No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Tree adjunct grammar
Abstract
In this paper, a tree generating system called a tree adjunct grammar is described and its formal properties are studied relating them to the tree generating systems of Brainerd (Information and Control14 (1969), 217–231) and Rounds (Mathematical Systems Theory 4 (1970), 257–287) and to the recognizable sets and local sets discussed by Thatcher (Journal of Computer and System Sciences1 (1967), 317–322; 4 (1970), 339–367) and Rounds. Linguistic relevance of these systems has been briefly discussed also.
... Tree Adjoining Grammars (TAGs) was proposed by Joshi et al [6] and later extended by Vijay-Shankar and Joshi [7] for linguistic modelling of natural languages. They form a non-Chomskian collection of mildly context sensitive languages. ...
... According to Joshi and Schabes [6,12], TAG is formally defined as a quintuple as per Formulation (1.2). TAGs have multilevel trees as their production rules, which can capture full phrase structure of a sentence hierarchically, using a single rule. ...
... The grammar generates parse trees or derived trees like in Figure 1.2, through two combination operations between any two elementary trees; namely substitution and adjunction. Here we will verbally and pictorially try and understand the effects of these combination operations as detailed in [6,7,14,18,19] Substitutions can only be done on nodes that are marked for it; when marked, the node will have to be compulsorily replaced by an initial tree with the same root node label Substitution is the basic context free insertion of an initial tree into another initial or auxiliary tree node. Essentially substitution replaces a node in the parent tree with an initial tree. ...
Tree adjoining grammars (TAGs) are psycholinguistic formalisms proposed by Prof. Arvind Joshi from University of Pennsylvania. These formalism are special in the way they generate tree languages with complex deep structure. They fall under the class of mildly context sensitive grammars. They were originally used to model and mimic natural language syntax for purposes such as machine translation, dependency parsing and understanding deep structure. Unlike context free grammars on which most of the computer programming languages are based on, tree adjoining grammars are aware of contextual information and capture dependencies between lexicons (words) that are far away in the original sentence. They can also separate recurring structures such as repeating adjectives or nouns into single tree productions that have a recurring root and a foot node; such recurring trees are called auxiliary trees and are literally inserted into other tree forms by exploding any matching node into root and foot node. This process of insertion of trees is called an adjunction; it is this particular operation used to combine trees, that makes it, mildly context sensitive. In this day and age of machine learning and deep learning, tree adjoining grammars remain relevant as they contribute to our understanding of linguistic and other deep structure cognition. Parsing such a grammar that generates trees rather than strings is quite complex to imagine. However considerable work has been done in this area and multiple parsing algorithms of varying efficiencies have been proposed; including machine learned models. Most of these algorithms also have robust implementations. The parsing process generates another tree structure called a derivation. This structure is of vital importance when analysing phrase structures. Since natural languages are finitely ambiguous, tree adjoining grammars are ideal to model them. But despite best efforts and numerous algorithms tree adjoining grammar parsing is a hard problem with the worst case complexity of . In reality most conventional TAG parsing algorithms run in and give the most probable parse, but trying to extract multiple ambiguous parses for a given phrase degrades the performance of the traditional TAG parsing algorithms toward worst case runtime; specially for longer sentences. Ambiguity in TAGs are not as well understood as in CFGs due to their complex derivation process; hence one of the ways to understand it is to find a finite set of ambiguous derivations for the given phrase structure. We introduce and extend the definition of a containing formalism called GATAGs on which a genetic algorithm can be deployed in order to find ambiguous derivation structures for longer sentences. We shall formalise the genotypes, phenotypes and the fitness functions for the entailing genetic algorithm, exploring various avenues for their efficient computations. Our main objective here is to explore the possibility of random derivations evolving to good derivations through a natural selection. Further we shall also investigate the properties and some behavioural characterisation of GATAGs in detail concluding on a minimal implementation of the formalism.
... CCG is a revised version of the CG with a mildly context-sensitivity: the syntactic structure of natural languages cannot be adequately described by contextfree grammars (Joshi, 1985). Weakly equivalent to the tree adjoining grammars (TAG) (Joshi, Levy, and Takahashi, 1975;Joshi and Schabes, 1991;Vijay-Shanker and Weir, 1994), it can generate a string beyond context-free languages such as a n b n c n . The different generative capacity of CCG based on rule restrictions has been well-studied Satta, 2010, 2015). ...
... combinatory categorial language. The similar idea of "multi-" as in multiset-CCG has been presented for Tree-adjoining grammars (Joshi et al., 1975) as multicomponent tree-adjoining grammars (MCTAG) (Weir, 1990). Accordingly, MCTAG for the word order variation of Korean was also proposed (Rambow and Lee, 1994). ...
This study discusses a role of functional morphemes in Korean categorial grammars, providing the reviews of various types of Korean categorial grammars that have never been conducted so far, notwithstanding many previous studies on them. Previous work has presented different morphological segmentation because of Korean’s agglutinative characteristics, implying that Korean words may contain a different segmentation sequence of morphemes. We focus on functional morphemes in Korean categorial grammars, which have been explored in different ways by previous work. We present detailed analyses for postpositions and verbal endings in categorial grammars, insisting that the functional morphemes in Korean should be treated as part of a word, with the result that their categories do not require to be assigned individually in a syntactic level, and also that it would be more efficient to assign the syntactic categories on the fully inflected lexical word derived by the lexical rule of the morphological processes in the lexicon.
... This resource should be used for syntax-semantic analysis and other different NLP applications. We chose the formalism of tree-adjoining grammar (TAG) (Joshi et al. 1975) which has a rich expressivity. Moreover, this formalism allows the integration of semantic information straightforwardly. ...
... Each formalism has its own characteristics, its strengths but also its weaknesses. Four formalisms of unification grammars with large-coverage retained our attention: the Tree-adjoining Grammar (TAG) (Joshi et al. 1975) and its extensions (Schabes and Joshi 1990), the Lexical Functional Grammar (LFG) (Bresnan and Kaplan 1982), the Generalized Phrase Structure Grammar (GPSG) (Gerald et al. 1985) and the Headdriven Phrase Structure Grammar (HPSG) (Pollard and Sag 1994). Sentence in LFG grammar is described by means of two distinct levels: the representation of grammatical functions (f-structure) and the structure of syntactic constituents (c-structure). ...
