Falsifying Serial and Parallel Parsing Models: Empirical Conundrums and An Overlooked Paradigm

When the human parser encounters a local structural ambiguity, are multiple structures pursued (parallel or breadth-first parsing), or just a single preferred structure (serial or depth-first parsing)? This note discusses four important classes of serial and parallel models: simple limited parallel, ranked limited parallel, deterministic serial with reanalysis, and probabilistic serial with reanalysis. It is argued that existing evidence is compatible only with probabilistic serial-reanalysis models, or ranked parallel models augmented with a reanalysis component. A new class of linguistic structures is introduced on which the behavior of serial and parallel parsers diverge the most radically: multiple local ambiguities are stacked to increase the number of viable alternatives in the ambiguous region from two to eight structures. This paradigm may provide the strongest test yet for parallel models.     

Locality and Parsing Complexity

Two experiments, an off-line acceptability judgment study and an on-line self-paced reading experiment, were conducted to tackle the question of locality-based preferences in sentence processing. The material consisted of German verb-final sentences containing a relative clause that was either host adjacent or extraposed. While the off-line data seem to reflect locality based integration cost, the on-line data revealed a highly different picture. It is argued that, in the present constructions at least, locality is relevant for production rather than perception. In perception, heads can be anticipated through incremental integration of their arguments.     

Parsing strategies and discourse context

Two experiments investigated syntactic processing during comprehension of sentences presented either in isolation or in a discourse context. Comprehension of a range of different types of surface structurally ambiguous sentences was studied. To explain the interpretations generally given to the sentences processed in isolation, two parsing principles were proposed: Kimball's (1973) Right Association and Verb Dominance. When the ambiguous sentences were read in context, the interpretations computed were still determined by these structurally based principles, even when this meant that the meanings were at variance with the prior context. These results indicate that surface structure parsing of sentences proceeds in the same way whether sentences are processed alone or in context.This research was supported by a grant from the Australian Research Grants Scheme. I thank Linda Cupples for her help in conducting the experiments and in analyzing the data. I am indebted to Marilyn Ford for her critical comments on an earlier, version of the article. Thanks are also due to Dianne Bradley, Ken Forster, and Roger Wales for their helpful comments on the topic.     

Parsing to Learn

Learning a language by parameter setting is almost certainly less onerous than composing a grammar from scratch. But recent computational modeling of how parameters are set has shown that it is not at all the simple mechanical process sometimes imagined. Sentences must be parsed to discover the properties that select between parameter values. But the sentences that drive learning cannot be parsed with the learner's current grammar. And there is not much point in parsing them with just one new grammar. They must apparently be parsed with all possible grammars, in order to find out which one is most successful at licensing the language. The research task is to reconcile this with the fact that the human sentence parsing mechanism, even in adults, has only very limited parallel parsing capacity. I have proposed that all possible grammars can be folded into one, if parameter values are fragments of sentential tree structures that the parser can make use of where necessary to assign a structure to an input sentence. However, the problem of capacity limitations remains. The combined grammar will afford multiple analyses for some sentences, too many to be computed on-line. I propose that the parser computes only one analysis per sentence but can detect ambiguity, and that the learner makes use of unambiguous input only. This provides secure information but relatively little of it, particularly at early stages of learning where few grammars have been excluded and ambiguity is rife. I consider three solutions: improving the parser's ability to extract unambiguous information from partially ambiguous sentences, assuming default parameter values to temporarily eliminate ambiguity, reconfiguring the parameters so that some are subordinate to others and do not present themselves to the learner until the others have been set. A more radical alternative is to give up the quest for error-free learning and permit parameters to be set without regard for whether the parser may have overlooked an alternative analysis of the sentence. If it can be assumed that the human parser keeps a running tally of the parameter values it has accessed, then the learner would do nothing other than parse sentences for comprehension, as adults do. The most useful parameter values would become more and more easily accessed; the noncontributors would drop out of the running. There would be no learning mechanism at all, over and above the parser. But how accurate this system would be remains to be established.     

Race-Based Parsing and Syntactic Disambiguation

We present a processing model that integrates same important psychological claims about the human sentence-parsing mechanism: namely, that processing is influenced by limitations an working memory and by various syntactic preferences. The model uses time-constraint information to resolve conflicting preferences in a psychologically plausible way. The starting paint far this proposal is the Sausage Machine model (Fodor & Frazier, 1980: Frazier & Fodor, 1978). From there, we attempt to overcome the original model's dependence an ad hoc aspects of its grammar, and its omission of verb-frame preferences. We also add mechanisms far lexical disambiguation and semantic processing in parallel with syntactic processing.     

Parsing the rainbow

Navigating the ontology of color used to be a simple affair. There was the naive view that colors really are in objects the way they appear, and the view that they are secondary qualities to cause certain experiences in us. Today, there are myriad well-developed views but no satisfactory taxonomy of philosophical theories on color. In this article, I first examine the two newest taxonomies on offer and argue that they are inadequate. In particular, I look at Brogaard’s taxonomy and then Cohen’s. One of the reasons I am displeased with Brogaard and Cohen’s taxonomies is that I find it implausible that dispositions are relational properties. I provide an argument against this way of classifying dispositions. Having learned from the vices and virtues of Brogaard and Cohens’ taxonomies, I provide what I believe is a much-enhanced way of taxonomizing philosophical views on color. My taxonomy rules out certain views, clarifies others, and shows that there is an unnoticed view worthy of consideration.     

