A probabilistic computational model of cross-situational word learning

Words are the essence of communication: They are the building blocks of any language. Learning the meaning of words is thus one of the most important aspects of language acquisition: Children must first learn words before they can combine them into complex utterances. Many theories have been developed to explain the impressive efficiency of young children in acquiring the vocabulary of their language, as well as the developmental patterns observed in the course of lexical acquisition. A major source of disagreement among the different theories is whether children are equipped with special mechanisms and biases for word learning, or their general cognitive abilities are adequate for the task. We present a novel computational model of early word learning to shed light on the mechanisms that might be at work in this process. The model learns word meanings as probabilistic associations between words and semantic elements, using an incremental and probabilistic learning mechanism, and drawing only on general cognitive abilities. The results presented here demonstrate that much about word meanings can be learned from naturally occurring child-directed utterances (paired with meaning representations), without using any special biases or constraints, and without any explicit developmental changes in the underlying learning mechanism. Furthermore, our model provides explanations for the occasionally contradictory child experimental data, and offers predictions for the behavior of young word learners in novel situations.     

Acquired affective associations induce emotion effects in word recognition: An ERP study

The present study examined how contextual learning and in particular emotionality conditioning impacts the neural processing of words, as possible key factors for the acquisition of words’ emotional connotation. 21 participants learned on five consecutive days associations between meaningless pseudowords and unpleasant or neutral pictures using an evaluative conditioning paradigm. Subsequently, event-related potentials were recorded while participants implicitly processed the learned emotional relevance in a lexical decision paradigm. Emotional and neutral words were presented together with the conditioned pseudowords and a set of new pseudowords. Conditioned and new pseudowords differed in the late positive complex. Emotionally and neutrally conditioned stimuli differed in an early time window (80–120 ms) and in the P300. These results replicate ERP effects known from emotion word recognition and indicate that contextual learning and in particular evaluative conditioning is suitable to establish emotional associations in words, and to explain early ERP effects in emotion word recognition.     

语境情绪韵律对新异词汇学习过程的影响

模拟自然学习语境,探讨情绪韵律对具体和抽象词学习的调节作用及其产生原因。被试在高兴、愤怒、中性语境情绪韵律条件下学习抽象和具体词,记录学习过程及测试过程中的行为与脑电反应。结果发现在愤怒韵律背景下学习的抽象词加工的正确率更低、反应时更长;并且诱发了更为显著的脑电成分。表明愤怒韵律对词汇学习尤其是抽象词学习有显著的消极作用。此外,情绪韵律对词汇学习的调节作用是在学习过程中对词汇语义提取及晚期语义整合产生影响而形成的。     

Children extend both words and non‐verbal actions to novel exemplars

Markson and Bloom (1997) found that some learning processes involved in children's acquisition of a new word are also involved in their acquisition of a new fact. They argued that these findings provided evidence against a domain‐specific system for word learning. However, Waxman and Booth (2000) found that whereas children quite readily extend newly learned words to novel exemplars within a category, they do not do this with newly learned facts. They therefore argued that because children did not extend some facts in a principled way, word learning and fact learning may result from different domain‐specific processes. In the current study, we argue that facts are a poor comparison in this argument since facts vary in whether they are tied to particular individuals. A more appropriate comparison is a conventional non‐verbal action on an object –‘what we do with things like this’– since they are routinely generalized categorically to new objects. Our study shows that 2¹/₂‐year‐old children extend novel non‐verbal actions to new objects in the same way that they extend novel words to new objects. The findings provide support for the view that word learning represents a unique configuration of more general learning processes.     

新颖词汇学习中大脑激活模式的fMRI研究

该研究利用功能磁共振脑成像(fMRI)技术,比较在新颖语言学习和母语学习过程中大脑皮层参与的异同。实验设置德语学习和汉语学习两个条件,采用跨通道联结学习范式,要求被试在听到听觉词汇后,按键判断屏幕上所呈现的图片是否与所听词汇匹配。行为数据显示,被试在六段实验后基本掌握了德语语音和语义之间的联结关系;脑成像数据显示,两种学习任务都激活了语言加工的主要脑区,而德语词汇学习过程相比汉语参照条件对应的颞叶激活强度与范围有较大的差异,并且比汉语条件表现出更显著的前额叶区域激活。     