Arabic presents many challenges for automatic processing. Although several research studies have addressed some issues, electronic resources for processing Arabic remain relatively rare or not widely available. In this paper, we propose a Tree-adjoining grammar with a syntax-semantic interface. It is applied to the modern standard Arabic, but it can be easily adapted to other languages. This grammar named “ArabTAG V2.0” (Arabic Tree Adjoining Grammar) is semi-automatically generated by means of an abstract representation called meta-grammar. To ensure its development, ArabTAG V2.0 benefits from a grammar testing environment that uses a corpus of phenomena. Further experiments were performed to check the coverage of this grammar as well as the syntax-semantic analysis. The results showed that ArabTAG V2.0 can cover the majority of syntactical structures and different linguistic phenomena with a precision rate of 88.76%. Moreover, we were able to semantically analyze sentences and build their semantic representations with a precision rate of about 95.63%.
... Sin embargo, la mayor parte de las gramáticas de restricciones fueron propuestas en los años 70 y 80 del siglo pasado como teorías alternativas a los modelos transformacionales de Chomsky (1965). Es el caso de las Tree Adjoining Grammars (TAG) (Joshi et al. 1975), de la Lexical Functional Grammar (LFG) (Kaplan y Bresnan 1982), de la Generalized Phrase Structure Grammar (GPSG) (Gazdar et al. 1985), de la Optimality Theory (Prince y Smolensky 1993) ...
... Parten de la idea de que la información lingüística debe ser descodificada de manera inmediata e independientemente de la forma final que presente una construcción. Esta idea, que está ya presente en las Tree-Adjoining Grammars (Joshi et al. 1975) y en las HPSG (Pollard y Sag 1994), permite a las Gramáticas de Propiedades poder describir cualquier input con independencia del nivel de gramaticalidad que presente. ...
Resumen En este trabajo, introducimos las Gramáticas de Propiedades, una teoría gramatical no generativa que define un formalismo basado exclusivamente en el concepto de restricción. Desde el punto de vista lingüístico, toda la información sintáctica es representada a través de restricciones. Desde el punto de vista computacional, se propone un parsing basado en la satisfacción de restricciones. Una de las principales ventajas de las Gramáticas de Propiedades es que tienen capacidad para analizar cualquier tipo de input, sea o no gramatical, y por ello resultan un formalismo útil para proponer un modelo sintáctico que tolere distintos niveles de gramaticalidad. Palabras clave: Restricciones, niveles de gramaticalidad, análisis sintáctico.
---------------------------------------
Abstract In this paper, we introduce Property Grammars, a non-generative grammar theory that defines a formalism based on the constraint concept. From a linguistic point of view, all the syntactic information is represented through restrictions and from the computational point of view, a parsing based on constraint satisfaction is proposed. One of the main advantages of Property Grammars is that they can parse any type of input, grammatical or ungrammatical, and, therefore, they are a useful tool to propose a syntactic model that tolerates different levels of grammaticality.
... It is known that context-free grammars are characterized by the non-associative Lambek calculus as categorial grammars studied in 1950s [7]. On the other hand, tree-adjoining grammars [5] and other mildly context-sensitive grammars have no such a characterization as a sequent calculus or a generalization of it. As a characterization of tree-adjoining languages, combinatory categorial grammars (CCG) ...
... The cross-serial dependency, a well-known counterexample in linguistics, is phenomenon cannot be captured in the context-free grammars but can be in treeadjoining grammars [5] (See more details in [6] for example). We show that this phenomenon can be indeed captured by HLG. ...
While context-free grammars are characterized by a simple proof-theoretic grammatical formalism namely categorial grammar and its logic the Lambek calculus, no such characterizations were known for tree-adjoining grammars, and even for any mildly context-sensitive languages classes in the last forty years despite some efforts. We settle this problem in this paper. On the basis of the existing fragment of the Lambek-Grishin calculus which captures tree-adjoining languages, we present a logic called HLG: a proof-theoretic characterization of tree-adjoining languages based on the Lambek-Grishin calculus restricted to Hyperedge-replacement grammar with rank two studied by Moot. HLG is defined in display calculus with cut-admissibility. Several new techniques are introduced for the proofs, such as purely structural connectives, usefulness, and a graph-theoretic argument on proof nets for HLG.
... Several formalisms in the scope of mild context-sensitivity have been proposed. Among these are tree adjoining grammar (TAG) (Joshi et al., 1975), head grammar (HG) (Pollard, 1984), linear indexed grammars (LIG) (Gazdar, 1988) and linear context-free rewriting systems (LCFRS) (Vijay-Shanker et al., 1987) as well as certain variants of combinatory categorial grammar (CCG) (Steedman, 1989(Steedman, , 1996(Steedman, , 2000. ...
Syntactic parsing is essential in natural-language processing, with constituent structure being one widely used description of syntax. Traditional views of constituency demand that constituents consist of adjacent words, but this poses challenges in analysing syntax with non-local dependencies, common in languages like German. Therefore, in a number of treebanks like NeGra and TIGER for German and DPTB for English, long-range dependencies are represented by crossing edges. Various grammar formalisms have been used to describe discontinuous trees - often with high time complexities for parsing. Transition-based parsing aims at reducing this factor by eliminating the need for an explicit grammar. Instead, neural networks are trained to produce trees given raw text input using supervised learning on large annotated corpora. An elegant proposal for a stack-free transition-based parser developed by Coavoux and Cohen (2019) successfully allows for the derivation of any discontinuous constituent tree over a sentence in worst-case quadratic time. The purpose of this work is to explore the introduction of supertag information into transition-based discontinuous constituent parsing. In lexicalised grammar formalisms like CCG (Steedman, 1989) informative categories are assigned to the words in a sentence and act as the building blocks for composing the sentence's syntax. These supertags indicate a word's structural role and syntactic relationship with surrounding items. The study examines incorporating supertag information by using a dedicated supertagger as additional input for a neural parser (pipeline) and by jointly training a neural model for both parsing and supertagging (multi-task). In addition to CCG, several other frameworks (LTAG-spinal, LCFRS) and sequence labelling tasks (chunking, dependency parsing) will be compared in terms of their suitability as auxiliary tasks for parsing.
... In a formal language, the set of symbol strings are defined by a grammar, a set of rules for taking symbol strings and constructing larger ones, loosely speaking. Grammars define compositional objects, whereby objects are composed of parts, which in turn, are composed of parts, etc. Grammars can also be used for constructing other compositional objects besides strings, for example graphs and trees (e.g., using graph grammars [68] or tree grammars [50]), which could be used as well. In classical logics, these symbol strings are used to represent binary functions, allowing for a compressed (and finite, under certain assumptions) representation of them. ...
