifications. A number of processing dependencies between the two classifications are explicitly formulated, such as the dependence of the location, orientation, and steepness of the class boundary for one classification on the outcome of the other classification. The significance of various types of dependencies can be tested statistically. Analyses of a data set from the literature shows that interesting dependencies in human speech recognition can be uncovered using the model.

Abstract

The temporal distribution of perceptually relevant information for consonant recognition in British English VCVs is investigated. The information distribution in the vicinity of consonantal closure and release was measured by presenting initial and final portions, respectively, of naturally produced VCV utterances to listeners for categorization. A multidimensional scaling analysis of the results provided highly interpretable, four-dimensional geometrical representations of the confusion patterns in the categorization data. In addition, transmitted information as a function of truncation point was calculated for the features manner place and voicing. The effects of speaker, vowel context, stress, and distinctive feature on the resulting information distributions were tested statistically. It was found that, although all factors are significant, the location and spread of the distributions depends principally on the distinctive feature, i.e., the temporal distribution of perceptually relevant information is very different for the features manner, place, and voicing.

Abstract

Finnish vowel harmony rules require that if the vowel in the first syllable of a word belongs to one of two vowel sets, then all subsequent vowels in that word must belong either to the same set or to a neutral set. A harmony mismatch between two syllables containing vowels from the opposing sets thus signals a likely word boundary. We report five experiments showing that Finnish listeners can exploit this information in an on-line speech segmentation task. Listeners found it easier to detect words likehymyat the end of the nonsense stringpuhymy(where there is a harmony mismatch between the first two syllables) than in the stringpyhymy(where there is no mismatch). There was no such effect, however, when the target words appeared at the beginning of the nonsense string (e.g.,hymypuvshymypy). Stronger harmony effects were found for targets containing front harmony vowels (e.g.,hymy) than for targets containing back harmony vowels (e.g.,paloinkypaloandkupalo). The same pattern of results appeared whether target position within the string was predictable or unpredictable. Harmony mismatch thus appears to provide a useful segmentation cue for the detection of word onsets in Finnish speech.

Abstract

In this study the N400 component of the event-related potential was used to investigate spoken sentence understanding in Broca's and Wernicke's aphasics. The aim of the study was to determine whether spoken sentence comprehension problems in these patients might result from a deficit in the on-line integration of lexical information. Subjects listened to sentences spoken at a normal rate. In half of these sentences, the meaning of the final word of the sentence matched the semantic specifications of the preceding sentence context. In the other half of the sentences, the sentence-final word was anomalous with respect to the preceding sentence context. The N400 was measured to the sentence-final words in both conditions. The results for the aphasic patients (n = 14) were analyzed according to the severity of their comprehension deficit and compared to a group of 12 neurologically unimpaired age-matched controls, as well as a group of 6 nonaphasic patients with a lesion in the right hemisphere. The nonaphasic brain damaged patients and the aphasic patients with a light comprehension deficit (high comprehenders, n = 7) showed an N400 effect that was comparable to that of the neurologically unimpaired subjects. In the aphasic patients with a moderate to severe comprehension deficit (low comprehenders, n = 7), a reduction and delay of the N400 effect was obtained. In addition, the P300 component was measured in a classical oddball paradigm, in which subjects were asked to count infrequent low tones in a random series of high and low tones. No correlation was found between the occurrence of N400 and P300 effects, indicating that changes in the N400 results were related to the patients' language deficit. Overall, the pattern of results was compatible with the idea that aphasic patients with moderate to severe comprehension problems are impaired in the integration of lexical information into a higher order representation of the preceding sentence context.

Abstract

This study investigates whether spoken sentence comprehension deficits in Broca's aphasics results from their inability to access the subordinate meaning of ambiguous words (e.g. bank), or alternatively, from a delay in their selection of the contextually appropriate meaning. Twelve Broca's aphasics and twelve elderly controls were presented with lexical ambiguities in three context conditions, each followed by the same target words. In the concordant condition, the sentence context biased the meaning of the sentence final ambiguous word that was related to the target. In the discordant condition, the sentence context biased the meaning of the sentence final ambiguous word that was incompatible with the target.In the unrelated condition, the sentence-final word was unambiguous and unrelated to the target. The task of the subjects was to listen attentively to the stimuli The activational status of the ambiguous sentence-final words was inferred from the amplitude of the N399 to the targets at two inter-stimulus intervals (ISIs) (100 ms and 1250 ms). At the short ISI, the Broca's aphasics showed clear evidence of activation of the subordinate meaning. In contrast to elderly controls, however, the Broca's aphasics were not successful at selecting the appropriate meaning of the ambiguity in the short ISI version of the experiment. But at the long ISI, in accordance with the performance of the elderly controls, the patients were able to successfully complete the contextual selection process. These results indicate that Broca's aphasics are delayed in the process of contextual selection. It is argued that this finding of delayed selection is compatible with the idea that comprehension deficits in Broca's aphasia result from a delay in the process of integrating lexical information.

