Two bracketing schemes for the Penn Treebank

The trees in the Penn Treebank have a standard representation that involves complete balanced bracketing. In this article, an alternative for this standard representation of the tree bank is proposed. The proposed representation for the trees is loss-less, but it reduces the total number of brackets by 28%. This is possible by omitting the redundant pairs of special brackets that encode initial and final embedding, using a technique proposed by Krauwer and des Tombe (1981). In terms of the paired brackets, the maximum nesting depth in sentences decreases by 78%. The 99.9% coverage is achieved with only five non-top levels of paired brackets. The observed shallowness of the reduced bracketing suggests that finite-state based methods for parsing and searching could be a feasible option for tree bank processing.

Jac. Ahrenberg ja historian perintö : Restaurointisuunnitelmat Viipurin ja Turun linnoihin 1800-luvun lopussa

Jac. Ahrenberg and Patrimony Restoration Plans for Viipuri and Turku Castles at the End of the 19th Century This dissertation examines the unrealized restoration plans for two castles in the Grand Duchy of Finland one located at Viipuri (Vyborg, nowadays in Russia), the other at Turku (in Swedish, Åbo) during the last decades of the 19th century. Both castles were used as prisons, barracks and warehouses. From the middle of the 19th century on, their restoration and transformation into museums and "national monuments" were demanded in the newspapers. The prison reform in the 1860s stimulated the documentation and debate concerning their future, but it was only at the beginning of the 1880s when their restoration became an official state-run project. The undertaking was carried out by Johan Jacob (Jac.) Ahrenberg (1847 1914), architect of the National Board of Public Buildings. By combining written sources with drawings and photographs, this dissertation examines the restoration projects, the two castles' significance and the ways in which they were investigated by scholars. The plans are analyzed in connection with restoration practices in France and Sweden and in the context of contemporary discussions concerning national art and patrimony. The thesis argues that these former castles of the Swedish crown were used to manifest the western roots of Finnish law and order, the lineage of power and the capacity of the nation to defend itself. However, because of their symbolism, their restoration became a politically delicate question concerning the role of the Swedish heritage in Finland's nation-building process. According to Jac. Ahrenberg's plans, the two castles were to be restored to their assumed appearance at the time of the Vasa dynasty. Consequently, the structures would have resembled castles in Sweden. It is suggested that one aim of the restoration plans was to transform the two buildings into monuments testifying to the common history of Sweden and Finland. They were meant to consolidate the Swedish basis of Finnish culture and autonomy and thus to secure them against the threatening implications of Russian imperialism. It seems that along with the changing ideals of architectural restoration and the need for an original Finnish architectural heritage, the political connotations associated with the castles were one reason why Jac. Ahrenberg's restoration plans were never realized.

Monikasvoinen TAVA-partisiippi : Tutkimus suomen TAVA-partisiipin käyttökonteksteista ja verbiliittojen kieliopillistumisesta

The multifaceted passive present participle in Finnish This study investigates the uses of the passive present participle in Finnish. The participle occurs in a variety of syntactic environments and exhibits a rich polysemy. Former descriptions have treated it as a mainly modal element, but it has several non-modal uses as well. The present study provides an overview of its uses and meanings, with the main focus on the factors which trigger the modal reading. In addition, the study contains two case studies on modal periphrastic constructions consisting of the verb 'to be' and the present passive participle, the Obligation construction, e.g., on men-tä-vä [is go-pass-ptc], and the Possiblity construction, e.g., on pelaste-tta-v-i-ssa [is save-pass-ptc-pl-ine]. The study is based on empirical data of 9000 sentences obtained from i) large collections of transcribed material from Finnish dialects, ii) a corpus of modern Finnish newspaper texts, iii) corpora of Old Finnish texts. Both in colloquial and standard Finnish the reading of the participle is highly dependent of the context and determined by such factors as the overall syntactic environment and other co-occurring elements. One of the main findings here is that the Finnish passive present participle is not modal per se. The contextual modal reading arises whenever the state of affairs is conceptualized from the viewpoint of the implied subject of the participle, and the meaning of possibility or obligation depends mostly on whether the situation is pleasant or undesirable. In sections examining the grammaticalization of the Possibility and Obligation constructions, the perspective is diachronic. Both constructions have derived from copula constructions with the passive present participle as a predicate (adjective or adverb). These sections show how a linguistic change can be investigated on the basis of the patterns of usage in the empirical data. The Possibility construction is currently going through a restructuration to a passive verbal complex. The source of this construction is reflected in its present-day use by the fact that it heavily biased towards a small set of verbs. The Obligation construction has grammaticalized to a construction comparable to a compound tense. Patterns of use of the construction show that grammaticalization originates in specific syntactic constructions with an implication of practical necessity. Furthermore, it is shown that the Obligation construction has grammaticalized in different directions in standard and colloquial Finnish. Differing from the study on most typical phenomena investigated in the literature on grammaticalization of modality, the present study opens new perspectives and methods for discussion on these questions.

