Hemispheric sensitivity to grammatical cues: evidence for bilateral processing of number agreement in noun phrases

The present experiment employed a grammatical priming task to explore the possible contributions of the left and right cerebral hemispheres to the processing of grammatical agreement. Stimuli were three-word noun phrases, with the prime centered above the fixation point and the target presented laterally to one visual field after a 600-ms stimulus onset asynchrony. Number agreement between primes and targets was varied such that the article of the prime could be consistent (i.e., each narrow shoe or all narrow shoes), inconsistent (i.e., all narrow shoe or each narrow shoes) or neutral (i.e., the narrow shoe(s)) with respect to the inflection of the target. Half of the subjects provided lexical decision responses and the other half pronunciation. The bilateral priming effect, obtained only in lexical decision, suggests that both the left and the right hemispheres are sensitive to certain grammatical cues. In addition to the task difference in priming, the inclusion of a neutral condition and of pseudo-inflected nonwords allowed these effects to be attributed to postlexical mechanisms.     

Making syntax of sense: number agreement in sentence production

Grammatical agreement flags the parts of sentences that belong together regardless of whether the parts appear together. In English, the major agreement controller is the sentence subject, the major agreement targets are verbs and pronouns, and the major agreement category is number. The authors expand an account of number agreement whose tenets are that pronouns acquire number lexically, whereas verbs acquire it syntactically but with similar contributions from number meaning and from the number morphology of agreement controllers. These tenets were instantiated in a model using existing verb agreement data. The model was then fit to a new, more extensive set of verb data and tested with a parallel set of pronoun data. The theory was supported by the model's outcomes. The results have implications for the integration of words and structures, for the workings of agreement categories, and for the nature of the transition from thought to language.     

Agrammatic production of subject-verb agreement: the effect of conceptual number.

Three experiments tested the hypothesis that the deficit underlying agrammatic sentence production difficulties can be characterized as a limitation of computational resources and that these resources are not restricted to syntactic processing. This hypothesis was tested by eliciting subject-verb agreement errors in a sentence fragment completion paradigm. Sentence fragments were complex noun phrases, containing a subject (head) noun and a modifying prepositional phrase, containing a "local" noun. We varied the number of "tokens" a singular head noun referred to. Therefore, in one condition, grammatical and conceptual number of the head noun mismatched, whereas these numbers were the same in another condition. In Experiments 1 and 2, we observed an effect of this variable (i.e., more agreement errors when conceptual number was plural and grammatical number singular) in normal controls. Broca's aphasics, on the other hand, showed no effect. Experiment 3 consisted of a sentence/picture matching test. This test showed that the lack of effect with Broca's aphasics cannot be attributed to a comprehension deficit. We argue that these results are incompatible with the notion of a limitation in resources specific for syntactic processing. Instead, we interpret this as the result of a trade-off: Broca's aphasics lack computational resources to take into account both grammatical and conceptual information in morphosyntactic processing and rely on grammatical information only.     

Facilitatory and inhibitory effects of grammatical agreement: evidence from readers' eye fixation patterns

The study examined how grammatical agreement affects reading in Finnish. Readers' eye fixation patterns were recorded when they read one of three alternative versions of the same sentences, where the critical difference was the type of preceding word of the target nouns. The preceding word was (a) an agreeing modifier (mainioksi orkesteriksi='for an excellent orchestra'), (b) a non-agreeing modifier that was grammatical, unambiguous and synonymous to the agreeing modifier (kelpo orkesteriksi='for an excellent orchestra'), or (c) a baseline condition without a modifier (orkesteriksi='for an orchestra'). Two different types of agreement were used, a modifier-head agreement and a possessive agreement. The results showed that the agreeing modifiers facilitate and the non-agreeing modifiers inhibit the reading of the target nouns compared to the neutral baseline condition. These effects appeared in the second-pass reading. The pattern was similar between the two agreement structures.     

Statistical language learning: mechanisms and constraints

What types of mechanisms underlie the acquisition of human language? Recent evidence suggests that learners, including infants, can use statistical properties of linguistic input to discover structure, including sound patterns, words, and the beginnings of grammar. These abilities appear to be both powerful and constrained, such that some statistical patterns are more readily detected and used than others. Implications for the structure of human languages are discussed.     

