masood ghayoomi

In the comprehensive scientific roadmap of the country (Iran), the promotion of Persian language as a science language among other international science languages is taken into consideration. One of the ways to reach this goal is suggested as expanding the usage of the Persian language. To achieve the goal, which can be raised in the field of Persian language policy making, it is necessary to understand more about the linguistic content properties and the basic concepts that are taught in the textbooks to students. The description of these features can be considered when preparing the language content. In this research, a corpus of textbooks from grades 1 to 6 (the primary school period) is developed that contains around 208,000 words and annotated. These courses include Farsi, Experimental Sciences, Social Studies and Heavenly Gifts. All the sentences of different courses are written in plain text files, separated by grade and course, and after normalization in the pre-processing process, they are annotated automatically at four levels: broad transliteration, lemmatization, part-of-speech and syntactic constituency parsing. The results of this research can help to know more about the content of textbooks and to be useful in the fields of education and policy making in language planning.

The present study aimed to study the preposition /az/ (i.e., from) in primary school textbooks. Polysemy of prepositions is one of the sources of many students’ perception. A text corpus developed for this study from Farsi, Practical Science, Heavenly Gifts, and Social Studies of elementary school textbooks grades 1 to 6. To analyze the data, Rosch’s (1975) prototype theory and Tyler and Evans’ (2003) systematic prepositions polysemy approaches in the cognitive semantics framework were used. At first, the sample sentences which contained the preposition /az/ (from) were extracted to form the sub-corpus. The set of senses that are obtained from the sub-corpus are “Type”, “Whole Part”, “Tense, Source”, “Comparation”, “with the help of, By, With”, “About”, “Orientation”, “Motion”, “Metaphor”, and “Complement”. Statistical distribution of meanings in the whole sub-corpus (general domain) was “Whole Part”. The senses of the proposition among lessons and grades (domain restricted) are not similar to the prototype sense. This research showed that the sense “Complement” is the prototype for grades 1 to 3, and the sense “Whole Part” is more frequent in grade 4 to 6. Different forms of this preposition are used at the beginning, middle, and end of sentence. Furthermore, the findings showed that the sense “Metaphor” is the prototype for the book Farsi, the sense “Whole Part” is the prototype sense for Practical Science and Social Studies, and the sense “Tense” is the prototype sense for Heavenly Gifts.

Dialectology studies a dialect scientifically along with its geographical distribution.Each dialect is a language; and to study a dialect various linguistic analyses are required.This property makes the study of a language a little long in terms of time.Collecting dialectical data is very time consuming and required a lot of effort.Raw data is not much usable in dialectology and it is required to add linguistic analyses to the data in the framework of structural linguistic analysis.Using a computer as a research tool causes to prepare the data in a specific structure.The main contribution of the current paper is proposing a standard to organize dialectic data and information.This standard contains the dialectic data, its relevant meta-data, and the linguistic information related to the analysis of this data.The meta-data and linguistic information are organized in the XML tree structure.This data structure is highly portable and it can be easily read into a database.

Nowadays, the existence of emerging communication tools has made communication between speakers possible through writing. The electronic, global and interactive nature of such emerging technologies has facilitated and increased the speed of communication. The linguistic interaction by using these tools and the relationship between speech and writing have causes a type of writing to be created by the users of a language, a writing type known as “neography”. The main aim of this research is to investigate the properties of neography in Persian and classify them into the categories based on a linguistic corpus developed from the data published in social media. To this end, the corpus is semi-automatically annotated based on the standard writing style, and the linguistic properties at phonetic, morphological and syntax levels. Then, the words whose written forms are different from the standard form and have a type of neography are studied based on the orthography properties and also linguistic features. The results of the analyzing the data and the assigned labels show that neography in Persian in the virtual space occurs at two levels of oryhography and morpho-phonetic, and the content words bear the highest amount of neography in the words.