Parsing mit inkonsistenten Grammatiken

A novel approach to the parsing of natural language based on weighted constraints is presented. Wellformedness conditions are exclusively expressed by means of constraints and no generative rule component is available. In general, constraints can be violated and contradictory requirements of a grammar are tolerated. Additional representational levels allow the integration of different linguistic perspectives into a coherent model, thus contributing jointly to the computation of the most plausible structural interpretation. A number of solution procedures is available for such a parsing problem. Their properties are discussed with respect to possible analogies to human language processing, focussing on robustness against deviating utterances and external temporal pressure.     

Parsing temporarily ambiguous complements

Holmes, Kennedy and Murray (1987) recently claimed that the empirical support for the Minimal Attachment Strategy of sentence parsing had been weakened by results they reported. They found that reading time for an ambiguous string of words did not decrease when it was preceded by an overt complementizer, which should have disambiguated it. Thus, they suggested that results that we (Frazier and Rayner, 1982) earlier attributed to Minimal Attachment were not due to “garden-path” effects, but rather reflected the extra complexity caused by having to process two sets of clausal relations instead of just one. In the present experiment, we replicated their experiment using eye movement data rather than the subject-paced reading task they used. We found that readers processed Nonminimal Attachment sentences with overt complementizers considerably faster than those without a complementizer. Our results showed that the complexity of Nonminimal Attachment sentences cannot be attributed to their clausal status per se. Differences between the tasks that might contribute to the different pattern of results across the experiments are discussed.     

Distinguishing Belief and Imagination

Some philosophers (including Urmson, Humberstone, Shah, and Velleman) hold that believing that p distinctively involves applying a norm according to which the truth of p is a criterion for the success or correctness of the attitude. On this view, imagining and assuming differ from believing in that no such norm is applied. I argue against this view with counterexamples showing that applying the norm of truth is neither necessary nor sufficient for distinguishing believing from imagining and assuming. Then I argue that the different functional properties of these mental states are enough to distinguish them, and that norm‐application doesn't help us draw the functional distinctions.     

Parsing surrounding space into regions

Surrounding space is not inherently organized, but we tend to treat it as though it consisted of regions (e.g., front, back, right, and left). The current studies show that these conceptual regions have characteristics that reflect our typical interactions with space. Three experiments examined the relative sizes and resolutions of front, back, left, and right around oneself. Front, argued to be the most important horizontal region, was found to be (a) largest, (b) recalled with the greatest precision, and (c) described with the greatest degree of detail. Our findings suggest that some of the characteristics of the category model proposed by Huttenlocher, Hedges, and Duncan (1991) regarding memory for pictured circular displays may be generalized to space around oneself. More broadly, our results support and extend thespatial framework analysis of representation of surrounding space (Franklin & Tversky, 1990).     

Distinguishing social and individual psychology

Abstract.— The problem analyzed is how (according to what rules) to disentangle social and individual aspects in, e.g., an act of communication. Taking as point of departure the everyday fact that social relationships as well as individuals may be assessed across variables, an attempt was made to classify variables as either social or individual. It was found, however, that they seem to be distributed along a continuum from individual via relational to social variables; that all variables seem to have both a social and an individual aspect; and that they may be classified along a continuum accordning to how easily they can be made socially relevant or salient. Instead of concluding either that a social psychology proper is impossible, since all variables can be ascribed to the individual(s), or that we are dealing only with social psychology, since all variables have a social potential, a pragmatic solution may be to make clear at each stage of research from which perspective, the social or the individual, one is exploring the variable in focus.     

Parsing as a Cue-Based Retrieval Model

This paper develops a novel psycholinguistic parser and tests it against experimental and corpus reading data. The parser builds on the recent research into memory structures, which argues that memory retrieval is content-addressable and cue-based. It is shown that the theory of cue-based memory systems can be combined with transition-based parsing to produce a parser that, when combined with the cognitive architecture ACT-R, can model reading and predict online behavioral measures (reading times and regressions). The parser's modeling capacities are tested against self-paced reading experimental data (Grodner & Gibson, 2005), eye-tracking experimental data (Staub, 2011), and a self-paced reading corpus (Futrell et al., 2018).     

Parsing silhouettes: The short-cut rule

Many researchers have proposed that, for the purpose of recognition, human vision parses shapes into component parts. Precisely how is not yet known. The minima rule for silhouettes (Hoffman & Richards, 1984) defines boundary points at which to parse but does not tell how to use these points to cut silhouettes and, therefore, does not tell what the parts are. In this paper, we propose the short-cut rule, which states that, other things being equal, human vision prefers to use the shortest possible cuts to parse silhouettes. We motivate this rule, and the well-known Petter’s rule for modal completion, by the principle of transversality. We present five psychophysical experiments that test the short-cut rule, show that it successfully predicts part cuts that connect boundary points given by the minima rule, and show that it can also create new boundary points.     

伦理与道德之辨正

伦理与道德两个概念的内涵及其相互关系在许多研究者那里是比较模糊的。本文试图从词源学的角度并结合黑格尔在《法哲学原理》中的相应观点对二者的内涵加以区别界定并对二者的关系加以厘定:道德是个体对自身完满性的价值追求,具有主体性与私人性;而伦理则是基于社会成员之间的互动交往而形成的规范体系,具有“互主体性”与公共性。道德是伦理的基础,伦理是道德的提升。笔者认为,对伦理与道德加以区分和厘定将有助于伦理冲突的化解、伦理关系的协调以及伦理秩序的维护。