There are two general conceptions on the relationship between probability and logic. In the first, these systems are viewed as complementary—having offsetting strengths and weaknesses—and there exists a fusion of the two that creates a reasoning system that improves upon each. In the second, probability is viewed as an instance of logic, given some sufficiently broad formulation of it, and it is this that should inform the development of more general reasoning systems. These two conceptions are in conflict with each other, where the root issue of contention is the proper abstraction of the concept of logical consequence. In this work, we put forth a proposal on this abstraction based on an extension of the subset relation through the use of projections, which in turn, allows for the formalization of valid inferences to more general settings. Our proposal results in a formalism that encompasses probability and classical logic, and importantly, does so with minimal machinery. This formalism makes assertions about the relationship between these two systems that are explicit, and suggests a path forward in the development of alternatives to them.
... They are introduced to correctly represent the scope of the case marker. Park (2006) considered case markers as independent elements within the formalism of Tree adjoining grammars (Joshi, Levy, and Takahashi, 1975). Therefore, he defined case markers as an auxiliary tree to be adjoined to a noun phrase. ...
This paper describes word segmentation granularity in Korean language processing. From a word separated by blank space, which is termed an eojeol, to a sequence of morphemes in Korean, there are multiple possible levels of word segmentation granularity in Korean. For specific language processing and corpus annotation tasks, several different granularity levels have been proposed and utilized, because the agglutinative languages including Korean language have a one-to-one mapping between functional morpheme and syntactic category. Thus, we analyze these different granularity levels, presenting the examples of Korean language processing systems for future reference. Interestingly, the granularity by separating only functional morphemes including case markers and verbal endings, and keeping other suffixes for morphological derivation results in the optimal performance for phrase structure parsing. This contradicts previous best practices for Korean language processing, which has been the de facto standard for various applications that require separating all morphemes.
... According to different theory-specific proposals, the syntactic structure of a sentence can therefore be analyzed assuming either binary branching (e.g., Chomsky, 1988Chomsky, , 1995 or n-ary branching (Pollard & Sag, 1994), which results in rather different-looking structural descriptions (see Figure 2A and 2C). Furthermore, a particular linguistic theory may start from the assumption that well-formed sentences have a so-called deep structure satisfying some structural requirements and a surface structure that complies with positional relationships (as in the early versions of the principles and parameters theory within the generative grammar tradition; Chomsky, 1988), or alternatively, that it should provide the complete derivation of a structure in its description (e.g., minimalism; Chomsky, 1995) at all times (see Figure 2A) or, conversely, that it should allow for the combination of "preassembled" structural elements already stored in memory (e.g., tree adjoining grammars; Joshi & Schabes, 1997;Joshi, Levy, & Takahashi, 1975; see Figure 2B). ...
The capacity for language is a defining property of our species, yet despite decades of research, evidence on its neural basis is still mixed and a generalized consensus is difficult to achieve. We suggest that this is partly caused by researchers defining “language” in different ways, with focus on a wide range of phenomena, properties, and levels of investigation. Accordingly, there is very little agreement among cognitive neuroscientists of language on the operationalization of fundamental concepts to be investigated in neuroscientific experiments. Here, we review chains of derivation in the cognitive neuroscience of language, focusing on how the hypothesis under consideration is defined by a combination of theoretical and methodological assumptions. We first attempt to disentangle the complex relationship between linguistics, psychology, and neuroscience in the field. Next, we focus on how conclusions that can be drawn from any experiment are inherently constrained by auxiliary assumptions, both theoretical and methodological, on which the validity of conclusions drawn rests. These issues are discussed in the context of classical experimental manipulations as well as study designs that employ novel approaches such as naturalistic stimuli and computational modeling. We conclude by proposing that a highly interdisciplinary field such as the cognitive neuroscience of language requires researchers to form explicit statements concerning the theoretical definitions, methodological choices, and other constraining factors involved in their work.
... Earlier, NLP Programmer need to rely on grammar rules where they analyze structure of source language and convert it into Parse derivation. As part of our research on Tree Adjoining Grammar, we have developed Multithreaded TAG Parser which is implementation of the 'Early-Type Parsing Algorithm' originally proposed by Arvind Joshi [1]. Basically, TAG Parser is a software program that analyze the source in order to determine its grammatical structure with respect to a given formal grammar. ...
The Revolution of the Artificial Intelligence (AI) has started when machines could decipher enigmatic symbols concealed within messages. Subsequently, with the progress of Natural Language Processing (NLP), machines attained the capacity to understand and comprehend human language. Tree Adjoining Grammar (TAG) has become powerful grammatical formalism for processing Large-scale Grammar. However, TAG mostly rely on Grammar which is created by Languages expert and due to structural ambiguity in Natural Languages computation complexity of TAG is very high o(n^6). We observed that rules-based approach has many serious flaws, firstly, language evolves with time and it is impossible to create grammar which is extensive enough to represent every structure of language in real world. Secondly, it takes too much time and language resources to develop a practical solution. These difficulties motivated us to explore an alternative approach instead of completely rely on the rule-based method. In this paper, we proposed a Statistical Parsing algorithm for Natural Languages (NL) using TAG formalism where Parser makes crucial use of data driven model for identifying Syntactic dependencies of complex structure. We observed that using probabilistic model along with limited training data can significantly improve both the quality and performance of TAG Parser. We also demonstrate that the newer parser outperforms previous rule-based parser on given sample corpus. Our experiment for many Indian Languages, also provides further support for the claim that above mentioned approach might be an awaiting solution for problem that require rich structural analysis of corpus and constructing syntactic dependencies of any Natural Language without much depending on manual process of creating grammar for same. Finally, we present result of our on-going research where probability model will be applying to appropriate selection of adjunction of any given node of elementary trees and state chart representations are shared across derivation.
... In this paper, we use probabilistic tree-substitution grammars as our model of lexical argument structure. A tree-substitution grammar formalizes the lexicon as an inventory of stored tree fragments, such as those shown in Figure 1 (Bod 1998;Joshi and Levy 1975;Scha 1990Scha , 1992. This figure shows the inventory of elementary trees that we will use as examples below. ...