Abstract

This volume on Inuit speech follows the evolution of a native language of the North American Arctic, from its historical roots to its present-day linguistic structure and patterns of use from Alaska to Greenland. Drawing on a wide range of research from the fields of linguistics, anthropology, and sociology, Dorais integrates these diverse perspectives in a comprehensive view of native language development, maintenance, and use under conditions of marginalization due to social transition.

Abstract

Although children's knowledge of the sound patterns of words has been a focus of debate for many years, little is known about the lexical representations very young children use in word recognition. In particular, researchers have questioned the degree of specificity encoded in early lexical representations. The current study addressed this issue by presenting 18–23-month-olds with object labels that were either correctly pronounced, or mispronounced. Mispronunciations involved replacement of one segment with a similar segment, as in ‘baby–vaby’. Children heard sentences containing these words while viewing two pictures, one of which was the referent of the sentence. Analyses of children's eye movements showed that children recognized the spoken words in both conditions, but that recognition was significantly poorer when words were mispronounced. The effects of mispronunciation on recognition were unrelated to age or to spoken vocabulary size. The results suggest that children's representations of familiar words are phonetically well-specified, and that this specification may not be a consequence of the need to differentiate similar words in production.

Abstract

The roles which word class (open/closed) and sentential stress play in the sentence comprehension processes of both agrammatic (Broca's) aphasics and normal listeners were examined with a word monitoring task. Overall, normal listeners responded more quickly to stressed than to unstressed items, but showed no effect of word class. Aphasics also responded more quickly to stressed than to unstressed materials, but, unlike the normals, responded faster to open than to closed class words regardless of their stress. The results are interpreted as support for the theory that Broca's aphasics lack the functional underlying open/closed class word distinction used in word recognition by normal listeners.

Abstract

A growing number of researchers in the sentence processing community are using eye movements to address issues in spoken language comprehension. Experiments using this paradigm have shown that visually presented referential information, including properties of referents relevant to specific actions, influences even the earliest moments of syntactic processing. Methodological concerns about task-specific strategies and the linking hypothesis between eye movements and linguistic processing are identified and discussed. These concerns are addressed in a review of recent studies of spoken word recognition which introduce and evaluate a detailed linking hypothesis between eye movements and lexical access. The results provide evidence about the time course of lexical activation that resolves some important theoretical issues in spoken-word recognition. They also demonstrate that fixations are sensitive to properties of the normal language-processing system that cannot be attributed to task-specific strategies

Abstract

This work presents a salvage grammar of the Biri language of Eastern Central Queensland, a Pama-Nyungan language belonging to the large Maric subgroup. As the language is no longer used, the grammatical description is based on old written sources and on recordings made by linguists in the 1960s and 1970s. Biri is in many ways typical of the Pama-Nyungan languages of Southern Queensland. It has split case marking systems, marking nouns according to an ergative/absolutive system and pronouns according to a nominative/accusative system. Unusually for its area, Biri also has bound pronouns on its verb, cross-referencing the person, number and case of core participants. As far as it is possible, the grammatical discussion is ‘theory neutral’. The first four chapters deal with the phonology, morphology, and syntax of the language. The last two chapters contain a substantial discussion of Biri’s place in the Pama-Nyungan family. In chapter 6 the numerous dialects of the Biri language are discussed. In chapter 7 the close linguistic relationship between Biri and the surrounding languages is examined.

Abstract

In normal conversation, speakers translate thoughts into words at high speed. To enable this speed, the retrieval of distinct types of linguistic knowledge has to be orchestrated with millisecond precision. The nature of this orchestration is still largely unknown. This report presents dynamic measures of the real-time activation of two basic types of linguistic knowledge, syntax and phonology. Electrophysiological data demonstrate that during noun-phrase production speakers retrieve the syntactic gender of a noun before its abstract phonological properties. This two-step process operates at high speed: the data show that phonological information is already available 40 milliseconds after syntactic properties have been retrieved.

