Conceptual graph grammar--a simple formalism for sublanguage

There are a wide variety of computer applications that deal with various aspects of medical language: concept representation, controlled vocabulary, natural language processing, and information retrieval. While technical and theoretical methods appear to differ, all approaches investigate different aspects of the same phenomenon: medical sublanguage. This paper surveys the properties of medical sublanguage from a formal perspective, based on detailed analyses cited in the literature. A review of several computer systems based on sublanguage approaches shows some of the difficulties in addressing the interaction between the syntactic and semantic aspects of sublanguage. A formalism called Conceptual Graph Grammar is presented that attempts to combine both syntax and semantics into a single notation by extending standard Conceptual Graph notation. Examples from the domain of pathology diagnoses are provided to illustrate the use of this formalism in medical language analysis. The strengths and weaknesses of the approach are then considered. Conceptual Graph Grammar is an attempt to synthesize the common properties of different approaches to sublanguage into a single formalism, and to begin to define a common foundation for language-related research in medical informatics.     

Understanding clinical narrative text

A linguistic analysis of the narrative which appears in patient documents, in the domain of urology, has shown that the information is conveyed by a small number of semantic statement types composed of syntactic combinations of medical and English word classes. It has been shown that words from different classes occurring in the diagnosis for the patient disease follow some regular grammatical rules. This impressive linguistic regularity in clinical narrative has made it encouraging to develop a recognizer which is capable of recognizing disease diagnosis in the narrative clinical documents in the domain of urology. The recognizer was implemented using the C language. This recognizer output information which could be refined to a form suitable for management by a database management system for further computerized information processing.     

Medical dictionaries for patient encoding systems: a methodology

Medical language is highly compositional and makes extensive use of common roots, especially Latino-Greek roots. Besides words devoted to common sense, medical language presents some typical characteristics, especially on morphological and semantic aspects of word formation. Morphological decomposition and identification precedes semantic analysis. It is only when these two prerequisites are fulfilled that an attempt to grasp the meaning of a whole expression is made possible. The main aim of the proposed approach is that of coping with 'the lack of coverage of the medical lexical knowledge', in order to help physicians find the correct international classification for diseases (ICD) codes for a written diagnosis. The proposed methodology allows the development of a powerful dynamic dictionary dedicated to natural language processing in the field of diagnoses and narrative procedures. It describes the design of an analyser that can profit from a dictionary. The methods used have proved to be efficient for various classifications, s well as for multiple languages, as the system presently supports French, German, English and Dutch for ICD-9 and ICD-10 classifications.     

The nature of lexical knowledge

This paper considers the nature of lexical knowledge and its role in language and information processing. The lexicon is the central component of language and plays a pivotal role in current linguistic theory [3, 4] and, increasingly, in natural language processing systems [5-7]. The lexicon embodies information about the lexical items of the language and serves as the foundation for morphologic, syntactic, and semantic processing. The differences as well as commonalities among dictionaries, thesauri, and lexicons are discussed, and distinctions between words, lexical items, and terms are drawn. Next, the scope and content of the SPECIALIST lexicon are presented, followed by a discussion of certain writing conventions that can be troublesome for text processing applications. One approach to handling orthographic and other lexical variation is discussed in a section that reports on the design and implementation of the SPECIALIST lexical programs. The paper concludes with a discussion of controlled terminologies for the medical domain. Throughout the discussion, examples are drawn from the SPECIALIST lexicon and from the other UMLS knowledge sources [8, 9].     

A reason to rhyme: Phonological and semantic influences on lexical access.

During on-line language production, speakers rapidly select a sequence of words to express their desired meaning. The current study examines whether this lexical selection is also dependent on the existing activation of surface properties of the words. Such surface properties clearly matter in various forms of wordplay, including poetry and musical lyrics. The experiments in this article explore whether language processing more generally is sensitive to these properties. Two experiments examined the interaction between phonological and semantic features for written and verbal productions. In Experiment 1, participants were given printed sentences with a missing word, and were asked to generate reasonable completions. The completions reflected both the semantic and the surface features of the preceding context. In Experiment 2, listeners heard sentence contexts, and were asked to rapidly produce a word to complete the utterance. These spontaneous completions again incorporated surface features activated by the context. The results suggest that lexical access in naturalistic language processing is influenced by an interaction between the surface and semantic features of language. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

From syntactic-semantic tagging to knowledge discovery in medical texts

In the GALEN project, the syntactic-semantic tagger MultiTALE is upgraded to extract knowledge from natural language surgical procedure expressions. In this paper, we describe the methodology applied and show that out of a randomly selected sample of such expressions coming from the procedure axis of Snomed International, 81% could be analysed correctly. The problems encountered fall in three different categories: unusual grammatical configurations within the Snomed terms, insufficient domain knowledge and different categorisation of concepts and semantic links in the domain and linguistic models used. It is concluded that the Multi-TALE system can be used to attach meaning to words that not have been encountered previously, but that an interface ontology mediating between domain models and linguistic models is needed to arrive at a higher level of independence from both particular languages and from particular domains.     