We present a learnability analysis of the argument-modifier distinction, asking whether there is information in the distribution of English constituents that could allow learners to identify which constituents are arguments and which are modifiers. We first develop a general description of some of the ways in which arguments and modifiers differ in distribution. We then identify two models from the literature that can capture these differences, which we call the argument-only model and the argument-modifier model. We employ these models using a common learning framework based on two simplicity biases which tradeoff against one another. The first bias favors a small lexicon with highly reusable lexical items, and the second, opposing, bias favors simple derivations of individual forms – those using small numbers of lexical items. Our first empirical study shows that the argument-modifier model is able to recover the argument-modifier status of many individual constituents when evaluated against a gold standard. This provides evidence in favor of our general account of the distributional differences between arguments and modifiers. It also suggests a kind of lower bound on the amount of information that a suitably equipped learner could use to identify which phrases are arguments or modifiers. We then present a series of analyses investigating how and why the argument-modifier model is able to recover the argument-modifier status of some constituents. In particular, we show that the argumentmodifier model is able to provide a simpler description of the input corpus than the argument-only model, both in terms of lexicon size, and in terms of the complexity of individual derivations. Intuitively, the argument-modifier model is able to do this because it is able to ignore spurious modifier structure when learning the lexicon. These analyses further support our general account of the differences between arguments and modifiers, as well as our simplicity-based approach to learning.
... The degree of grammaticality is the theoretical value resulting from the satisfaction and violation of the linguistic rules that characterize the linguistic knowledge in any linguistic domain (Keller [8], Blache [20], Chomsky [21], Joshi et al. [29]). In contrast, the degree of acceptability is essentially a subjective evaluation (Sorace and Keller [23], Schutze [30]). ...
This paper introduces a new grammatical framework, Fuzzy Property Grammars (FPGr). This is a model based on Property Grammars and Fuzzy Natural Logic. Such grammatical framework is constraint-based and provides a new way to formally characterize gradience by representing grammaticality degrees regarding linguistic competence (without involving speakers judgments). The paper provides a formal-logical characterization of FPGr. A test of the framework is presented by implementing an FPGr for Spanish. FPGr is a formal theory that may serve linguists, computing scientists, and mathematicians since it can capture infinite grammatical structures within the variability of a language.
... Much of the foundational work on large-scale, broad-coverage grammar engineering was carried out in the 1980s and 1990s by researchers trained in a variety of computational linguistics formalisms, in particular Tree-Adjoining Grammar (TAG) [74], Lexical-Functional Grammar (LFG) [75], Combinatory Categorial Grammar (CCG) [76] and Head-Driven Phrase Structure Grammar (HPSG) [77]. Since this period, many research papers have been published that explicitly tackle issues related to large-scale grammar engineering or that present tools for supporting the grammar engineering process. ...
Since its inception in the mid-eighties, the field of construction grammar has been steadily growing and constructionist approaches to language have by now become a mainstream paradigm for linguistic research. While the construction grammar community has traditionally focused on theoretical, experimental and corpus-based research, the importance of computational methodologies is now rapidly increasing. This movement has led to the establishment of a number of exploratory computational construction grammar formalisms, which facilitate the implementation of construction grammars, as well as their use for language processing purposes. Yet, implementing large grammars using these formalisms still remains a challenging task, partly due to a lack of powerful and user-friendly tools for computational construction grammar engineering. In order to overcome this obstacle, this paper introduces the FCG Editor, a dedicated and innovative integrated development environment for the Fluid Construction Grammar formalism. Offering a straightforward installation and a user-friendly, interactive interface, the FCG Editor is an accessible, yet powerful tool for construction grammarians who wish to operationalise their construction grammar insights and analyses in order to computationally verify them, corroborate them with corpus data, or integrate them in language technology applications.
... Williams calls this notion the derivational clock (or 'F-clock'). To illustrate, consider that-clause embedding in (79) Under the LEC, embedding is a substitution operation (though Williams does not explicitly call it such), analogous to Chomsky's (1955Chomsky's ( , 1957 theory of generalized transformations (for a reprise, see also Chomsky 1995b:173-174) and to substitution in Tree Adjoining Grammar (Joshi et al. 1975;Kroch and Joshi 1985). On this proposal, the WC follows from the strict cycle. ...
This paper argues that case assignment is impossible in configurations that parallel generalized improper-movement configurations. Thus, like improper movement, there is “improper case.” The empirical motivation comes from (i) the interaction between case and movement and (ii) crossclausal case assignment in Finnish. I propose that improper case is ruled out by the Ban on Improper Case : a DP in [Spec, XP] cannot establish a dependent-case relationship with a lower DP across YP if Y is higher than X in the functional sequence. I show that this constraint falls under a strong version of the Williams Cycle (Williams 1974, 2003, 2013; van Riemsdijk and Williams 1981) and is derived under Williams’s (2003, 2013) analysis of embedding.
... Research in computational linguistics has demonstrated that quite different grammar formalisms, such as tree-adjoining grammar (Joshi et al., 1975), multiple context-free grammar (Seki et al., 1991), range concatenation grammar (Boullier, 2005), and minimalist grammar (Stabler, 1997;Stabler and Keenan, 2003) converge toward universal description models (Joshi et al., 1990;Michaelis, 2001;Stabler, 2011a;Kuhlmann et al., 2015). Minimalist grammar has been developed by Stabler (1997) to mathematically codify Chomsky's Minimalist Program (Chomsky, 1995) in the generative grammar framework. ...
Cognitive agents that act independently and solve problems in their environment on behalf of a user are referred to as autonomous. In order to increase the degree of autonomy, advanced cognitive architectures also contain higher-level psychological modules with which needs and motives of the agent are also taken into account and with which the behavior of the agent can be controlled. Regardless of the level of autonomy, successful behavior is based on interacting with the environment and being able to communicate with other agents or users. The agent can use these skills to learn a truthful knowledge model of the environment and thus predict the consequences of its own actions. For this purpose, the symbolic information received during the interaction and communication must be converted into representational data structures so that they can be stored in the knowledge model, processed logically and retrieved from there. Here, we firstly outline a grammar-based transformation mechanism that unifies the description of physical interaction and linguistic communication and on which the language acquisition is based. Specifically, we use minimalist grammar (MG) for this aim, which is a recent computational implementation of generative linguistics. In order to develop proper cognitive information and communication technologies, we are using utterance meaning transducers (UMT) that are based on semantic parsers and a mental lexicon , comprising syntactic and semantic features of the language under consideration. This lexicon must be acquired by a cognitive agent during interaction with its users. To this aim we outline a reinforcement learning algorithm for the acquisition of syntax and semantics of English utterances. English declarative sentences are presented to the agent by a teacher in form of utterance meaning pairs (UMP) where the meanings are encoded as formulas of predicate logic. Since MG codifies universal linguistic competence through inference rules, thereby separating innate linguistic knowledge from the contingently acquired lexicon, our approach unifies generative grammar and reinforcement learning, hence potentially resolving the still pending Chomsky-Skinner controversy.