Neurons in Human Temporal Cortex Active with Verbal Associative Learning

In neuronal activity recorded from human middle temporal gyrus during learning of associations between word pairs, a population was identified that had greater activity for associations that were learned rapidly during initial encoding compared to those learned slowly or not at all by an individual subject. This population can be separated from other neurons by the combination of inhibition during word reading when no learning is required and excitation during recent memory for words. These neurons are present in both hemispheres, predominately in deeper layers of cortex. During initial encoding, the increased activity appears at presentation of all word pairs but persists for several seconds only for the rapidly learned pairs, likely reflecting rehersal of items being learned. Human associative learning is related to activity of this specific population of “association” neurons, identified here for the first time.     

Age of acquisition effects in adult lexical processing reflect loss of plasticity in maturing systems: insights from connectionist networks

Early learned words are recognized and produced faster than later learned words. The authors showed that such age of acquisition effects are a natural property of connectionist models trained by back-propagation when patterns are introduced at different points into training and learning of early and late patterns is cumulative and interleaved. Analysis of hidden unit activations indicated that the age of acquisition effect reflects a gradual reduction in network plasticity and a consequent failure to differentiate late items as effectively as early ones. Further simulations examined the effects of vocabulary size, learning rate, sparseness of coding, use of a modified learning algorithm, loss of early items, acquisition of very late items, and lesioning the network. The relationship between age of acquisition and word frequency was explored, including analyses of how the relative influence of these factors is modulated by introducing weight decay.     

Consolidation of novel word learning in native English-speaking adults

Sleep has been shown to improve the retention of newly learned words. However, most methodologies have used artificial or foreign language stimuli, with learning limited to word/novel word or word/image pairs. Such stimuli differ from many word-learning scenarios in which definition strings are learned with novel words. Thus, we examined sleep's benefit on learning new words within a native language by using very low-frequency words. Participants learned 45 low-frequency English words and, at subsequent recall, attempted to recall the words when given the corresponding definitions. Participants either learned in the morning with recall in the evening (wake group), or learned in the evening with recall the following morning (sleep group). Performance change across the delay was significantly better in the sleep than the wake group. Additionally, the Levenshtein distance, a measure of correctness of the typed word compared with the target word, became significantly worse following wake, whereas sleep protected correctness of recall. Polysomnographic data from a subsample of participants suggested that rapid eye movement (REM) sleep may be particularly important for this benefit. These results lend further support for sleep's function on semantic learning even for word/definition pairs within a native language.     

Basic Concepts for a Psychology of Personality

Previous research has indicated a difference between taboo and nontaboo words rated as upsetting in that the former are learned more easily than matched neutral words, while the latter are not. In an effort to determine the source of this differential effect, a four trial study was performed on 64 female college students. The effects of rate and organization of list presentation on the learning of the two types of emotional words were compared to the learning of neutral words. Earlier results demonstrating the significant superiority of taboo but not of upsetting words over matched neutral counterparts were replicated. It was also determined that experimental conditions which result in generally poor performance and low rates of clustering (e.g., early trials, uncategorized list presentation) accentuated the taboo word effect.     

Morphological priming in the German mental lexicon.

We present results from cross-modal priming experiments on German participles and noun plurals. The experiments produced parallel results for both inflectional systems. Regular inflection exhibits full priming whereas irregularly inflected word forms show only partial priming: after hearing regularly inflected words (-t participles and -s plurals), lexical decision times on morphologically related word forms (presented visually) were similar to reaction times for a base-line condition in which prime and target were identical, but significantly shorter than in a control condition where prime and target were unrelated. In contrast, prior presentation of irregular words (-n participles and -er plurals) led to significantly longer response times on morphologically related word forms than the prior presentation of the target itself. Hence, there are clear priming differences between regularly and irregularly inflected German words. We compare the findings on German with experimental results on regular and irregular inflection in English and Italian, and discuss theoretical implications for single versus dual-mechanism models of inflection.     