"Miksi lehmät pitää tappaa?" : Etnografinen tutkimus 2000-luvun alun huumemarkkinoista Helsingissä

The present study focuses on the drug market in Helsinki in the early 2000s, mainly on the dealing in and use of amphetamines, cannabis and the pharmaceutical Subutex. The drug market is usually analysed into upper, middle and lower level markets. These levels are very different in terms of their operating practices, although there may be some mingling. The present study is mainly concerned with drug dealers and users in the lower and middle level markets. Operations also differ depending on whether the dealing involves just one drug or several. Dealing in and using Subutex is a very different business from dealing and using home grown cannabis, for instance: both the customers and the dealers are mostly quite different. The study material was mostly collected through ethnographical field work, including observations and interviews. Interviews with officials and minutes of pre-trial investigations concerning aggravated drug crimes are also included. The study discusses the roles of dealers on the various levels of the drug market in Helsinki and traces activities at various levels. Ethnographical methods are employed to observe day-to-day drug dealing and use and leisure pursuits in private homes and in public premises. The study takes note of the risks inherent in drug dealing and estimates what kind of drug dealers can last the longest on the market without the authorities intervening. At the same time, the study discusses how small groups on the middle and lower levels of the drug market avoid control measures undertaken by the authorities and how the authorities address these groups. Moreover, the study discusses what the drug market is like in prison from the perspective of a drug dealer sent to prison, what their everyday lives are like after release, and how much money dealers on various levels of the drug market make. The study demonstrates that drug dealing in Helsinki, whether we consider the very top or the very bottom of the pyramid, is a far from rational pursuit. The undertakings are not very systematic; they are more a reaction to intoxicant addiction( s) and other problems caused by other dealers, the dealers own actions and the actions of the police. The everyday lives of drug dealers are often chaos only alleviated by drug use in the company of buyers or alone. If a drug dealer uses drugs himself/herself, things become even more complicated and a vicious circle develops. At the same time, everyday life is certainly exciting, and a drug dealer often has a highly eventful if brief life. Drug dealing is a very masculine pursuit, and there is a sort of macho code governing it, although this does not nearly always work as it should. This macho code, typically for illegal activities, involves the threat of violence as a control measure. Hence the untranslatable slang expression Kill the cows : the Finnish word for calf has the slang meaning snitch or police informant . No more cows, no more calves. But informing on others to the authorities is a fact of life in the drug-dealing world. Contributing factors to being reported to the authorities are the dealer s own mistakes and the actions of other dealers and the police. A determined drug dealer will not be deterred from drug dealing by a prison sentence. However, following time in prison only few dealers manage to gain an income from drug dealing commensurate with its risks.

"Pitää kuitenkin lähentyä sitä vangin roolia, vaikka ite ois toista mieltä" : Diskurssianalyyttinen tutkimus toiseuden rakentumisesta vankilaopetuksessa

