Do Humans Really Learn AnBn Artificial Grammars From Exemplars?

An important topic in the evolution of language is the kinds of grammars that can be computed by humans and other animals. Fitch and Hauser (F&H; 2004) approached this question by assessing the ability of different species to learn 2 grammars, (AB)(n) and A(n) B(n) . A(n) B(n) was taken to indicate a phrase structure grammar, eliciting a center-embedded pattern. (AB)(n) indicates a grammar whose strings entail only local relations between the categories of constituents. F&H's data suggest that humans, but not tamarin monkeys, learn an A(n) B(n) grammar, whereas both learn a simpler (AB)(n) grammar (Fitch & Hauser, 2004). In their experiments, the A constituents were syllables pronounced by a female voice, whereas the B constituents were syllables pronounced by a male voice. This study proposes that what characterizes the A(n) B(n) exemplars is the distributional regularities of the syllables pronounced by either a male or a female rather than the underlying, more abstract patterns. This article replicates F&H's data and reports new controls using either categories similar to those in F&H or less salient ones. This article shows that distributional regularities explain the data better than grammar learning. Indeed, when familiarized with A(n) B(n) exemplars, participants failed to discriminate A(3) B(2) and A(2) B(3) from A(n) B(n) items, missing the crucial feature that the number of As must equal the number of Bs. Therefore, contrary to F&H, this study concludes that no syntactic rules implementing embedded nonadjacent dependencies were learned in these experiments. The difference between human linguistic abilities and the putative precursors in monkeys deserves further exploration.     

Intentional control based on familiarity in artificial grammar learning

It is commonly held that implicit learning is based largely on familiarity. It is also commonly held that familiarity is not affected by intentions. It follows that people should not be able to use familiarity to distinguish strings from two different implicitly learned grammars. In two experiments, subjects were trained on two grammars and then asked to endorse strings from only one of the grammars. Subjects also rated how familiar each string felt and reported whether or not they used familiarity to make their grammaticality judgment. We found subjects could endorse the strings of just one grammar and ignore the strings from the other. Importantly, when subjects said they were using familiarity, the rated familiarity for test strings consistent with their chosen grammar was greater than that for strings from the other grammar. Familiarity, subjectively defined, is sensitive to intentions and can play a key role in strategic control.     

Adjoints and emergence: applications of a new theory of adjoint functors

Since its formal definition over sixty years ago, category theory has been increasingly recognized as having a foundational role in mathematics. It provides the conceptual lens to isolate and characterize the structures with importance and universality in mathematics. The notion of an adjunction (a pair of adjoint functors) has moved to center-stage as the principal lens. The central feature of an adjunction is what might be called “determination through universals” based on universal mapping properties. A recently developed “heteromorphic” theory about adjoints suggests a conceptual structure, albeit abstract and atemporal, for how new relatively autonomous behavior can emerge within a system obeying certain laws. The focus here is on applications in the life sciences (e.g., selectionist mechanisms) and human sciences (e.g., the generative grammar view of language).

人工语法学习中意识加工的渐进发展

内隐学习研究致力于探讨学习活动的意识加工程度, 当前相关的研究逻辑从传统的意识二分观转向渐进意识假设。本研究通过对人工语法范式进行改造, 并借鉴加工分离程序的有关思想, 在学习阶段引入双重测量任务, 并根据双重任务成绩计算分离出相应学习时段的意识与无意识成分的贡献分数, 从而考察在内隐学习过程中两种加工成分的动态变化特征。研究结果支持渐进意识假设, 在学习进程中, 无意识成分和意识成分都呈现出渐进发展的趋势; 而随学习的深入, 两种成分呈现出不同的变化模式, 中后期意识加工快速增长, 无意识加工则保持平缓发展, 渐进意识系统整体呈现向外显学习推进的发展态势。     

The P600 in Implicit Artificial Grammar Learning

The suitability of the artificial grammar learning (AGL) paradigm to capture relevant aspects of the acquisition of linguistic structures has been empirically tested in a number of EEG studies. Some have shown a syntax‐related P600 component, but it has not been ruled out that the AGL P600 effect is a response to surface features (e.g., subsequence familiarity) rather than the underlying syntax structure. Therefore, in this study, we controlled for the surface characteristics of the test sequences (associative chunk strength) and recorded the EEG before (baseline preference classification) and after (preference and grammaticality classification) exposure to a grammar. After exposure, a typical, centroparietal P600 effect was elicited by grammatical violations and not by unfamiliar subsequences, suggesting that the AGL P600 effect signals a response to structural irregularities. Moreover, preference and grammaticality classification showed a qualitatively similar ERP profile, strengthening the idea that the implicit structural mere‐exposure paradigm in combination with preference classification is a suitable alternative to the traditional grammaticality classification test.     

Understanding how input matters: verb learning and the footprint of universal grammar

Studies under the heading "syntactic bootstrapping" have demonstrated that syntax guides young children's interpretations during verb learning. We evaluate two hypotheses concerning the origins of syntactic bootstrapping effects. The "universalist" view, holding that syntactic bootstrapping falls out from universal properties of the syntax-semantics mapping, is shown to be superior to the "emergentist" view, which holds that argument structure patterns emerge from a process of categorization and generalization over the input. These theories diverge in their predictions about a language in which syntactic structure is not the most reliable cue to a certain meaning. In Kannada, causative morphology is a better predictor of causative meaning than transitivity is. Hence, the emergentist view predicts that Kannada-speaking children will associate causative morphology (in favor of transitive syntax) with causative meaning. The universalist theory, however, predicts the opposite pattern. Using an act-out task, we found that 3-year-old native speakers of Kannada associate argument number and not morphological form with causativity, supporting the universalist approach.     