汉语母语者英语动词三种屈折变化形式加工机制的差异

采用ERP技术考察了汉语母语者,英语动词三种屈折变化形式加工机制的差异。结果发现,英语动词进行式的屈折变化形式- ing的错误使用诱发了P600成分;完成式的屈折变化形式- ed的错误使用诱发了N400成分;完成式中没有明显屈折变化标识的不规则词的错误使用没有诱发典型的ERP成分。研究结果表明：对于缺乏英语使用环境、母语为汉语的晚期英语学习者来说,屈折变化形式- ing,完成式规则动词的变化形式- ed和没有明显形态变化标识的不规则动词的变化形式具有不同的加工机制,前者可能处于由陈述性记忆系统向程序性记忆系统加工的转变过程中,而后两者更多地由陈述性系统进行加工。这种加工机制的差异可能与动词屈折变化规则的复杂程度有关。     

Gavagai Is as Gavagai Does: Learning Nouns and Verbs From Cross‐Situational Statistics

Learning to map words onto their referents is difficult, because there are multiple possibilities for forming these mappings. Cross‐situational learning studies have shown that word‐object mappings can be learned across multiple situations, as can verbs when presented in a syntactic context. However, these previous studies have presented either nouns or verbs in ambiguous contexts and thus bypass much of the complexity of multiple grammatical categories in speech. We show that noun word learning in adults is robust when objects are moving, and that verbs can also be learned from similar scenes without additional syntactic information. Furthermore, we show that both nouns and verbs can be acquired simultaneously, thus resolving category‐level as well as individual word‐level ambiguity. However, nouns were learned more quickly than verbs, and we discuss this in light of previous studies investigating the noun advantage in word learning.     

INCIDENTAL LANGUAGE LEARNING:

Abstract —Two experiments investigated the performance of first-grade children and adults on an incidental language-learning task. Learning entailed word segmentation from continuous speech, an initial and crucial component of language acquisition. Subjects were briefly exposed to an unsegmented artificial language, presented auditorily, in which the only cues to word boundaries were the transitional probabilities between syllables. Subjects were not told that they were listening to a language, or even to listen at all; rather, they were engaged in a cover task of creating computer illustrations. Both adults and children learned the words of the language. Moreover, the children performed as well as the adults. These data suggest that a statistical learning mechanism (transitional probability computation) is able to operate incidentally and, surprisingly, as well in children as in adults.     

Phonological Short-term Memory and the Learning of Novel Words: The Effect of Phonological Similarity and Item Length

The investigation of a patient with a selective impairment of phonological short-term memory has recently provided evidence that this system may be involved in long-term learning of novel words, for which a pre-existing semantic representation is not available (Baddeley, Papagno, & Vallar, 1988). The present series of experiments in normal subjects explored this hypothesis. We assessed the effects of phonological similarity and item length, which reflect the operation of the phonological short-term store and the rehearsal component of verbal memory, upon paired associate long-term learning of auditorily presented words and non-words. Phonological similarity affected the learning of novel words more than known words (Experiment 1); when a delay was interposed between presentation and recall, the disruptive effect was confined to novel words (Experiment 2). Also word length disrupted the learning of novel words, but not of known words (Experiment 3). These results tie in with neuropsychological evidence to suggest a role for phonological short-term memory in the learning of new words, and they have developmental implications for the study of language acquisition.     

Enhancing Spelling Performance in Students with Learning Disabilities

The present study, using a single subject adapted alternating treatments design, compared the effects of two spelling strategies (cover, copy, compare, and copy-only) used to enhance spelling performance in 2, third-grade students with learning disabilities. The cover, copy compare (CCC) method required the students to say the word, point to the word, repeat the word, cover the word, print the word, compare the word to the correct model, and correct errors if necessary. The copy-only method required the students to say the word, point to the word, repeat the word, and print the word. Overall, the CCC strategy was more effective in words learned and words retained. Results are discussed in light of CCC requiring a self-evaluation/self-correction component not required in the more traditional method of instruction. By incorporating a simple, self-management component to spelling instruction, teachers may improve spelling performance in the classroom.     