... The translation rules used by the system are motivated by linguistic considerations. For this, a tree generating system called tree adjoining grammar (TAG) was studied which was introduced by Joshi et al. (1975) and Joshi (1985) [21]. In the TAG algorithm, insertion of the absolute and unbounded quantity of stuff in the existing tree structure is granted by using adjunction operation [22]. ...
... The various proposals about grammars (e.g., Stabler, 2004) and experiments on the learning of artificial grammars (Westphal-Fitch et al., 2018) have converged on similar results. Natural language has mildly context-sensitive rules of the sort in various systems, including tree-adjoining grammars (Joshi et al., 1975) and combinatory categorial grammars (Steedman, 2019). They can be parsed in a time proportional to a polynomial of the number of words in a sentence. ...
This article presents a theory of recursion in thinking and language. In the logic of computability, a function maps one or more sets to another, and it can have a recursive definition that is semi-circular, i.e., referring in part to the function itself. Any function that is computable – and many are not – can be computed in an infinite number of distinct programs. Some of these programs are semi-circular too, but they needn’t be, because repeated loops of instructions can compute any recursive function. Our theory aims to explain how naive individuals devise informal programs in natural language, and is itself implemented in a computer program that creates programs. Participants in our experiments spontaneously simulate loops of instructions in kinematic mental models. They rely on such loops to compute recursive functions for rearranging the order of cars in trains on a track with a siding. Kolmogorov complexity predicts the relative difficulty of abducing such programs – for easy rearrangements, such as reversing the order of the cars, to difficult ones, such as splitting a train in two and interleaving the two resulting halves (equivalent to a faro shuffle). This rearrangement uses both the siding and part of the track as working memories, shuffling cars between them, and so it relies on the power of a linear-bounded computer. Linguistic evidence implies that this power is more than necessary to compose the meanings of sentences in natural language from those of their grammatical constituents.
... And here we see a real difference between the frameworks: TAG (Joshi et al., 1975) is a framework in which structure is assembled: the basic operations are substitution and adjunction. The lexicon consists of ready-made building blocks that are combined to yield the trees we want to have in the end. ...
This paper compares a recent TAG-based analysis of complex predicates in Hindi/Urdu with its HPSG analog. It points out that TAG combines actual structure while HPSG (and Categorial Grammar and other valence-based frameworks) specify valence of lexical items and hence potential structure. This makes it possible to have light verbs decide which arguments of embedded heads get realized, something that is not possible in TAG. TAG has to retreat to disjunctions instead. While this allows straight-forward analyses of active/passive alternations based on the light verb in valence-based frameworks, such an option does not exist for TAG and it has to be assumed that preverbs come with different sets of arguments.
... A TAG is essentially a rewriting system working on predefined elementary trees that can be augmented and combined with one another at the frontier or "counter-cyclically"; these two cases correspond to the (generalised) operations substitution [15,16] and adjunction [17,18], respectively. Trees may be either elementary or derived: the latter are obtained by means of applying composition operations to the former. ...
Contemporary generative grammar assumes that syntactic structure is best described in terms of sets, and that locality conditions, as well as cross-linguistic variation, is determined at the level of designated functional heads. Syntactic operations (merge, MERGE, etc.) build a structure by deriving sets from lexical atoms and recursively (and monotonically) yielding sets of sets. Additional restrictions over the format of structural descriptions limit the number of elements involved in each operation to two at each derivational step, a head and a non-head. In this paper, we will explore an alternative direction for minimalist inquiry based on previous work, e.g., Frank (2002, 2006), albeit under novel assumptions. We propose a view of syntactic structure as a specification of relations in graphs, which correspond to the extended projection of lexical heads; these are elementary trees in Tree Adjoining Grammars. We present empirical motivation for a lexicalised approach to structure building, where the units of the grammar are elementary trees. Our proposal will be based on cross-linguistic evidence; we will consider the structure of elementary trees in Spanish, English and German. We will also explore the consequences of assuming that nodes in elementary trees are addresses for purposes of tree composition operations, substitution and adjunction.
... Bajo el término genérico de gramáticas de restricciones encontramos modelos como los siguientes: Functional Unification Grammar (Kay 1979), Lexical Functional Grammar (LFG) (Kaplan y Bresnan 1982), Categorial Grammar (Buszkowski et al. 1988), Head-Driven Phrase-Structure Grammar (HPSG) (Pollard y Sag 1994), Tree Adjoining Grammar (TAG) (Joshi et al. 1975), Optimality Theory (Prince y Smolensky 1993), etc. ...
RESTRICCIONES Y GRADIENCE EN EL PROCESAMIENTO DEL LENGUAJE NATURAL M. DOLORES JIMÉNEZ-LÓPEZ ADRIÀ TORRENS-URRUTIA Universitat Rovira i Virgili, Tarragona RESUMEN En este trabajo, analizamos el uso del concepto de restricción en lingüística formal con el objetivo de poner de relieve las ventajas que los modelos basados en restricciones presentan para el desarrollo de sistemas de procesamiento del lenguaje natural. En general, se reconoce que el concepto de restricción es de gran utilidad tanto en la descripción como en el procesamiento del lenguaje natural. Son muchos los modelos gramaticales que incorporan esta noción. En este trabajo, tras analizar los distintos tipos de restricciones que aparecen en los diferentes modelos, presentamos dos formalismos que consideramos de especial interés para el procesamiento del lenguaje natural: las gramáticas de propiedades y las womb grammar. ABSTRACT In this paper, we analyze the use of constraints in formal linguistics with the aim of highlighting the advantages of constraints based models for the development of natural language processing systems. In general, it is recognized that the concept of restriction is useful both in the description and processing of natural language. There are many grammatical models that incorporate this notion. In this paper, after analyzing the different types of restrictions that appear in different models, we present two formalisms that we consider of particular interest for the processing of natural language: property grammars and womb grammars.
... Bajo el término genérico de gramáticas de restricciones encontramos modelos como los siguientes: Functional Unification Grammar (Kay 1979), Lexical Functional Grammar (LFG) (Kaplan y Bresnan 1982), Categorial Grammar (Buszkowski et al. 1988), Head-Driven Phrase-Structure Grammar (HPSG) (Pollard y Sag 1994), Tree Adjoining Grammar (TAG) (Joshi et al. 1975), Optimality Theory (Prince y Smolensky 1993), etc. ...