Domain-specific learning of grammatical structure in musical and phonological sequences

Artificial grammar learning depends on acquisition of abstract structural representations rather than domain-specific representational constraints, or so many studies tell us. Using an artificial grammar task, we compared learning performance in two stimulus domains in which respondents have differing tacit prior knowledge. We found that despite grammatically identical sequence structures, learning was better for harmonically related chord sequences than for letter name sequences or harmonically unrelated chord sequences. We also found transfer effects within the musical and letter name tasks, but not across the domains. We conclude that knowledge acquired in implicit learning depends not only on abstract features of structured stimuli, but that the learning of regularities is in some respects domain-specific and strongly linked to particular features of the stimulus domain.     

One or more: psycholinguistic evidence for divergence of numerosity and grammatical number assignment.

Naturalistic slips of the tongue in Finnish were studied in order to reveal the production of grammatical number. It appears that speakers may consider plural forms referring to single events as singulars and collective nouns as plurals. These result in violations in number agreement. It was found that errors in numerosity are both structural and semantic in nature and that one type can be traced to the level of morphological encoding and the other type to message formulation.     

Relationships between language structure and language learning: the suffixing preference and grammatical categorization

It is a reasonable assumption that universal properties of natural languages are not accidental. They occur either because they are underwritten by genetic code, because they assist in language processing or language learning, or due to some combination of the two. In this paper we investigate one such language universal: the suffixing preference across the world's languages, whereby inflections tend to be added to the end of words. A corpus analysis of child-directed speech in English found that suffixes were more accurate at cuing the grammatical category of the root word than were prefixes. An artificial language experiment found that there was a learning advantage for suffixes over prefixes in terms of grammatical categorization within an artificial language. The results are consistent with an account of language universals that originate in general purpose learning mechanisms.     

Native language effects in learning second-language grammatical gender: A training study

We investigated cross-language influences in the representation and acquisition of Dutch word gender by native speakers of German. Participants named pictures in Dutch, using gender-marked noun phrases, and were trained on this task using feedback. Nouns differed in gender compatibility and cognate status with respect to German. The results show clear effects of cross-language gender compatibility and cognate status on response accuracy, certainty, and consistency. Feedback during training reduced gender errors approximately by half, and affected the different item conditions similarly. Furthermore, relative to the initial error rates, incorrect gender responses given with great certainty were not harder to modify than those with lower certainty. The results provide insights into the nature and stability of correct and incorrect gender representations in L2, and demonstrate the pervasiveness of transfer from the first to the second language even after intensive training.     

Type and number of violations and the grammatical congruency effect in lexical decision

Summary An experiment was conducted in the Serbo-Croatian language in which native speakers/readers made lexical decisions on inflected nouns and legally inflected pseudonouns following inflected possessive pronouns. A possessive pronoun and the noun or pseudonoun that followed it could agree in case, gender, and number (0 violations), disagree in either case or gender or number (1 violation) or disagree simultaneously on two of the three (2 violations). A grammatical congruency effect was observed for both nouns and pseudonouns. Acceptance latencies were shorter and rejection latencies were longer for inflectional agreement than inflectional disagreement. However, for neither nouns nor pseudonouns was the magnitude of the effect influenced by the type or number of violations. The results are discussed in terms of (1) the automaticity of syntactic processes and (2) the properties of a decision making device (specially tailored to rapid lexical evaluations) relative to the properties of the language processor.This research was supported in part by NICHD Grants HD-08495 and HD-01994 to the University of Belgrade and Haskins Laboratories, respectively     

Role of grammatical gender and semantics in German word production

Semantic substitution errors (e.g., saying "arm" when "leg" is intended) are among the most common types of errors occurring during spontaneous speech. It has been shown that grammatical gender of German target nouns is preserved in the errors (E. Mane, 1999). In 3 experiments, the authors explored different accounts of the grammatical gender preservation effect in German. In all experiments, semantic substitution errors were induced using a continuous naming paradigm. In Experiment 1, it was found that gender preservation disappeared when speakers produced bare nouns. Gender preservation was found when speakers produced phrases with determiners marked for gender (Experiment 2) but not when the produced determiners were not marked for gender (Experiment 3). These results are discussed in the context of models of lexical retrieval during production.     