The focus of this discourse analytical study was to review the meanings which prison education and its participants are given in the formal educational policy and to find out whether these meanings are agreed by the prison students. The intrestes of this thesis were to examine what kind of social and individual promises are related to the prison education and what kind of subject positions are possible to the prison students. The thesis was also interested in the meanings of education in different contexts. The aim of this study was to examine prison education especially for women. This thesis was based on Michel Foucault s ideas of power and the view of this study was directly critical. The prison education was seen as normalizing governance which tends to prepare its subjects to the normal. In this process of differentiating the normal and the abnormal the subjects of prison education are reconstructed as the others . The three research questions of this thesis were: how and what kind of prison education is reconstructed firstly in the strategy of prison education and secondly in the interviews. Thirdly it was questioned how and what kind of meanings gender receive in the discourse of prison education. The main data was consisted by the Finnish Criminal Sanctions Agency s document The Strategy of the Prison education for years 2008 2012 and the three interviews of women who participated to prison education. The interviews were made for this thesis in Autumn 2010. Two other Finnish Criminal Sanctions Agency s documents Prisoners in education (2007) and Visible women (2008) - the report of the working group on female prisoners were also used as subtext materials. The methods of critical and realistic discourse analysis and rhetorical analysis were applied in the analysis. The results of the thesis support the outlook of prison education as a form of normalizing governance which reconstruct the others position for the prison students. Prison education was seen supportive for personal life control and for integrating to the society. The participants of prison education repeated the official discourses in their accounts which is a signal of internalized governance. The interviewees also used different types of anti-discourses when forming implications of prison education.

Investigating online reading with eye tracking and EEG : The influence of text format, reading task and parafoveal stimuli on reading processes

Research on reading has been successful in revealing how attention guides eye movements when people read single sentences or text paragraphs in simplified and strictly controlled experimental conditions. However, less is known about reading processes in more naturalistic and applied settings, such as reading Web pages. This thesis investigates online reading processes by recording participants eye movements. The thesis consists of four experimental studies that examine how location of stimuli presented outside the currently fixated region (Study I and III), text format (Study II), animation and abrupt onset of online advertisements (Study III), and phase of an online information search task (Study IV) affect written language processing. Furthermore, the studies investigate how the goal of the reading task affects attention allocation during reading by comparing reading for comprehension with free browsing, and by varying the difficulty of an information search task. The results show that text format affects the reading process, that is, vertical text (word/line) is read at a slower rate than a standard horizontal text, and the mean fixation durations are longer for vertical text than for horizontal text. Furthermore, animated online ads and abrupt ad onsets capture online readers attention and direct their gaze toward the ads, and distract the reading process. Compared to a reading-for-comprehension task, online ads are attended to more in a free browsing task. Moreover, in both tasks abrupt ad onsets result in rather immediate fixations toward the ads. This effect is enhanced when the ad is presented in the proximity of the text being read. In addition, the reading processes vary when Web users proceed in online information search tasks, for example when they are searching for a specific keyword, looking for an answer to a question, or trying to find a subjectively most interesting topic. A scanning type of behavior is typical at the beginning of the tasks, after which participants tend to switch to a more careful reading state before finishing the tasks in the states referred to as decision states. Furthermore, the results also provided evidence that left-to-right readers extract more parafoveal information to the right of the fixated word than to the left, suggesting that learning biases attentional orienting towards the reading direction.

Cohésion modale et subordination : Le conditionnel et le jussif finnois au miroir de la valeur sémantique et discursive du subjonctif français

Modal cohesion and subordination. The Finnish conditional and jussive moods in comparison to the French subjunctive This study examines verb moods in subordinate clauses in French and Finnish. The first part of the analysis deals with the syntax and semantics of the French subjunctive, mood occurring mostly in subordinate positions. The second part investigates Finnish verb moods. Although subordinate positions in Finnish grammar have no special finite verb form, certain uses of Finnish verb moods have been compared to those of subjunctives and conjunctives in other languages. The present study focuses on the subordinate uses of the Finnish conditional and jussive (i.e. the third person singular and plural of the imperative mood). The third part of the analysis discusses the functions of subordinate moods in contexts beyond complex sentences. The data used for the analysis include 1834 complex sentences gathered from newspapers, online discussion groups and blog texts, as well as audio-recorded interviews and conversations. The data thus consist of both written and oral texts as well as standard and non-standard variants. The analysis shows that the French subjunctive codes theoretical modality. The subjunctive does not determine the temporal and modal meaning of the event, but displays the event as virtual. In a complex sentence, the main clause determines the temporal and modal space within which the event coded by the subjunctive clause is interpreted. The subjunctive explicitly indicates that the space constructed in the main clause extends its scope over the subordinate clause. The subjunctive can therefore serve as a means for creating modal cohesion in the discourse. The Finnish conditional shares the function of making explicit the modal link between the components of a complex construction with the French subjunctive, but the two moods differ in their semantics. The conditional codes future time and can therefore occur only in non-factual or counterfactual contexts, whereas the event expressed by French subjunctive clauses can also be interpreted as realized. Such is the case when, for instance, generic and habitual meaning is involved. The Finnish jussive mood is used in a relatively limited number of subordinate clause types, but in these contexts its modal meaning is strikingly close to that of the French subjunctive. The permissive meaning, typical of the jussive in main clause positions, is modified in complex sentences so that it entails inter-clausal relation, namely concession. Like the French subjunctive, the jussive codes theoretical modal meaning with no implication of the truth value of the proposition. Finally, the analysis shows that verb moods mark modal cohesion, not only on the syntagmatic level (namely in complexe sentences), but also on the paradigmatic axis of discourse in order to create semantic links over entire segments of talk. In this study, the subjunctive thus appears, not as an empty category without function, as it is sometimes described, but as an open form that conveys the temporal and modal meanings emerging from the context.