The role of partial knowledge in statistical word learning

A critical question about the nature of human learning is whether it is an all-or-none or a gradual, accumulative process. Associative and statistical theories of word learning rely critically on the later assumption: that the process of learning a word’s meaning unfolds over time. That is, learning the correct referent for a word involves the accumulation of partial knowledge across multiple instances. Some theories also make an even stronger claim: Partial knowledge of one word–object mapping can speed up the acquisition of other word–object mappings. We present three experiments that test and verify these claims by exposing learners to two consecutive blocks of cross-situational learning, in which half of the words and objects in the second block were those that participants failed to learn in Block 1. In line with an accumulative account, Re-exposure to these mis-mapped items accelerated the acquisition of both previously experienced mappings and wholly new word–object mappings. But how does partial knowledge of some words speed the acquisition of others? We consider two hypotheses. First, partial knowledge of a word could reduce the amount of information required for it to reach threshold, and the supra-threshold mapping could subsequently aid in the acquisition of new mappings. Alternatively, partial knowledge of a word’s meaning could be useful for disambiguating the meanings of other words even before the threshold of learning is reached. We construct and compare computational models embodying each of these hypotheses and show that the latter provides a better explanation of the empirical data.     

汉字频率和构词数对非汉字圈学生汉字学习的影响

探讨汉字频率和构词数对非汉字文化圈的留学生汉字学习效果的影响。实验要求被试对已学过的不同频率和构词数的汉字写出拼音并组词,结果显示,汉字频率对汉字学习效果有影响,而且频率效应的大小受笔画数的制约;实验未发现构词数对汉字学习效果的影响。研究结果表明,语言输入的频率是影响汉语习得的一个重要因素     

Human Learning on a Four-Section,Elevated, Finger Maze

Forty Ss learned 10 sentences composed of adjective, noun, verb, and adverb and were subsequently tested for their recall of the sentences and their ability to generate new sentences based on an association rule for words within the sentences. The rule could be discovered from the sentences learned and was comparable to grammatical rules for sentence structure. Subjects also rated the meaning of the words from the sentences before and after learning. Eight Ss were in each of five experimental conditions, which differed in terms of the degree to which the words in the 10 sentences were in a natural language order. The five orders were natural, reversed, 20% random, 50% random, and 100% random orders.

The results showed that the closer the sentence order was to natural language order, the more Ss recalled the sentences they learned and the more accurately they generated unlearned sentences, apparently as a result of discovering and using the association rule. Another finding was that the rated evaluative meaning of words changed in a predictable direction, toward the mean rating of the words associated with each word. Such meaning conditioning appears to be an automatic process comparable to classical conditioning in that it is unaffected by the order of words within sentences and occurs for different word forms.     

The acquisition of allophonic rules: statistical learning with linguistic constraints

Phonological rules relate surface phonetic word forms to abstract underlying forms that are stored in the lexicon. Infants must thus acquire these rules in order to infer the abstract representation of words. We implement a statistical learning algorithm for the acquisition of one type of rule, namely allophony, which introduces context-sensitive phonetic variants of phonemes. This algorithm is based on the observation that different realizations of a single phoneme typically do not appear in the same contexts (ideally, they have complementary distributions). In particular, it measures the discrepancies in context probabilities for each pair of phonetic segments. In Experiment 1, we test the algorithm's performances on a pseudo-language and show that it is robust to statistical noise due to sampling and coding errors, and to non-systematic rule application. In Experiment 2, we show that a natural corpus of semiphonetically transcribed child-directed speech in French presents a very large number of near-complementary distributions that do not correspond to existing allophonic rules. These spurious allophonic rules can be eliminated by a linguistically motivated filtering mechanism based on a phonetic representation of segments. We discuss the role of a priori linguistic knowledge in the statistical learning of phonology.     

A computer model of inflection learning

This paper presents a computer program that creates and modifies inflectional paradigms. A paradigm is a linguistic device used to index lexical entries that share a common inflection along a given dimension such as tense or case. Paradigms are useful because they offer a convenient method for predicting the form of a novel word and for explaining some aspects of language development. The model presented here is based on Pinker’s (1984) inflection learning theory. The model covers a subset of the rules for learning inflectional paradigms that are outlined in the theory. The primary concern of the model is to understand how elements that are crucial to the theory, such as lexical entries and feature equations, may interact to form and expand word-specific paradigms. The model also represents an attempt to assess implementation details not specified by the theory. A brief overview of the model is given, followed by a sample run. Finally, some practical and theoretical considerations, such as the structure of paradigms, paradigm assignment to novel words, and suppletion, are discussed.