Annually, researchers in various scientific fields publish their research results as technical reports or articles in proceedings or journals. The collocation of this type of data is used by search engines and digital libraries to search and access research publications, which usually retrieve related articles based on the query keywords instead of the article’s subjects. Consequently, accurate classification of scientific articles can increase the quality of users’ searches when seeking a scientific document in databases. The primary purpose of this paper is to provide a classification model to determine the scope of scientific articles. To this end, we proposed a model which uses the enriched contextualized knowledge of Persian articles through distributional semantics. Accordingly, identifying the specific field of each document and defining its domain by prominent enriched knowledge enhances the accuracy of scientific articles’ classification. To reach the goal, we enriched the contextualized embedding models, either ParsBERT or XLM-RoBERTa, with the latent topics to train a multilayer perceptron model. According to the experimental results, overall performance of the ParsBERT-NMF-1HT was 72.37% (macro) and 75.21% (micro) according to F-measure, with a statistical significance compared to the baseline (p<0.05).

The verb particles of the verb phrases in Persian were simple verbs in which new role don't have their primary meaning. The purpose of this research is to identify what verbs in New Persian have gone through this passage, which ones have been more stable in the language, and also which ones have had strong tendencies to have new roles in Persian verb phrases. For this, we have used a quantitative statistical criterion called “coefficient of variance”. This parameter creates a new ranking for verbs, which shows a numerical presentation of their behavioral changes. The results show that “zadan” has had the most behavioral change and “dāštan” has been the most stable verb particle.

Verb is one of the essential elements of sentence.  Like the other linguistic components, verbs are also born, go through changes in their lifetime and then they may die or may have a longer lifetime. From one point of view, structurally, verbs can be classified into two general categories: simple verbs and verbs with complex structures. Generally, simple verbs, which are usually one word, have the same semantic content represented lexically.  The primary goal of this paper is to investigate which verbs have been used as verb particles in the structure of modern Persian non-simple verbs in the last twelve centuries. Therefore, instead of the term “light verb” used for the verb component in compound verbs, the term “verb particle” has been used to include more verb constructions. Besides, this research seeks to identify verbs used as verb participles and find those verbs that have been continuously present during recent 12 centuries. Obviously, the frequency of verbs that have this condition is not the same. So, if we are looking for a quantitative measure for it, it is clear that frequency is not a suitable one. Therefore, another goal of this research is to provide a quantitative measure to determine which of these verbs, which have always been verb particle, have a growing tendency to change from simple verbs to verb participles, and which ones show almost constant behavior.

In order to observe language change, Paul Baker has done a corpus survey on English. His study deals with the concept of “lockwords”. A lockword is a word that may change its meaning or even use, but still have an approximately equal number in the corpus, especially in multi-part historical corpuses. The aim of Baker's research was to examine how English vocabulary has developed over the past eight decades. His main research questions were to identify words whose frequency was increasing, decreasing, or remaining constant over time relative to other words. He tries to provide acceptable explanations for such phenomena. Unlike previous studies, Baker was also interested in stability including which words have not changed much over time in recent English corpora.

Conducting this research required a historical corpus. This corpus contains 1200 pages of complete text, with 100 pages from each century as random samples. That is, 100 pages with 12 B Nazanin font have been selected from the texts of each century, excluding any type of paragraphing and punctuation marks. This body contains more than 1,000,000 words in total. But even with a text corpus, this research wouldn’t be completed. Rather, it was necessary to create a labeled body.

182 different verb participles were found in this corpus from which only 24 were used in all centuries. Their appropriate quantitative measure is the “coefficient of variance” that quantitatively determines their tendency during these centuries. Therefore, the frequency factor alone cannot show this trend accurately.

This research sought to find a quantitative measure to show which verbs have had a greater tendency to be used in more complicated verb structures, as well as in simple ones, from 10th century until now. First, we had to investigate which verbs participated in this process. Next, 182 verbs were found from which only 24 were continuously found in all centuries.
Therefore, the investigation was continued on these 24 verb components. Then, by examining the tools provided by statistics, a search was made to find the best measure to quantify the behavior of these verb components. According to Baker, neither the frequency nor the standard deviation was appropriate for this research, but the appropriate measure was the coefficient of variance. Based on the results obtained from this criterion, a new ranking of the behavior of the verb components in terms of participation in more complex constructions was obtained. In this ranking, the verb “zadan” showed the most tendency to participate in this type of verb constructions and the verb “dāštan” had the least change. That is, “dāštan” has always participated in these types of verb constructions in a stable way during these centuries. Therefore, it can be concluded that “zaden” has shown the most tendency to participate in more complex verb constructions and the verb “dāštan” is the most stable lockword in the verb class in New Persian.