Implicit learning shapes new conscious percepts and representations

We present here the lineaments of a new account of implicit learning, an account that does not rely on the notion of “implicit knowledge.” In this account, improved performance depends on the action of unconscious mechanisms that structure the phenomenal, conscious experience of the world. This integrative view makes groundless the search for dissociations between conscious and unconscious influences that has been at the core of the research on implicit learning and memory. We contrast this view, on the one hand, to Dienes and Berry’s (1997) proposal, which defines implicit learning by analogy with subliminal perception, and, on the other, to Neal and Hesketh’s (1997) episodic account, in which subjective experience is a starting point for inquiry, rather than the phenomenon requiring explanation.     

Lexical availability and grammatical encoding scope during spoken sentence production

Three sentence production experiments investigate the relationship between lexical and structural processing scope. Speakers generated sentences with varying phrase structures in response to visual displays (e.g., The dog and the hat move above the fork and the tree/The dog moves above the hat and the fork and the tree). On half of the trials, one of the pictures in the arrays was previewed. Filler sentences varied preview position and sentence structure from trial to trial. When speakers could not anticipate the position of the previewed picture in the upcoming sentence (Experiment 1), preview benefit for pictures corresponding to the second noun to be produced was limited to pictures that fell within the sentence-initial phrase. When the linear position of the previewed picture was predictable, preview benefits were observed for the second noun to be produced, irrespective of phrase position (Experiment 2). However, no preview benefits were observed for the third noun to be produced (Experiment 3). In contrast, significant effects of initial phrase structure were observed in all experiments, with latencies increasing with initial phrase length. The results are consistent with speakers operating a phrasal scope for structural planning within which the scope of lexical access can vary.     

Production of grammatical number in specific language impairment: An elicitation experiment on Finnish.

A wug type of plural production task with Finnish-speaking 10- to 11-year-old normals and 10-year-old SLI children shows that formal transparency of affixation and the lexicality (real words vs pseudowords) of the stimuli affect the performance of both groups. Not unexpectedly, nontransparent pseudowords receive drastically low correctness rates among the SLI speakers. The failures of the SLI group corroborate the claim that they have not internalized the productive inflectional morphology involved in plural formation to the extent that their normal, age-matched peers have.     

Planning scope in spoken sentence production: the role of grammatical units

Four experiments investigate the scope of grammatical planning during spoken sentence production in Japanese and English. Experiment 1 shows that sentence latencies vary with length of sentence-initial subject phrase. Exploiting the head-final property of Japanese, Experiments 2 and 3 extend this result by showing that in a 2-phrase subject phrase, sentence latency varies with the length of the sentence-initial phrase rather than that of the whole subject phrase or its head phrase. Experiment 4 confirms this finding in English. The authors' interpretation suggests that these effects derive from grammatical encoding processes. Planning scope varies according to the relation between the 2 phrases composing the subject phrase. A thematically defined functional phrase is suggested as defining this scope.     

Shared learning shapes human performance: Transfer effects in task sharing

We investigated whether performing a task with a co-actor shapes the way a subsequent task is performed. In four experiments participants were administered a Simon task after practicing a spatial compatibility task with an incompatible S-R mapping. In Experiment 1 they performed both tasks alongside another person; in Experiment 2 they performed the spatial compatibility task alone, responding to only one stimulus position, and the Simon task with another person; in Experiment 3, they performed the spatial compatibility task with another person and the Simon task alone; finally, in Experiment 4, they performed the spatial compatibility task alone and the Simon task with another person. The incompatible practice eliminated the Simon effect in Experiments 1 and 4. These results indicate that when a task is distributed between two participants with each one performing a different part of it, they tend to represent the whole task rather than their own part of it. This experience can influence the way a subsequent task is performed, as long as this latter occurs in a social context.     

Statistical inference: learning in artificial neural networks

Artificial neural networks (ANNs) are widely used to model low-level neural activities and high-level cognitive functions. In this article, we review the applications of statistical inference for learning in ANNs. Statistical inference provides an objective way to derive learning algorithms both for training and for evaluation of the performance of trained ANNs. Solutions to the over-fitting problem by model-selection methods, based on either conventional statistical approaches or on a Bayesian approach, are discussed. The use of supervised and unsupervised learning algorithms for ANNs are reviewed. Training a multilayer ANN by supervised learning is equivalent to nonlinear regression. The ensemble methods, bagging and arching, described here, can be applied to combine ANNs to form a new predictor with improved performance. Unsupervised learning algorithms that are derived either by the Hebbian law for bottom-up self-organization, or by global objective functions for top-down self-organization are also discussed.