Abstract

The temporal properties of semantic and phonological processes in speech production were investigated in a new experimental paradigm using movement-related brain potentials. The main experimental task was picture naming. In addition, a 2-choice reaction go/no-go procedure was included, involving a semantic and a phonological categorization of the picture name. Lateralized readiness potentials (LRPs) were derived to test whether semantic and phonological information activated motor processes at separate moments in time. An LRP was only observed on no-go trials when the semantic (not the phonological) decision determined the response hand. Varying the position of the critical phoneme in the picture name did not affect the onset of the LRP but rather influenced when the LRP began to differ on go and no-go trials and allowed the duration of phonological encoding of a word to be estimated. These results provide electrophysiological evidence for early semantic activation and later phonological encoding.

Abstract

This paper reports two experiments with vowels and consonants as phoneme detection targets in real words. In the first experiment, two relatively distinct vowels were compared with two confusible stop consonants. Response times to the vowels were longer than to the consonants. Response times correlated negatively with target phoneme length. In the second, two relatively distinct vowels were compared with their corresponding semivowels. This time, the vowels were detected faster than the semivowels. We conclude that response time differences between vowels and stop consonants in this task may reflect differences between phoneme categories in the variability of tokens, both in the acoustic realisation of targets and in the' representation of targets by subjects.

Abstract

Computer simulations in an instructional context can be characterized according to four aspects (themes): simulation models, learning goals, learning processes and learner activity. The present paper provides an outline of these four themes. The main classification criterion for simulation models is quantitative vs. qualitative models. For quantitative models a further subdivision can be made by classifying the independent and dependent variables as continuous or discrete. A second criterion is whether one of the independent variables is time, thus distinguishing dynamic and static models. Qualitative models on the other hand use propositions about non-quantitative properties of a system or they describe quantitative aspects in a qualitative way. Related to the underlying model is the interaction with it. When this interaction has a normative counterpart in the real world we call it a procedure. The second theme of learning with computer simulation concerns learning goals. A learning goal is principally classified along three dimensions, which specify different aspects of the knowledge involved. The first dimension, knowledge category, indicates that a learning goal can address principles, concepts and/or facts (conceptual knowledge) or procedures (performance sequences). The second dimension, knowledge representation, captures the fact that knowledge can be represented in a more declarative (articulate, explicit), or in a more compiled (implicit) format, each one having its own advantages and drawbacks. The third dimension, knowledge scope, involves the learning goal's relation with the simulation domain; knowledge can be specific to a particular domain, or generalizable over classes of domains (generic). A more or less separate type of learning goal refers to knowledge acquisition skills that are pertinent to learning in an exploratory environment. Learning processes constitute the third theme. Learning processes are defined as cognitive actions of the learner. Learning processes can be classified using a multilevel scheme. The first (highest) of these levels gives four main categories: orientation, hypothesis generation, testing and evaluation. Examples of more specific processes are model exploration and output interpretation. The fourth theme of learning with computer simulations is learner activity. Learner activity is defined as the ‘physical’ interaction of the learner with the simulations (as opposed to the mental interaction that was described in the learning processes). Five main categories of learner activity are distinguished: defining experimental settings (variables, parameters etc.), interaction process choices (deciding a next step), collecting data, choice of data presentation and metacontrol over the simulation.