News discourse is a type of discourse analysis that deals with the analysis of news discourse. Due to the fact that in the formatting of news there are two hidden features of selection and prominence in the communication representation of news, the inverted pyramid of news is used to grade the importance of the discourse parts of the news. Although it is desirable to meet the structure of the inverted pyramid of news, sometimes this structure may change. In this article, we put an effort to analyse the discourse analysis of Persian news websites with the help of statistical analysis. To research the goal, data science can be used. This inter-discipline deals with data analysis from a scientific aspect, finding implicit concepts to be obtained from data analysis and extracting knowledge from the data. In the framework of data science, we examined the Persian news corpus and studied the existence of semantic correlation between the news title and the news content based on the structure of the news inverted pyramid. To achieve the goal, by using the crawling method, a relatively large news corpus with a volume of 14 billion words has been obtained from 24 news websites. After pre-processing and normalizing the corpus, in the framework of distributional semantics, the vector of title news and content have been created by using the Word2Vec tool for creating the vector model to have the vector representation of each news. After segmenting news content into three parts (lead, body and further explanation about the lead) according to the inverted pyramid, the Pearson correlation coefficient has been used to calculate the correlation between the title and each part of the news. Although Pearson's correlation coefficient was positive for a large number of news, zero value and no correlation was found for the news. On average, the correlation between the headline and the news lead and body was higher than the correlation between the headline and the lead development. This research can be used as a method to carefully select the title and content and filter the news according to the inverted pyramid structure.

In this research, an attempt is made to investigate the characteristics of Persian fake news related to Covid-19 by using statistical analysis.  To this end, first, a language corpus containing reliable and fake news in Persian in the field of Corona is prepared. Then, the language patterns of these two data sets, as well as two statistical analyzes of the amount of information and the readability of reliable and fake news, are examined and compared with each other. According to the exteracted information and the experimental results achieved from the developed corpus on COVID-19 fake news, there are common language patterns in these two datasets. Moreover, the amount of information in reliable news is more than fake news based on two measures of entropy and surprise. Based on the results, the readability level of the fake news is measured based on the readability formulas. According to the results, the text of fake news is simpler than real news. In the process of automatic labeling of reliable and fake news based on the level of difficulty, most news is recognized as simple texts. The results show that fake news is mostly simple and not difficult compared to reliable news. In addition to this achievement, to study linguistic properties of fake news statistically based on the information amount and readability, the applicablity of this statistical information was studied to detect fake news using machine learning methods.

Coronavirus pandemic caused changes in the daily lifestyle, such as reducing social interactions and creating social distancing. In this research, we pursue two goals. One is algorithmic content analysis of comments/posts in Persian related to the Coronavirus on two social media, namely Tweeter and Instagram. To this end, topic modeling is used as a method for content analysis to cluster the data into abstract topics. The other goal is finding the correlation between topics and hashtags in the comments/posts. To this end, we developed a corpus from these two social media. We found 24 abstract topics by algorithmic content analysis of this corpus and they were manually labeled to be comprehensive. According to the corpus and the statistical information of the extracted topics, it can be speculated that about 25% of the comments/posts in this corpus focused on political and social issues of the virus. 10 fine-grained topics which contained 35% of the comments were related to the Coronavirus itself and its pandemic property. This indicates the importance of the attention that has been paid to social media for informing and disseminating information. Furthermore, the hypothesis of existing correlation between topics and hashtags was studied from statistical point of view by using the Pearson correlation coefficient. For 20 topics, a high correlation score between topics and hashtags was found; but this correlation was not found for 4 topics. The outcome of this research can be used to increase the internal coherence of a text and to make the hashtags predictable.