Research bed for unit selection based text to speech synthesis

The paper describes a modular, unit selection based TTS framework, which can be used as a research bed for developing TTS in any new language, as well as studying the effect of changing any parameter during synthesis. Using this framework, TTS has been developed for Tamil. Synthesis database consists of 1027 phonetically rich prerecorded sentences. This framework has already been tested for Kannada. Our TTS synthesizes intelligible and acceptably natural speech, as supported by high mean opinion scores. The framework is further optimized to suit embedded applications like mobiles and PDAs. We compressed the synthesis speech database with standard speech compression algorithms used in commercial GSM phones and evaluated the quality of the resultant synthesized sentences. Even with a highly compressed database, the synthesized output is perceptually close to that with uncompressed database. Through experiments, we explored the ambiguities in human perception when listening to Tamil phones and syllables uttered in isolation,thus proposing to exploit the misperception to substitute for missing phone contexts in the database. Listening experiments have been conducted on sentences synthesized by deliberately replacing phones with their confused ones.

Methods for Automatic Evaluation of Sentence Extract Summaries

This paper describes three novel techniques to automatically evaluate sentence extract summaries. Two of these techniques called FuSE and DeFuSE evaluate the quality of the generated extract summary based on the degree of similarity to the model summary. They use a fuzzy set theoretic basis to generate a match score. DeFuSE is an enhancement to FuSE and uses WordNet based hypernymy structures to detect similarity between sentences at abstracted levels. The third technique focuses on quantifying the quality of an extract summary based on the difficulty in generating such a summary. Advantages of these techniques are described with examples.

Improving query focused summarization using look-ahead strategy

Query focused summarization is the task of producing a compressed text of original set of documents based on a query. Documents can be viewed as graph with sentences as nodes and edges can be added based on sentence similarity. Graph based ranking algorithms which use 'Biased random surfer model' like topic-sensitive LexRank have been successfully applied to query focused summarization. In these algorithms, random walk will be biased towards the sentences which contain query relevant words. Specifically, it is assumed that random surfer knows the query relevance score of the sentence to where he jumps. However, neighbourhood information of the sentence to where he jumps is completely ignored. In this paper, we propose look-ahead version of topic-sensitive LexRank. We assume that random surfer not only knows the query relevance of the sentence to where he jumps but he can also look N-step ahead from that sentence to find query relevance scores of future set of sentences. Using this look ahead information, we figure out the sentences which are indirectly related to the query by looking at number of hops to reach a sentence which has query relevant words. Then we make the random walk biased towards even to the indirect query relevant sentences along with the sentences which have query relevant words. Experimental results show 20.2% increase in ROUGE-2 score compared to topic-sensitive LexRank on DUC 2007 data set. Further, our system outperforms best systems in DUC 2006 and results are comparable to state of the art systems.

Obtaining single document summaries using latent dirichlet allocation

In this paper, we present a novel approach that makes use of topic models based on Latent Dirichlet allocation(LDA) for generating single document summaries. Our approach is distinguished from other LDA based approaches in that we identify the summary topics which best describe a given document and only extract sentences from those paragraphs within the document which are highly correlated given the summary topics. This ensures that our summaries always highlight the crux of the document without paying any attention to the grammar and the structure of the documents. Finally, we evaluate our summaries on the DUC 2002 Single document summarization data corpus using ROUGE measures. Our summaries had higher ROUGE values and better semantic similarity with the documents than the DUC summaries.