... Tree-adjoining grammar is a formalism introduced by Joshi et al. [1] for describing linguistic structure of natural languages. It is a tree generating and more powerful formalism as opposed to string generating formalisms such as a context-free grammar. ...
We present an initial study into the representation of tree-adjoining grammar formalism for parsing Manipuri language. Being a low resource and computationally less researched language, it is difficult to achieve a natural language parser for Manipuri. Treebanks, which are the main requirement for inducing data-driven parsers, are not available for Manipuri. In this paper, we present an extensive analysis of the Manipuri language structure and formulate a lexicalized tree-adjoining grammar. A generalized structure of Manipuri phrases, clauses and the structure of basic and derived sentences have been presented. The sentence types covered in our analysis are that of simple, compound and complex sentences. Using the tree-adjoining grammar we have formulated, one can implement a Manipuri parser whose results can be of immense help in creating a Treebank for Manipuri.
... H EAD-DRIVEN Phrase Structure Grammar (HPSG) [1], Generative Grammar [2], Tree Adjoining Grammar (TAG) [3], Lexical Function Grammar (LFG) [4], and Combinatory Categorial Grammar (CCG) [5] are known as the most sophisticated grammar frameworks for syntactic phrase structures. HPSG, a lexicalized grammar, explains linguistic phenomena elegantly with a small number of grammar rules (phrase structure rules and lexical rules) and a number of complex lexical entries. ...
Constituent and dependency parsing, the two classic forms of syntactic parsing, have been found to benefit from joint training and decoding under a uniform formalism, Head-driven Phrase Structure Grammar (HPSG). However, decoding this unified grammar has a higher time complexity () than decoding either form individually () since more factors have to be considered during decoding. We thus propose an improved head scorer that helps achieve a novel performance-preserved parser in O() time complexity. Furthermore, on the basis of this proposed practical HPSG parser, we investigated the strengths of HPSG-based parsing and explored the general method of training an HPSG-based parser from only a constituent or dependency annotations in a multilingual scenario. We thus present a more effective, more in-depth, and general work on HPSG parsing.
... Tree Adjoining Grammars (TAGs) were introduced by Joshi et al. [21]. Even though they are not part of the traditional Chomsky hierarchy, they are considered as a class of mildly context sensitive grammars [20]. ...
Tree Adjoining Grammars (TAGs) are very useful psycholinguistic formalisms for syntax and dependency analysis of phrase structures. Since natural languages are finitely ambiguous, TAGs are ideal to model them being mildly context sensitive. But these grammars are very hard to parse as they have a worst case complexity of O(n6). In reality most conventional TAG parsing algorithms run in O(n3) and give the most probable parse, but trying to extract multiple ambiguous parses for a given phrase degrades the performance of the traditional TAG parsing algorithms toward worst case runtime, especially for longer sentences. Ambiguity in TAGs are not as well understood as in CFGs due to their complex derivation process; hence one of the ways to understand it is to find a finite set of ambiguous derivations for the given phrase structure. In this article we extend the definition of the containing formalism introduced as GATAGs on which a genetic algorithm can be deployed in order to find ambiguous derivation structures for longer sentences. We shall formalise the genotypes, phenotypes and the fitness functions for the entailing genetic algorithm, exploring various avenues for their efficient computations. Our main objective here is to explore the possibility of random derivations evolving to good derivations though a natural selection.
... 15 In terms of the Chomsky hierarchy, this means that it falls in a class between the context-free and context-sensitive classes, but close enough to the context-free class such that parsing remains tractable. There is much debate about what the right formalism is: two contenders are Tree Adjoining Grammar (TAG;Joshi et al., 1975) and Combinatory Category Grammar (CCG; Steedman, 1996). ...
NLP is deeply intertwined with the formal study of language, both conceptually and historically. Arguably, this connection goes all the way back to Chomsky's Syntactic Structures in 1957. This still holds true today, with a strand of recent works building formal analysis of modern neural networks methods in terms of formal languages. In this document, I aim to explain background about formal languages as they relate to to this recent work. I will by necessity ignore large parts of the rich history of this field, instead focusing on presenting formal language theoretic concepts from the perspective of modern deep learning-based NLP.
... • Tree Adjoining Grammar (TAG, Joshi, Levy, and Takahashi, 1975), a tree-rewriting system where a grammar consists of a set of elementary trees divided in initial and auxiliary trees which are combined of two tree rewriting operations called adjunction and substitution. The set of grammar rules is somewhat similar to CFG, but TAG constitutes a tree-generating rather than a string-generating system such as CFG. ...
In the rise of the internet, user-generated content from social networking services is becoming a giant source of information that can be useful to businesses on the aspect where users are viewed as customers or potential customers for companies. Exploitation of user-generated texts can help identify their feelings, intentions, or reduce the effort of the agents who are responsible for collecting or receiving information on social networking services. As part of this thesis, the content of texts such as speeches, statements, conversations from interactive communication on social media platforms become the main data object of our study. We deepen an analysis of structures and components of sentences in texts on the basis of Combinatory Categorial Grammar (CCG) and the Discourse Representation Structure (DRS). We propose a method for extracting a CCG tree from the dependency structure of the sentence, and a general architecture to build a bridge of relationship between syntaxes and semantics of French sentences. As a result, our study achieves representations of natural language texts in a new form of first order logic or the box of DRS.
... Besides these unsuccessful attempts, a few generalized grammar models capable of giving meaningful descriptions of syntax were discovered. The first such models were Aho's indexed grammars [1], Fischer's macro grammars [11] and tree-adjoining grammars by Joshi et al. [16]. Later, the ideas behind these models led to the more practical multi-component grammars [42,47], which became a standard model in computational linguistics and receive continued attention. ...
A classical result by Floyd ("On the non-existence of a phrase structure grammar for ALGOL 60", 1962) states that the complete syntax of any sensible programming language cannot be described by the ordinary kind of formal grammars (Chomsky's ``context-free''). This paper uses grammars extended with conjunction and negation operators, known as conjunctive grammars and Boolean grammars, to describe the set of well-formed programs in a simple typeless procedural programming language. A complete Boolean grammar, which defines such concepts as declaration of variables and functions before their use, is constructed and explained. Using the Generalized LR parsing algorithm for Boolean grammars, a program can then be parsed in time in its length, while another known algorithm allows subcubic-time parsing. Next, it is shown how to transform this grammar to an unambiguous conjunctive grammar, with square-time parsing. This becomes apparently the first specification of the syntax of a programming language entirely by a computationally feasible formal grammar.