Vocabulary differences between languages indicate differences in the cultures associated with each language. Polysemous words in any culture can be an expression of such lexical differences in a language. In cognitive semantics, there are different approaches to the polysemous phenomenon in language; The theory of Frame semantics is one of these approaches in cognitive semantics that has looked at the issue of polysemy from the perspective of semantic frames. The revised form of this theory has been realized in FrameNet . Semantic frames are semantic packages and each frame is composed of semantic components called frame elements. Semantic relationships in this network are presented as relationships between semantic frames. In this article, an attempt is made to provide a semantic and corpus-based analysis of the polysemy of the conceptual domain of the verb "Listening" in Persian based on the principles of the FrameNet network in English. Accordingly, the keywords "šeno", " šenid" and "guš" have been studied in the Persian database . Among the concepts found in sentences extracted from the corpus, six semantic frames of Perception_active, Perception_experience, Awareness, Compliance, Seeking and Attention and two "inheritance" Frame to frame relations between "Perception-Active" and "Perception-Experience" Frames and "use" Frame to frame relations between "Attention" and Perception-Active" Frames were extracted. The results of the analysis show that the current approach to the phenomenon of polysemy is an efficient approach to study the meaning of the verb "Listening" in Persian and can provide a picture of Persian language culture in this conceptual area.

This paper provides a comparative analysis of cross-lingual word embedding by studying the impact of different variables on the quality of the embedding models within the distributional semantics framework. Distributional semantics is a method for the semantic representation of words, phrases, sentences, and documents. This method aims at capturing as much information as possible from the contextual information in a vector space. The early study in this domain focused on monolingual word embedding. Further progress used cross-lingual data to capture the contextual semantic information across different languages. The main contribution of this research is to make a comparative study to find out the superior impact of the learning methods, supervised and unsupervised in training and post-training approaches in different embedding algorithms, to capture semantic properties of the words in cross-lingual embedding models to be applicable in tasks that deal with multi-languages, such as question retrieval. To this end, we study the cross-lingual embedding models created by BilBOWA, VecMap, and MUSE embedding algorithms along with the variables that impact the embedding models' quality, namely the size of the training data and the window size of the local context. In our study, we use the unsupervised monolingual Word2Vec embedding model as the baseline and evaluate the quality of embeddings on three data sets: Google analogy, mono- and cross-lingual words similar lists. We further investigated the impact of the embedding models in the question retrieval task.

Since the advent of the Internet, we have come across a large amount of data that has been placed on this platform. This has caused the way data is studied and examined to evolve from manual to automatic. This research aims to extract the topics raised in articles in the field of language and linguistics and to study the evolution of topics over time with the help of the topic modeling algorithm. To this end, the Latent Dirichlet allocation algorithm is used. The corpus of this research was developed through crawling; and after filtering and pre-processing of the data, the number of 5, 10 and 15 topics are extracted from the articles; and based on the words of each topic, they are manually labeled. Due to the distribution of articles from 1927 to 2020 (1306 to 1399 according to the solar calendar), the time is divided into 5 time span and the topics related to each time span is identified after the labeling process. Among the 5 extracted topics, including Education, Discourse Analysis, Psychology, Syntax and Dialectology, all topics, except Discourse Analysis, are progressively studied over time. The growth and decline of the topics extracted from the articles can be seen in 10 and 15 topics. The practical achievement of this research is policy-making in the field of science. In addition to introducing a practical, technological methodology in do research, it is possible to identify hot topics among researchers in a scientific field and to find the gaps in research topics and to try to diversify and balance research topics.

Since 2001s (1380s according to the Iran’s solar calendar), the increasing rate of writing and publishing scientific articles in Iran has become very intense. This caused in addition to the governmental organizations, such as Irandoc & the National Library and Archives of the Islamic Republic of Iran, numerous other online systems, such as the General Portal of Humanities, Noormags, Magiran, Elmnet, Civilica, etc, to manage knowledge and to provide structured archives of the scientific documents. Each of these archives provides facilities to the user. One of these facilities is searching on the documents. An accurate search can greatly improve the usage of these online systems. To increase the accuracy of the search result, it is necessary to determine the scientific field of articles. Classifying large volumes of scientific resources in different fields is very time-consuming. Using machinery methods can be a solution to reduce the severity of the task.The main contribution of this paper is to provide a classification model to classify Persian scientific articles. Although in previous studies, the classification task has been mainly used for simple texts, in this study, the neural network-based classification models, such as convolutional and perceptron neural networks, are used with the contextualized semantic representation, such as ParsBERT; and the results are compared with the other common method utilized for vectorization, namely Word2Vec. To this end, we use the data from the General Portal of Humanities, which includes various articles in the Humanities and each article contains the label of the field. One of the neural network characteristics is that a set of hidden features from the data in the vector space is created and used to train the model. According to the experimental results, the Perceptron classifier that utilized ParsBERT representation obtained the highest performance which is 74.71% based on the Micro F-score, and 72.55% based on the Macro F-score.