Idiomatic syntactic constructions and language learning

This article explores the influence of idiomatic syntactic constructions (i.e., constructions whose phrase structure rules violate the rules that underlie the construction of other kinds of sentences in the language) on the acquisition of phrase structure. In Experiment 1, participants were trained on an artificial language generated from hierarchical phrase structure rules. Some participants were given exposure to an idiomatic construction (IC) during training, whereas others were not. Under some circumstances, the presence of an idiomatic construction in the input aided learners in acquiring the phrase structure of the language. Experiment 2 provides a replication of the first experiment and extends the findings by showing that idiomatic constructions that strongly violate the predictive dependencies that define the phrase structure of the language do not aid learners in acquiring the structure of the language. Together, our data suggest that (a) idiomatic constructions aid learners in acquiring the phrase structure of a language by highlighting relevant structural elements in the language, and (b) such constructions are useful cues to learning to the extent that learners can keep their knowledge of the idiomatic construction separate from their knowledge of the rest of the language.     

The seeds of spatial grammar in the manual modality

Sign languages modulate the production of signs in space and use this spatial modulation to refer back to entities—to maintain coreference. We ask here whether spatial modulation is so fundamental to language in the manual modality that it will be invented by individuals asked to create gestures on the spot. English speakers were asked to describe vignettes under 2 conditions: using gesture without speech, and using speech with spontaneous gestures. When using gesture alone, adults placed gestures for particular entities in non-neutral locations and then used those locations to refer back to the entities. When using gesture plus speech, adults also produced gestures in non-neutral locations but used the locations coreferentially far less often. When gesture is forced to take on the full burden of communication, it exploits space for coreference. Coreference thus appears to be a resilient property of language, likely to emerge in communication systems no matter how simple.     

The Declarative/Procedural Model of Lexicon and Grammar

Our use of language depends upon two capacities: a mental lexicon of memorized words and a mental grammar of rules that underlie the sequential and hierarchical composition of lexical forms into predictably structured larger words, phrases, and sentences. The declarative/procedural model posits that the lexicon/grammar distinction in language is tied to the distinction between two well-studied brain memory systems. On this view, the memorization and use of at least simple words (those with noncompositional, that is, arbitrary form-meaning pairings) depends upon an associative memory of distributed representations that is subserved by temporal-lobe circuits previously implicated in the learning and use of fact and event knowledge. This declarative memory system appears to be specialized for learning arbitrarily related information (i.e., for associative binding). In contrast, the acquisition and use of grammatical rules that underlie symbol manipulation is subserved by frontal/basal-ganglia circuits previously implicated in the implicit (nonconscious) learning and expression of motor and cognitive skills and habits (e.g., from simple motor acts to skilled game playing). This procedural system may be specialized for computing sequences. This novel view of lexicon and grammar offers an alternative to the two main competing theoretical frameworks. It shares the perspective of traditional dual-mechanism theories in positing that the mental lexicon and a symbol-manipulating mental grammar are subserved by distinct computational components that may be linked to distinct brain structures. However, it diverges from these theories where they assume components dedicated to each of the two language capacities (that is, domain-specific) and in their common assumption that lexical memory is a rote list of items. Conversely, while it shares with single-mechanism theories the perspective that the two capacities are subserved by domain-independent computational mechanisms, it diverges from them where they link both capacities to a single associative memory system with broad anatomic distribution. The declarative/procedural model, but neither traditional dual- nor single-mechanism models, predicts double dissociations between lexicon and grammar, with associations among associative memory properties, memorized words and facts, and temporal-lobe structures, and among symbol-manipulation properties, grammatical rule products, motor skills, and frontal/basal-ganglia structures. In order to contrast lexicon and grammar while holding other factors constant, we have focused our investigations of the declarative/procedural model on morphologically complex word forms. Morphological transformations that are (largely) unproductive (e.g., in go—went, solemn—solemnity) are hypothesized to depend upon declarative memory. These have been contrasted with morphological transformations that are fully productive (e.g., in walk—walked, happy—happiness), whose computation is posited to be solely dependent upon grammatical rules subserved by the procedural system. Here evidence is presented from studies that use a range of psycholinguistic and neurolinguistic approaches with children and adults. It is argued that converging evidence from these studies supports the declarative/procedural model of lexicon and grammar.     

Learning a Generative Probabilistic Grammar of Experience: A Process‐Level Model of Language Acquisition

We introduce a set of biologically and computationally motivated design choices for modeling the learning of language, or of other types of sequential, hierarchically structured experience and behavior, and describe an implemented system that conforms to these choices and is capable of unsupervised learning from raw natural‐language corpora. Given a stream of linguistic input, our model incrementally learns a grammar that captures its statistical patterns, which can then be used to parse or generate new data. The grammar constructed in this manner takes the form of a directed weighted graph, whose nodes are recursively (hierarchically) defined patterns over the elements of the input stream. We evaluated the model in seventeen experiments, grouped into five studies, which examined, respectively, (a) the generative ability of grammar learned from a corpus of natural language, (b) the characteristics of the learned representation, (c) sequence segmentation and chunking, (d) artificial grammar learning, and (e) certain types of structure dependence. The model's performance largely vindicates our design choices, suggesting that progress in modeling language acquisition can be made on a broad front—ranging from issues of generativity to the replication of human experimental findings—by bringing biological and computational considerations, as well as lessons from prior efforts, to bear on the modeling approach.