Abstract

The use of computer simulations in education and training can have substantial advantages over other approaches. In comparison with alternatives such as textbooks, lectures, and tutorial courseware, a simulation-based approach offers the opportunity to learn in a relatively realistic problem-solving context, to practise task performance without stress, to systematically explore both realistic and hypothetical situations, to change the time-scale of events, and to interact with simplified versions of the process or system being simulated. However, learners are often unable to cope with the freedom offered by, and the complexity of, a simulation. As a result many of them resort to an unsystematic, unproductive mode of exploration. There is evidence that simulation-based learning can be improved if the learner is supported while working with the simulation. Constructing such an instructional environment around simulations seems to run counter to the freedom the learner is allowed to in ‘stand alone’ simulations. The present article explores instructional measures that allow for an optimal freedom for the learner. An extensive discussion of learning goals brings two main types of learning goals to the fore: conceptual knowledge and operational knowledge. A third type of learning goal refers to the knowledge acquisition (exploratory learning) process. Cognitive theory has implications for the design of instructional environments around simulations. Most of these implications are quite general, but they can also be related to the three types of learning goals. For conceptual knowledge the sequence and choice of models and problems is important, as is providing the learner with explanations and minimization of error. For operational knowledge cognitive theory recommends learning to take place in a problem solving context, the explicit tracing of the behaviour of the learner, providing immediate feedback and minimization of working memory load. For knowledge acquisition goals, it is recommended that the tutor takes the role of a model and coach, and that learning takes place together with a companion. A second source of inspiration for designing instructional environments can be found in Instructional Design Theories. Reviewing these shows that interacting with a simulation can be a part of a more comprehensive instructional strategy, in which for example also prerequisite knowledge is taught. Moreover, information present in a simulation can also be represented in a more structural or static way and these two forms of presentation provoked to perform specific learning processes and learner activities by tutor controlled variations in the simulation, and by tutor initiated prodding techniques. And finally, instructional design theories showed that complex models and procedures can be taught by starting with central and simple elements of these models and procedures and subsequently presenting more complex models and procedures. Most of the recent simulation-based intelligent tutoring systems involve troubleshooting of complex technical systems. Learners are supposed to acquire knowledge of particular system principles, of troubleshooting procedures, or of both. Commonly encountered instructional features include (a) the sequencing of increasingly complex problems to be solved, (b) the availability of a range of help information on request, (c) the presence of an expert troubleshooting module which can step in to provide criticism on learner performance, hints on the problem nature, or suggestions on how to proceed, (d) the option of having the expert module demonstrate optimal performance afterwards, and (e) the use of different ways of depicting the simulated system. A selection of findings is summarized by placing them under the four themes we think to be characteristic of learning with computer simulations (see de Jong, this volume).

Abstract

Pitch, intonation, and speech rate were analyzed in a collection of everyday speech heard by one Dutch infant between the ages of six and nine months. Components of each of these variables were measured in the speech of three adult speakers (mother, father, baby-sitter) when they addressed the infant, and when they addressed another adult. The results are in line with previously reported findings which are usually based on laboratory or prearranged settings: infant-directed speech in a natural setting exhibits more pitch variation, a larger number of simple intonation contours, and slower speech rate than does adult-directed speech.

Abstract

Dutch distinguishes at least four sentence types: statements and questions, the latter type being subdivided into wh-questions (beginning with a question word), yes/no-questions (with inversion of subject and finite), and declarative questions (lexico-syntactically identical to statement). Acoustically, each of these (sub)types was found to have clearly distinct global F0-patterns, as well as a characteristic distribution of final rises [1,2]. The present paper explores the separate contribution of parameters of global downtrend and size of accent-lending pitch movements versus aspects of the terminal rise to the human identification of the four sentence (sub)types, at various positions in the time-course of the utterance. The results show that interrogativity in Dutch can be identified at an early point in the utterance. However, wh-questions are not distinct from statements.

Abstract

Jescheniak and Levelt (Jescheniak, J.-D., Levelt, W.J.M. 1994. Journal of Experimental Psychology: Learning, Memory and Cognition 20 (4), 824–843) have suggested that the speed with which native speakers of a gender-marking language retrieve the grammatical gender of a noun from their mental lexicon may depend on the recency of earlier access to that same noun's gender, as the result of a mechanism that is dedicated to facilitate gender-marked anaphoric reference to recently introduced discourse entities. This hypothesis was tested in two picture naming experiments. Recent gender access did not facilitate the production of gender-marked adjective noun phrases (Experiment 1), nor that of gender-marked definite article noun phrases (Experiment 2), even though naming times for the latter utterances were sensitive to the gender of a written distractor word superimposed on the picture to be named. This last result replicates and extends earlier gender-specific picture-word interference results (Schriefers, H. 1993. Journal of Experimental Psychology: Learning, Memory, and Cognition 19 (4), 841–850), showing that one can selectively tap into the production of grammatical gender agreement during speaking. The findings are relevant to theories of speech production and the representation of grammatical gender for that process.

Abstract

Based on the results of an event-related brain potentials (ERP) experiment (van Berkum, Brown, & Hagoort. 1999a, b), we have recently argued that discourse-level referential context can be taken into account extremely rapidly by the parser. Moreover, our ERP results indicated that local grammatical gender information, although available within a few hundred milliseconds from word onset, is not always used quickly enough to prevent the parser from considering a discourse-supported, but agreement-violating, syntactic analysis. In a comment on our work, Brysbaert and Mitchell (2000) have raised concerns about the methodology of our ERP experiment and have challenged our interpretation of the results. In this reply, we argue that these concerns are unwarranted and, that, in contrast to our own interpretation, the alternative explanations provided by Brysbaert and Mitchell do not account for the full pattern of ERP results.