Fillmore's “Frame theory” (1982) suggests a major principle in Cognitive Semantics that believes the meanings of words are perceived within a system of knowledge arising from the human's cultural experience. It maintains that “Semantic Frames” display sections of an event used for connecting a group of words to a set of meanings. Verbs are present as the most important element in most of the situations of an event and play significant role in the interpretation of the meaning. The purpose of this study is to present a quantitative and corpus-based analysis in order to recognize the semantic frames of “∫odæn” (to become) in Persian within the framework of this theory. Thus, 14 semantic frames were recognized using encyclopedias, Persian contemporary corpora and FrameNet. Then, the concept of “tæɁɢir-e hαlæt dαdæn” (to turn to a state) was determined as the prototype of this verb and the main frame was created.

One subfield of assessment of language proficiency is predicting language proficiency level.This research aims at proposing a computational linguistic model to predict language proficiency level and to explore the general properties of the levels. To this end, we collect the data from Persian learners' textbooks and extract statistical and linguistic features from this text corpus to train 3 classifiers as learners. The performance of the model varies based on the learning algorithm and the feature set(s) used for training the model. For evaluating the models, four standard metrics, namely accuracy, precision, recall, and F-measure are used. Based on the results, the model created by the Random Forest classifier performed the best when statistical features extracted from raw text is used. The Support Vector Machine classifier performed the best by using linguistic features extracted from the corpus annotated automatically. This determines that enriching the model and providing various kinds of information do not guarantee that a classifier (learner) performs the best.To discover the latent teaching methodology of the textbooks, we studied the general performance of the classifiers with respect to the language level and the linguistic knowledge used for creating the model. Based on the obtained results, the amount of extracted features plays an important role during to training a classifier. Furthermore, the average best performance of the classifiers is extending the linguistic knowledge from syntactic patterns at level A to all linguistic information at levels B and C.

Words in the natural language have forms and meanings, and there might not always be a one-to-one match between them. This property of the language causes words to have more than one meaning; as a result, a text processing system faces challenges to determine the precise meaning of the target word in a sentence. Using lexical resources or lexical databases, such as WordNet, might be a help, but due to their manual development, they become outdated by passage of time and language change. Moreover, the lexical resources might be domain dependent which are unusable for open domain natural language processing tasks. These drawbacks are a strong motivation to use unsupervised machine learning approaches to induce word senses from the natural data. To reach the goal, the clustering approach can be utilized such that each cluster resembles a sense. In this paper, we study the performance of a word sense induction model by using three variables: a) the target language: in our experiments, we run the induction process on Persian and English; b) the type of the clustering algorithm: both parametric clustering algorithms, including hierarchical and partitioning, and non-parametric clustering algorithms, including probabilistic and density-based, are utilized to induce senses; c) the context of the target words to capture the information in vectors created for clustering: for the input of the clustering algorithms, the vectors are created either based on the whole sentence in which the target word is located; or based on the limited surrounding words of the target word. We evaluate the clustering performance externally. Moreover, we introduce a normalized, joint evaluation metric to compare the models. The experimental results for both Persian and English test data showed that the windowbased partitioningK-means algorithm obtained the best performance.