Expressing the inexpressible and measuring the incommensurable: On language, literature and psychology.

Presents a broad historical survey of the various reflective, analytical, and scientific approaches to an intellectual understanding of language and literature. Within this overall picture, the contribution of psychology is seen as limited for the area of language and negligible for that of literature. It is argued that psychology has not done justice to language as a complex, hierarchically structured system by concentrating on micro- rather than macro-analysis and on isolated words rather than texts or sequences of natural language. Various characteristics of poetic language (in its semantic, emotional, rhythmic, and phonetic aspects) are offered as prime examples of the type of interpretive challenge to which a maturing science of psychology might be expected to rise. (21 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Alternative ways for knowledge collection, indexing and robust language retrieval

Definitions are provided of the key entities in knowledge representation for Natural Language Processing (NLP). Starting from the words, which are the natural components of any sentence, both the role of expressions and the decomposition of words into their parts are emphasized. This leads to the notion of concepts, which are either primitive or composite depending on the model where they are created. The problem of finding the most adequate degree of granularity for a concept is studied. From this reflection on basic Natural Language Processing components, four categories of linguistic knowledge are recognized, that are considered to be the building blocks of a Medical Linguistic Knowledge Base (MLKB). Following on the tracks of a recent experience in building a natural language-based patient encoding browser, a robust method for conceptual indexing and query of medical texts is presented with particular attention to the scheme of knowledge representation.     

Manual response set in a Stroop-like task involving categorization of English and Japanese words indicates a common semantic representation

The present study examined the semantic representation for English and Japanese words. A previous study using the interlingual word categorization showed Japanese words affected the semantic processing of English words, but English words did not affect the semantic processing of Japanese words. The present study tested the interlingual word categorization by a manual response instead of the vocal response. Analysis showed English words also affected the semantic processing of Japanese words. The common semantic representation for English and Japanese words was discussed.     

Using sound to solve syntactic problems: The role of phonology in grammatical category assignments.

One ubiquitous problem in language processing involves the assignment of words to the correct grammatical category, such as noun or verb. In general, semantic and syntactic cues have been cited as the principal information for grammatical category assignment, to the neglect of possible phonological cues. This neglect is unwarranted, and the following claims are made: (1) Numerous correlations between phonology and grammatical class exist, (2) some of these correlations are large and can pervade the entire lexicon of a language and hence can involve thousands of words, (3) experiments have repeatedly found that adults and children have learned these correlations, and (4) explanations for how these correlations arose can be proposed and evaluated. Implications of these phenonema for language representation and processing are discussed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

The heart of what's the matter. The semantics of illness in Iran

Our understanding of the psychosocial and cultural dimensions of disease and illness is limited not merely by a lack of empirical knowledge but also by an inadequate medical semantics. The empiricist theories of medical language commonly employed both by comparative ethnosemantic studies and by medical theory are unable to account for the integration of illness and the language of high medical traditions into distinctive social and symbolic contexts. A semantic network analysis conceives the meaning of illness categories to be constituted not primarily as an ostensive relationship between signs and natural disease entities but as a 'syndrome' of symbols and experiences which typically 'run together' for the members of a society. Such analysis dirests our attention to the patterns of associations which provide meaning to elements of a medical lexicon and to the constitution of that meaning through the use of medical discourse to articulate distinctive configurations of social stress and to negotiate relief for the sufferer. This paper provides a critical discussion of medical semantics and develops a semantic network analysis of 'heart distress', a folk illness in Iran.     

Encoding, retrieval, main effects and interactions: Were Lockhart and Craik (1990) on the level?

Investigated whether a main effect of encoding emerges when retrieval tests are equated for difficulty. 132 introductory psychology students were tested in 2 experiments. Ss encoded words semantically or by rhyme, and then judged whether unstudied words were related to studied words. A list structure (L. Jacoby, 1975) was adapted to remove the overall difference in difficulty across semantic and rhyme retrieval tasks. Results indicate no main effect of retrieval task, and the main effect of encoding was not dependent on the retrieval task. Semantic encoding led to a greater likelihood that Ss would judge an unstudied word to be related to a studied word. It was concluded that test words shared either a semantic or a rhyme relation with studied words, the advantage of semantic encoding was not dependent on the retrieval criterion. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Attributes of theories of meaning.

This article reviews and evaluates empirical evidence on psychological theories of semantic representation. Four theories are identified. These theories can be distinguished on the basis of (a) whether comprehension of a word requires decomposition into semantic components and (b) whether the units of representation used in comprehension (semantic components) are conceptual primitives. The review covers research on semantic complexity, semantic structure, speech and comprehension errors, effects of context on semantic interpretation, and semantic acquisition. A theoretical framework for understanding semantic acquisition is proposed and defended. The authors conclude that a decompositional theory of psychological semantics is needed to explain language acquisition, production, and comprehension. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Abstract word deficits in aphasia: Evidence from semantic priming.