Languages are governed by syntactic constraints—structural rules that determine which sentences are grammatical in the language. In English, one such constraint is subject-verb agreement, which dictates that the number of a verb must match the number of its corresponding subject: “the dogs run”, but “the dog runs”. While this constraint appears to be simple, in practice speakers make agreement errors, particularly when a noun phrase near the verb differs in number from the subject (for example, a speaker might produce the ungrammatical sentence “the key to the cabinets are rusty”). This phenomenon, referred to as agreement attraction, is sensitive to a wide range of properties of the sentence; no single existing model is able to generate predictions for the wide variety of materials studied in the human experimental literature. We explore the viability of neural network language models—broad-coverage systems trained to predict the next word in a corpus—as a framework for addressing this limitation. We analyze the agreement errors made by Long Short-Term Memory (LSTM) networks and compare them to those of humans. The models successfully simulate certain results, such as the so-called number asymmetry and the difference between attraction strength in grammatical and ungrammatical sentences, but failed to simulate others, such as the effect of syntactic distance or notional (conceptual) number. We further evaluate networks trained with explicit syntactic supervision, and find that this form of supervision does not always lead to more human-like syntactic behavior. Finally, we show that the corpus used to train a network significantly affects the pattern of agreement errors produced by the network, and discuss the strengths and limitations of neural networks as a tool for understanding human syntactic processing.
Ce texte traite de l’inscription des grammaires formelles à l’aide de diagrammes. Au travers de l’examen d’exemples tirés de la grammaire catégorielle, de la tree-adjoining grammar (TAG) et de la grammaire d’unification polarisée (GUP), nous montrons comment les diagrammes inscrivent les opérations nécessaires pour effectuer des calculs tels que la vérification de la structure d’une phrase ou la génération de structures syntaxiques à partir de structure sémantique. Ces opérations sont assemblées dans un métadiagramme, qui établit des relations entre plusieurs diagrammes. Le métadiagramme inscrit visuellement des opérations abstraites ; ce faisant, il autorise l’utilisateur à assigner une valeur à la position relative des entités graphiques sur le plan dans le processus de réflexion, ce qui peut mener à une utilisation créative qui exploite les contraintes formelles posées.
Lexical Functional Grammar (LFG) is a lexicalist, constraint-based grammatical theory that shares a lot of the basic assumptions of Construction Grammar (CxG), such as a commitment to surface-oriented descriptions (no transformations), and the simultaneous representation of form, meaning, and other grammatical information (no derivations). Nevertheless, LFG is not standardly viewed as a kind of CxG, in particular since its adherence to the principle of Lexical Integrity means that it insists on a strict morphology-syntax distinction where CxG canonically rejects such a divide. However, such a distinction is in fact entirely compatible with CxG assumptions; the actual problem with viewing LFG as a CxG is the difficulty it has in describing the more substantive end of the schematic-substantive spectrum of constructions. I suggest that by replacing the limited context-free grammar base of LFG responsible for this shortcoming with a more expressive formalism (in this case a description-based tree-adjoining grammar), we can obtain a fully constructional LFG, suitable as a formal framework for CxG.
Five recall-based structural priming experiments tested the predictions about dative structural priming derived from a new theory of structure building in sentence production. When both prime and target sentences contained direct object filler-gap dependencies, repeating a dative verb enhanced dative priming (the lexical boost). In contrast, the lexical boost was not observed when only target sentences contained object filler-gap dependencies. Additionally, the lexical boost was not observed when prime and target sentences contained object filler-gap dependencies but had mismatching tenses/aspects. In contrast, when neither prime nor target sentences contained object filler-gap dependencies, the lexical boost was observed despite prime and target sentences having different tenses/aspects. These findings confirm the unique set of predictions of the proposed theory, which posits that the size of compositional units is affected by the dependency structures of sentences.
In the current trend of unstructured language usage in simpler and shorter conversations, we identified a need of language analysis tool (a language parser) to find semantic relations within the sentence constituents. Earley’s algorithm is used in this paper to propose a parsing approach based on the Tree Adjoining Grammar (TAG) formalism. It is able to extract dependency relations in the derivation tree form. A generalized TAG tree-grammar pertaining to Subject-Object-Verb (SOV) syntactic pattern is also presented corresponding to the proposed algorithm. This grammar is targeted to be suitable for short dialogues in low resource languages which have scarcity of enough data for ML/DL based approaches. The process flow is presented using the combinatory tree-set formation and parsing. Two implementations are carried out using XML and LISP notation data structure. Examples from an Indian language are presented in a step-by-step manner that illustrates parsing in a controlled grammar environment for linguistically not-very complex syntactic constructions. Results along with findings are presented for execution time and memory consumption.KeywordsTree adjoining grammarNatural Language ProcessingSyntactic ParserRule-based approachesEaley’s algorithm
In most rule-based natural language processing systems, phrase structure grammar (PSG), which is also called context-free grammar (CFG), is currently most widely used. This is a very important formal model in the NLP.
Arabic is a challenging language when it comes to grammar production and parsing. It combines complex linguistic phenomena with a rich morphology that make its processing particularly ambiguous. This leaded us to choose the Tree-Adjoining Grammar (TAG) formalism. Indeed, TAG provides sufficient constraints for handling diverse linguistic phenomena and seems to be adequate to represent Arabic syntactic structures. In this paper, we present a semi-automatically generated TAG for modern standard Arabic using a compiler and a metagrammatical description language called XMG (eXtensible MetaGrammar). We describe the linguistic coverage of our grammar, and show how we used TAG and XMG’s properties to define in an expressive and concise way different linguistic phenomena. To check the coverage of our grammar, we have set up a development environment including a parser and using a test corpus of linguistic phenomena gathering both grammatical and ungrammatical sentences.
Understanding what someone says requires relating words in a sentence to one another as instructed by the grammatical rules of a language. In recent years, the neurophysiological basis for this process has become a prominent topic of discussion in cognitive neuroscience. Current proposals about the neural mechanisms of syntactic structure building converge on a key role for neural oscillations in this process, but they differ in terms of the exact function that is assigned to them. In this Perspective, we discuss two proposed functions for neural oscillations — chunking and multiscale information integration — and evaluate their merits and limitations taking into account a fundamentally hierarchical nature of syntactic representations in natural languages. We highlight insights that provide a tangible starting point for a neurocognitive model of syntactic structure building. Neural oscillations are thought to have an important role in syntactic structure building but views differ on their exact function in this context. In this Perspective, Kazanina and Tavano explore two proposed functions for neural oscillations in this process, namely chunking and multiscale information integration.