Recent attitude towards studying a language and a linguistic phenomenon is based upon the existence of a collection of data; therefore it is required to develop a linguistic corpus that is naturally occurred and it is not collected from the one’s intuition. This research methodology is highly important to study linguistic historical data, which is dead and has no speaker.The current research puts an effort to develop a linguistic corpus of middle Persian and to organize the data in a data-base. To this end, six information levels are determined in the annotation process, including transliteration of the Pahlavi texts, transcription of the words along with their Persian translation, defining fine-grained syntactic category of the words, lemmatizing the words, and identifying whether the word is huzwāreš or not. To define fine-grained syntactic categories, the tag set for contemporary Persian developed by Bijankhan et al (2011) and organized by Ghayoomi (2004) are modified and adapted to the Pahlavi language according to the requirements. The new tag set is used to label Pahlavi words. After annotating words and organizing the information, extracting the statistical information is possible to deepen the insight over the text’s content.

The dependency grammar is one of the grammar formalisms that is originated according to the dependency relations of lexical elements in a sentence. This dependency is based on the valence of the lexicon; therefore, in this formalism the semantic representation of the words as valence is the backbone of the grammar. The current paper reviews the book entitled “Persian Grammar based on the Autonomous Theory in Dependency Grammar” according to three major contents. Firstly, the chapters of the book are briefly overviewed. Then, the book’s content is evaluated and analyzed and the weak and strong points of the book are studied from different perspectives. In this section, some suggestions are provided for the terminology and expressed concepts. Since dependency grammar does not belong to generative grammar, a comparison is made between the dependency and phrase structure grammar formalisms to understand the content better and to ease the mind of the reader with the generative grammar background about the basic concepts of the dependency grammar. At the end, suggestions are provided to extend the dependency grammar in the book.

Reading is undoubtedly one of the most valuable skills of human beings. This complex behavior is composed of several distinct skills. Different theories about reading methods have been proposed and many researchers believe that reading methods varies in languages because of the different transparency in their orthography, so the purpose of this study is to investigate the Persian word reading pattern in children with dyslexia and normal children and compare their abilities.

For conducting this cross-sectional study, after issuing required permits, the elementary school children with dyslexia were identified by referring to learning disorders schools in Tehran City, Iran. After evaluating and diagnosing by the psychologist and speech therapist, 16 students with dyslexia were found. Then 32 normal age-matched students of Tehran’s public schools were randomly selected from the available population as the control group. The reading and phonological awareness tests were then performed in different sessions. The results of the tests were recorded and the data were analyzed by the Mann-Whitney and Spearman tests in SPSS version 20.

There was a significant difference between children with dyslexia and their age-matched normal group in reading skills and phonological awareness (P<0.05). In both groups, the mean percentage in reading irregular words was higher than the mean percentage of non-word reading. There is a significant correlation between reading speed and reading irregular words in both groups, and also between reading accuracy and reading non-words (P<0.05). There was a significant correlation between the reading speed and reading non-words in the group with dyslexia (P<0.05). But in normal children, there is no significant correlation between these tasks. This difference shows the problem of children with dyslexia pertains to reading non-words.

Based on the results of the present study, in children with dyslexia, there is a problem with reading non-words, and the dual-route of word reading has changed to only the whole word reading route. So it is important to consider phonological awareness skills training in children with dyslexia.

Word Sense Induction (WSI) aims at inducing word senses from data without using a prior knowledge. Utilizing no labeled data motivated researchers to use clustering techniques for this task. There exist two types of clustering algorithm: parametric or non-parametric. Although non-parametric clustering algorithms are more suitable for inducing word senses, their shortcomings make them useless. Meanwhile, parametric clustering algorithms show competitive results, but they suffer from a major problem that is requiring to set a predefined fixed number of clusters in advance.
Word Sense Induction (WSI) aims at inducing word senses from data without using a prior knowledge. Utilizing no labeled data motivated researchers to use clustering techniques for this task. There exist two types of clustering algorithm: parametric or non-parametric. Although non-parametric clustering algorithms are more suitable for inducing word senses, their shortcomings make them useless. Meanwhile, parametric clustering algorithms show competitive results, but they suffer from a major problem that is requiring to set a predefined fixed number of clusters in advance.
The main contribution of this paper is to show that utilizing the silhouette score normally used as an internal evaluation metric to measure the clusters’ density in a parametric clustering algorithm, such as K-means, in the WSI task captures words’ senses better than the state-of-the-art models. To this end, word embedding approach is utilized to represent words’ contextual information as vectors. To capture the context in the vectors, we propose two modes of experiments: either using the whole sentence, or limited number of surrounding words in the local context of the target word to build the vectors. The experimental results based on V-measure evaluation metric show that the two modes of our proposed model beat the state-of-the-art models by 4.48% and 5.39% improvement. Moreover, the average number of clusters and the maximum number of clusters in the outputs of our proposed models are relatively equal to the gold data