J.G. and D.E. are nonfluent aphasic patients who appear to have selective problems with abstract words on a variety of standard tests. Such a pattern would normally be interpreted as indicating a central semantic deficit for abstract words. The authors show that this is not the case by means of a semantic priming task that tests for implicit knowledge of the meanings of abstract and concrete words. Spoken word pairs that were either abstract or concrete synonyms (e.g., street-road or luck-chance) were presented; both Ss showed priming for the abstract and concrete pairs. The researchers followed up by asking the Ss to produce definitions to spoken abstract and concrete words; these definitions were also normal. The priming and definition data suggest that the semantic representations of abstract words in these Ss were relatively unimpaired. The researchers found that the Ss have problems only with spoken abstract words in just those tasks where normal controls also have difficulty. In contrast, they clearly have deficits in reading abstract words aloud, which may be due to problems with output phonology. Implications of these data for claims concerning hemispheric differences in the representation of abstract and concrete words are discussed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

基于领域词典与CRF双层标注的中文电子病历实体识别

医疗实体识别是电子病历文本信息抽取的基本任务。针对中文电子病历文本复合实体较多、实体长度较长、句子成分缺失严重、实体边界不清的语言特点以及标注语料难以获取的现状,提出了一种基于领域词典和条件随机场（CRF）的双层标注模型。该模型通过对外部资源的统计分析构建医疗领域词典,再结合条件随机场,进行了两次不同粒度的标注,将领域词典识别的准确性和机器学习的自动性融为一体,从中文电子病历文本中识别出疾病、症状、药品、操作四类医疗实体。该模型在测试数据中的宏精确率为96.7%、宏召回率为97.7%、宏F1值为97.2%。同时对比分析了采用注意力机制的深度神经网络的识别效果,因受到领域数据集大小的限制,在该测试数据集中后者表现不佳。实验结果表明了该双层标注模型对中文医疗实体识别的高效性。      

Primary progressive semantic aphasia: a case study

A longitudinal case study of a patient with a progressive loss of meaning of objects with preserved phonology and syntax is presented. Repeated measures of language, praxis, visual cognition, and semantic processing were carried out. The patient still has preserved conversational speech, social skills, and orientation in her 8th year of her illness, but shows severe anomia and comprehension deficit in all modalities of stimulus presentation. In addition to standardized tests of language, cognition, and memory, specific experiments of categorization, modalities of word access, item consistency, category specificity, and definition of words were carried out. Results indicate a frequency dependent loss of meaning that was consistent in all modalities and throughout all object categories. However, the relative preservation of visual categorization of all categories tested and the language based categorization of animals suggested some fractionation of semantic memory. Relative preservation of autobiographical and personal memories versus semantic memory was a striking observation. Evidence for selective impairment of central semantic processing was obtained from experiments indicating item consistency of loss and the lack of semantic cuing. Neuroimaging evidence of left temporal lobe atrophy and the classical picture is compatible with similar cases published under the term semantic dementia or "transcortical sensory aphasia with visual agnosia" and suggest the diagnosis of Pick's disease.     

Gender differences in the organization of sexual information

One widely used model of knowledge representation is that of a network in which concepts are portrayed as nodes with links between the nodes representing associations. Schvaneveldt (1990) developed a method (Pathfinder) that generates associative networks from individual's ratings of similarity of word pairs. We had 51 female and 47 male undergraduates rate for similarity all paired combinations of 16 words judged as relevant to the domain of sexuality. Using a measure of network similarity, we found that each gender's networks were more similar to each other than they were to the other gender's. Using the number of links on words as the dependent variable there were gender differences in the number of links within clusters of words, between clusters of words, and on specific words. These differences, for the most part, are consistent with gender stereotypes and prior research showing gender differences in the processing of sexual information.     

Clues to the cognitive organization of reading and writing from developmental hyperlexia.

Controversies regarding the organization of the cognitive systems used for processing written language were investigated in a case study of a mentally retarded hyperlexic child. Despite severe impairments in semantic processing, this child demonstrated intact phonological and orthographic processing. Detailed assessments of his nonword reading abilities provided support for the hypothesis that phonological processing of written words was accomplished by using lexical representations instead of applying nonlexical grapheme-phoneme transcoding rules. Longitudinal investigation of the hyperlexic child's development of writing supports the notion that reading and spelling rely on common mental representations rather than separate input and output mental lexicons. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Semantic priming in major depressive state

We studied semantic priming in 20 major depressive subjects. The methodology used was a visual lexical decision task. Semantic priming is the facilitation of target word recognition (shortening of response time) by the prior presentation of a semantically related context (a prime word). It relies on semantic processing of words and context, facilitating early cognitive stages of response. Varying the temporal interval between prime and target words onset allows us to distinguish between two priming mechanisms, relying on more automatic (test 1) or more controlled (i.e. attention dependent) (test 2) information processing. We observe a significant retardation for words and pseudo-words in depressives (in relation to controls) in both tests. In spite of a general retardation and increase of response times in depressives, semantic priming is evident in both groups and both tests, and does not differ significantly between depressive and control groups in either automatic or controlled conditions. Theses results confirm that semantic processing is not impaired in depression, and are discussed with regard to the hypothesis of an effortful processing impairment in depression, and to depressive retardation.