Abstract

We present the design, implementation and simulation results of a psycholinguistic model of human syntactic processing that meets major empirical criteria. The parser operates in conjunction with a lexicalist grammar and is driven by syntactic information associated with heads of phrases. The dynamics of the model are based on competition by lateral inhibition ('competitive inhibition'). Input words activate lexical frames (i.e. elementary trees anchored to input words) in the mental lexicon, and a network of candidate 'unification links' is set up between frame nodes. These links represent tentative attachments that are graded rather than all-or-none. Candidate links that, due to grammatical or 'treehood' constraints, are incompatible, compete for inclusion in the final syntactic tree by sending each other inhibitory signals that reduce the competitor's attachment strength. The outcome of these local and simultaneous competitions is controlled by dynamic parameters, in particular by the Entry Activation and the Activation Decay rate of syntactic nodes, and by the Strength and Strength Build-up rate of Unification links. In case of a successful parse, a single syntactic tree is returned that covers the whole input string and consists of lexical frames connected by winning Unification links. Simulations are reported of a significant range of psycholinguistic parsing phenomena in both normal and aphasic speakers of English: (i) various effects of linguistic complexity (single versus double, center versus right-hand self-embeddings of relative clauses; the difference between relative clauses with subject and object extraction; the contrast between a complement clause embedded within a relative clause versus a relative clause embedded within a complement clause); (ii) effects of local and global ambiguity, and of word-class and syntactic ambiguity (including recency and length effects); (iii) certain difficulty-of-reanalysis effects (contrasts between local ambiguities that are easy to resolve versus ones that lead to serious garden-path effects); (iv) effects of agrammatism on parsing performance, in particular the performance of various groups of aphasic patients on several sentence types.

Abstract

Recent studies using phoneme detection tasks have shown that spoken-language processing is neither facilitated nor interfered with by optional assimilation, but is inhibited by violation of obligatory assimilation. Interpretation of these results depends on an assessment of their generality, specifically, whether they also obtain when listeners are processing nonnative language. Two separate experiments are presented in which native listeners of German and native listeners of Dutch had to detect a target fricative in legal monosyllabic Dutch nonwords. All of the nonwords were correct realisations in standard Dutch. For German listeners, however, half of the nonwords contained phoneme strings which violate the German fricative assimilation rule. Whereas the Dutch listeners showed no significant effects, German listeners detected the target fricative faster when the German fricative assimilation was violated than when no violation occurred. The results might suggest that violation of assimilation rules does not have to make processing more difficult per se.

Abstract

This study investigates the influence of both phonotactic and acoustic cues on the segmentation of spoken English. Listeners detected embedded English words in nonsense sequences (word spotting). Words aligned with phonotactic boundaries were easier to detect than words without such alignment. Acoustic cues to boundaries could also have signaled word boundaries, especially when word onsets lacked phonotactic alignment. However, only one of several durational boundary cues showed a marginally significant correlation with response times (RTs). The results suggest that word segmentation in English is influenced primarily by phonotactic constraints and only secondarily by acoustic aspects of the speech signal.

Abstract

Previous research has shown that listeners make use of their knowledge of phonotactic constraints to segment speech into individual words. The present study investigates the influence of phonotactics when segmenting a non-native language. German and English listeners detected embedded English words in nonsense sequences. German listeners also had knowledge of English, but English listeners had no knowledge of German. Word onsets were either aligned with a syllable boundary or not, according to the phonotactics of the two languages. Words aligned with either German or English phonotactic boundaries were easier for German listeners to detect than words without such alignment. Responses of English listeners were influenced primarily by English phonotactic alignment. The results suggest that both native and non-native phonotactic constraints influence lexical segmentation of a non-native, but familiar, language.

Abstract

L'A. étudie les corrélations entre structure syntaxique et fonction pragmatique dans les alternances de voix en akatek, une langue maya appartenant au sous-groupe Q'anjob'ala. Les alternances pragmatiques de voix sont les mécanismes par lesquels les langues encodent les différents degrés de topicalité des deux principaux participants d'un événement sémantiquement transitif, l'agent et le patient. A l'aide d'une analyse quantitative, l'A. évalue la topicalité de ces participants et identifie les structures syntaxiques permettant d'exprimer les quatre principales fonctions de voix en akatek : active-directe, inverse, passive et antipassive