In linguistics, a tree bank is a parsed text corpus that annotates syntactic or semantic sentence structure. The exploitation of tree bank data has been important ever since the first large-scale tree bank, The Penn Treebank, was published. However, although originating in computational linguistics, the value of tree bank is becoming more widely appreciated in linguistics research as a whole. For example, annotated tree bank data has been crucial in syntactic research to test linguistic theories of sentence structure against large quantities of naturally occurring examples. The natural language parser consists of two basic parts, POS tagger and the syntax parser. A Part-Of-Speech Tagger (POS Tagger) is a piece of software that reads text in some languages and assigns parts of speech to each word (and other token), such as noun, verb, adjective, etc., although generally computational applications use more fine-grained POS tags like 'noun-plural'. A natural language parser is a program that works out the grammatical structure of sentences, for instance, which groups of words go together (as "phrases") and which words are the subject or object of a verb. Probabilistic parsers use knowledge of language gained from hand-parsed sentences to try to produce the most likely analysis of new sentences. These statistical parsers still make some mistakes, but commonly work rather well. Inaccurate design of context-free grammars and using bad structures such as Chomsky normal form can reduce accuracy of probabilistic context-free grammar parser. Weak independence assumption is one of the problems related to CFG. We have tried to improve this problem with parent and child annotation, which copies the label of a parent node onto the labels of its children, and it can improve the performance of a PCFG. In grammar, a conjunction (conj) is a part of speech that connects words, phrases, or clauses that are called the conjuncts of the conjunctions. In this study, we examined the conjunction phrases in the Persian tree bank. The results of this study show that adding structural dependencies to grammars and modifying the basic rules can remove conjunction ambiguity and increase accuracy of probabilistic context-free grammar parser. When a part-of-speech (PoS) tagger assigns word class labels to tokens, it has to select from a set of possible labels whose size usually ranges from fifty to several hundred labels depending on the language. In this study, we have investigated the effect of fine and coarse grain POS tags and merging non-terminals on Persian PCFG parser.

A word is the smallest unit in the language that has 'form' and 'meaning'. The word might have more than one meaning in which its exact meaning is determined according to the context it is appeared. Collecting all words’ senses manually is a tedious and time consuming task. Moreover, it is possible that the words’ meanings change over time such that the meaning of an existing word will become unusable or a new meaning will be added to the word. Computational methods is one of the approaches used for identifying words’ senses with respect to the linguistic contexts.In this paper, we put an effort to propose an algorithm to identify senses of Persian words automatically without a human supervision. To reach this goal, we utilize the word embedding method in a vector space model. To build words’ vectors, we use an algorithm based on the neural network approach to gather the context information of the words in the vectors. In the proposed model of this research, the divisive clustering algorithm as one of hierarchical clustering algorithms fits with the requirements of our research question. In the proposed model, two modes, namely the Sentence-based and the Context-based, are introduced to identify words’ senses. In the Sentence-based mode, all of the words in a sentence that contain the target word are involved to build the sentence vector; while in the Context-based mode, only a limited number of surrounding words of the target word is involved to build the sentence vector. Two evaluation methods, namely internal and external, are required to evaluate the performance of the clustering algorithm. The silhouette score for each cluster is computed as the internal evaluation metric for both modes of the proposed model. The external evaluation requires a gold standard data for which a data set containing 20 ambiguous words and 100 sentences for each target word is developed.According to the obtained results of the internal evaluation, the Sentence-based mode has higher density of clusters than the Context-based mode, and the difference between them is statistically significant. According to the V- and F-measure evaluation metrics in the external evaluation, the Context-based mode has obtained higher performance against the baselines with statistically significant difference.