Simultaneously Uncovering the Patterns of Brain Regions Involved in Different Story Reading Subprocesses

Story understanding involves many perceptual and cognitive subprocesses, from perceiving individual words, to parsing sentences, to understanding the relationships among the story characters. We present an integrated computational model of reading that incorporates these and additional subprocesses, simultaneously discovering their fMRI signatures. Our model predicts the fMRI activity associated with reading arbitrary text passages, well enough to distinguish which of two story segments is being read with 74% accuracy. This approach is the first to simultaneously track diverse reading subprocesses during complex story processing and predict the detailed neural representation of diverse story features, ranging from visual word properties to the mention of different story characters and different actions they perform. We construct brain representation maps that replicate many results from a wide range of classical studies that focus each on one aspect of language processing and offer new insights on which type of information is processed by different areas involved in language processing. Additionally, this approach is promising for studying individual differences: it can be used to create single subject maps that may potentially be used to measure reading comprehension and diagnose reading disorders.

SPEECH ENHANCEMENT USING DISCRIMINATIVE RANDOM FIELDS

Speech enhancement in stationary noise is addressed using the ideal channel selection framework. In order to estimate the binary mask, we propose to classify each time-frequency (T-F) bin of the noisy signal as speech or noise using Discriminative Random Fields (DRF). The DRF function contains two terms - an enhancement function and a smoothing term. On each T-F bin, we propose to use an enhancement function based on likelihood ratio test for speech presence, while Ising model is used as smoothing function for spectro-temporal continuity in the estimated binary mask. The effect of the smoothing function over successive iterations is found to reduce musical noise as opposed to using only enhancement function. The binary mask is inferred from the noisy signal using Iterated Conditional Modes (ICM) algorithm. Sentences from NOIZEUS corpus are evaluated from 0 dB to 15 dB Signal to Noise Ratio (SNR) in 4 kinds of additive noise settings: additive white Gaussian noise, car noise, street noise and pink noise. The reconstructed speech using the proposed technique is evaluated in terms of average segmental SNR, Perceptual Evaluation of Speech Quality (PESQ) and Mean opinion Score (MOS).

El uso de detalles biográficos como información seductora en el texto de ciencias

Resumen: Se estudió el efecto de incluir detalles biográficos seductores en un texto expositivo de ciencias. 66 estudiantes con bajo conocimiento sobre el tema leyeron el texto sin detalle seductor o con una anécdota biográfica interesante. El interés asociado a los materiales fue examinado en un estudio preliminar. Se recolectaron medidas de retención (recuerdo del texto y verificación de afirmaciones literales), comprensión (verificación de afirmaciones inferidas) y síntesis del texto (selección de un título). Los resultados indicaron que la condición que recibió el detalle biográfico tuvo más dificultades para recordar los contenidos y para contestar afirmaciones de la sección del texto próxima al detalle. Estos resultados son interpretados a la luz de la hipótesis de la integración desviada.

La reinserción social de los ex-presidiarios en la Argentina de las primeras décadas del siglo XX : algunos comentarios biográficos sobre Jorge H. Frías, fundador del Patronato de Liberados y Excarcelados de la Capital Federal

Resumen: Este trabajo pretende avanzar sobre una problemática escasamente abordada por la historiografía jurídica y social argentina: la reinserción social de los egresados de las prisiones. Así, partiendo de la utilidad que tienen para la historiografía los estudios biográficos y de instituciones penitenciarias y post-penitenciarias, analizaremos algunos aspectos de la labor profesional de Jorge H. Frías, distinguido jurista, presidente de la Cámara de Apelaciones en lo Criminal y Correccional y fundador en 1918 de la primera institución de reinserción social de la Argentina: el Patronato de Liberados y Excarcelados de la Capital Federal. Esperamos que este breve trabajo contribuya a alentar la realización de otros estudios sobre diferentes aspectos del universo penitenciario argentino (directores de presidios, viajeros penitenciarios, congresos penitenciarios nacionales e internacionales, revistas de los funcionarios de prisiones, etc.) que aún hoy en día permanecen desconocidos.