The movement of  “Cognitive Semantics” appears against the independence of the syntax from lexical domain and calls into question the semantics being peripheral as it is mentioned in Chomsky's (1982) “Generative theory”. In this regard, Fillmore's (1982) “Frame Semantics theory” is one of the major achievements of cognitive semantics. In this theory Fillmore states that understanding the meanings of words is possible through frames that each word calls for. He also maintains that “Semantic Frames” display sections of an event used for connecting a group of words to a set of meanings and the meanings of words are perceived within a system of knowledge arising from the human's social and cultural experience. In fact, the difference between Frame Semantics with other lexico semantic hypotheses is its emphasis on the background knowledge based on which the meaning of words can be interpreted (Fillmore & Atkins, 1992, 1994, 2000; Fillmore & Baker, 2010). Also, the “FrameNet Project” is a Corpus-based study based on Fillmore's theory (Fillmore, et al., 2003). It is a lexicographic research project housed and administered at the International Computer Science Institute in Berkeley, California.  In this system, the meaning of words can be understood on the basis of semantic frames that are mental concepts. On the other hand, verbs are present as the most important element in most of the situations of an event and play significant role in the interpretation of the meaning; thus, working on it is a useful idea.
        The purpose of this study is to present a quantitative and corpus-based analysis in order to recognize the semantic domains and frames of the verb “Goftan” (to Tell) in Persian and sketch its main frame within the framework of Frame Semantics theory and the fact-finding model of FrameNet lexicography. In fact, it attempts to investigate what Fillmore and his colleagues did in English FrameNet and others in different languages as well as in the domain of Frame Semantics to be considered in Persian too. So, the main questions of this study are:
What are the semantic domains and frames of the verb “Goftan” (to Tell) in Persian based on the major concepts of Frame Semantics theory and FrameNet?
 Is it possible to design the main semantic frame of the verb “Goftan” (to Tell) in Persian?

To answer these questions, the authors investigate the background of studies and introduce Frame semantics theory and FrameNet lexicography. Then, they tried to be familiar with the verb “Goftan” and it’s frequency in Persian based on Persian Language Database (PLDB). To this end, 44 semantic domains and frames of the verb “Goftan” were recognized using encyclopedias, Persian contemporary corpora and FrameNet database. Then, the concept of “bar zabân Ɂâvardan va bayân kardan” (to express) was determined as the main meaning and prototype of this verb and for the first time its “Radial category” the authors sketched. Finally, the main frame of “Goftan” was created using principles and concepts proposed in Fillmore's Frame Semantics theory

Lexical Functional Gramma (LFG) is a non-transformational generative grammar, which excludes concepts such as deep structure, surface structure and transformation. Rather than shifting a deep structure into a surface form through transformations, LFG maintains the idea that several structures exist in parallel levels. The two main structures in the LFG are constituent structure and function structure, which are abbreviated as c-structure and f- structure, respectively. LFG is also comprised of other structures including argument structure, semantic structure, and information structure. The present paper mainly focuses on the f-structure to prove the capability of LFG in explaining some linguistic phenomena and characteristics of Persian language such as passivization, non-configurationality, and topicalization. Certain Persian structures such as simple and compound sentences, complement clauses, and genitive structures are studied and, following 10 grammatical roles are introduced for Persian Lodrup’s (2011) model, which are classified as argument vs. non-argument and discourse vs. non-discourse.

In this paper, the role of Part-of-Speech (POS) tagging for parsing in automatic processing of the Persian language will be studied. To this end, the impact of the quality of POS tagging as well as the impact of the quantity of information available in the POS tags on parsing are studied. To reach the goals, three parsing scenarios are proposed and compared. In this study, various evaluation metrics are used to show the impacts from different points of views. The experimental results show that the quality of the POS tagger and the quantity of the information available in the POS tags have a direct effect on parsing performance. The high quality of the POS tags causes error reduction in parsing and also it increases parsing performance. Moreover, lack of morpho-syntactic information in the POS tags has a high negative impact on parsing performance. This impact is more pronounced than the impact of POS tagger performance.