This is a review of
Vijay-Shanker, K.; Weir, D. J.
The equivalence of four extensions of context-free grammars. (English) Zbl 0813.68129
Math. Syst. Theory 27, No. 6, 511-546 (1994).
A previous model of long-distance dependency production claims that speakers use two distinct pieces of structures containing clause-taking verbs like believe and the complementizer that or the null complementizer when planning sentences with cross-clausal filler-gap dependencies (e.g., Who did the breeder believe (that) the dog bit?) vs. when planning sentences without (e.g., The breeder believed (that) the dog bit them.). Under a certain assumption about the lexical boost effect, this model predicts that the lexical boost effect for that-priming occurs only when prime and target sentences both contain a cross-clausal filler-gap dependency or when neither does. In the current study, a computational model of structural priming implementing the core claims of the previous filler-gap dependency production model was built to show that this prediction coherently follows from the model. The prediction of the model was then tested in five recall-based structural priming experiments. Speakers showed a larger complementizer priming effect when prime and target sentences share a clause-taking verb (i.e., the lexical boost effect). But the lexical boost effect was selective to when both prime and target sentences contained cross-clausal filler-gap dependencies (Experiment 3) and when neither did (Experiment 1). Critically, the lexical boost effect was absent when only either prime or target sentences contained filler-gap dependencies crossing the complementizer structure (Experiments 2, 4, and 5), confirming the prediction of the model.
In Chap. 3, we introduced the preliminaries of GP, a generic optimization tool that can generate and search for optimal solutions in a general solution space. GP is the first key ingredient that will be used to realize the conceptualized identification methodology. Recall that in the context of automated SI, the desired solution space would consist of all model structures (with varying degrees of model complexity) that are relevant for the identification task at hand. Furthermore, when the user has prior knowledge of, or preference for the desired model structure, it should be possible to suitably adapt the search space of the desired SI method. These ideas connect to Research Sub-questions 1.1 and 1.2 proposed in Chap. 1.
Rule-based machine translation uses a set of linguistic rules in the process of translation. The results of these systems are usually better than the results of statistical models from grammatical and word order perspective, But it has been shown that statistical models are more powerful in selecting proper words and generating more fluent translations. In this paper our goal is to improve the word choice in rule-based machine translation. This is done by a set of lexical syntactic rules based on Tree Adjoining Grammar. These probabilistic rules are statistically extracted from a large parallel corpus. In the proposed system, the input sentence is first reordered by a rule-based system, and them the decoding is carried out monotonically by using dynamic programming. In this system, the best translation is chosen based on the extracted rules and the language model score. The experiments on English-Persian translation showed that the proposed method resulted in an improvement of 1.3 in BLEU score in comparison to our baseline rule-based method.
Constituent and dependency parsing, the two classic forms of syntactic parsing, have been found to benefit from joint training and decoding under a uniform formalism, inspired by Head-driven Phrase Structure Grammar (HPSG). We thus refer to this joint parsing of constituency and dependency as HPSG-like parsing. However, in HPSG-like parsing decoding this unified grammar has a higher time complexity (
) than decoding either form individually (
) since more factors have to be considered during decoding. We thus propose an improved head scorer that helps achieve a novel performance-preserved parser in O(
) time complexity. Furthermore, on the basis of this proposed practical HPSG-like parser, we investigated the strengths of HPSG-like parsing and explored the general method of training an HPSG-like parser from only a constituent or dependency annotations in a multilingual scenario. We thus present a more effective, more in-depth, and general work on HPSG-like parsing.
Code generation is an important research field of software engineering, aiming to reduce development costs and improve program quality. Nowadays, more and more researchers intend to implement code generation by natural language understanding. In this paper, we propose a generation method to convert natural language descriptions to the program code based on deep learning. We use an encoder-decoder model with gated attention mechanism. Here, the decoder is an InterAction-LSTM. The gated attention combines the previous decoding cell state with source representations to improve the limitation of invariant source representations. The decoder makes the information interact each other before putting them into the gate of the LSTM. The code generation is verified on two datasets, Conala and Django. Compared with known models, our model outperforms the baselines both in Accuracy and Bleu.
In a sentence like Who does the artist think chased the chef?, the who at the beginning depends on the last bit of the sentence, chased the chef. This is an instance of a long-distance dependency. What is the nature of the cognitive process that allows speakers to produce sentences that include distant elements that form dependencies? In four experiments, speakers described drawings that elicited long-distance dependencies. Critically, speakers were sometimes primed to produce a that in sentences where that was ungrammatical due to a grammatical constraint known as the that-trace constraint (e.g.,*Who does the artist think that chased the chef). Results showed that, when primed to say an ungrammatical that, speakers were slower to start to speak. Because the that-trace constraint applies selectively to certain configurations of long-distance dependencies, this suggests that the grammatical details of the long-distance dependency are already planned before speakers start to speak the sentences involving long-distance dependencies. I propose a formal model that explains how speakers plan long-distance dependencies in advance of speaking them while also managing the cognitive pressure to speak sentences incrementally.
The classification of grammars that became known as the Chomsky hierarchy was an exploration of what kinds of regularities could arise from grammars that had various conditions imposed on their structure. Intersubstitutability is closely related to the way different levels on the Chomsky hierarchy correspond to different kinds of memory. This chapter deals with the general concept of a string‐rewriting grammar, which provides the setting in which the Chomsky hierarchy can be formulated. An unrestricted rewriting grammar works with a specified set of nonterminal symbols, and specified set of terminal symbols. From the very outset there were doubts about about whether context‐free grammars (CFGs) could form the basis of a theory of natural language syntax. The grammar's rewrite rules correspond to the automaton's transitions. Chomsky argued that even if the generative capacity of CFGs turned out to be sufficient for English, the resulting grammars would be unreasonably complex.
We introduce context-free grammars with lookahead. The grammars are an extension of both context-free grammars and parsing expression grammars, hence we can handle the two grammars in a unified way. To accommodate lookahead, we use a language with lookahead, which is a set of string pairs. We considered the grammar as a system of equations and give the language with lookahead by the limit of iterations from the empty set. The language class is closed under union, intersection, complement, and a weak version of concatenation and Kleene star.
Information‐theoretic complexity metrics, such as Surprisal (Hale, 2001; Levy, 2008) and Entropy Reduction (Hale, 2003), are linking hypotheses that bridge theorized expectations about sentences and observed processing difficulty in comprehension. These expectations can be viewed as syntactic derivations constrained by a grammar. However, this expectation‐based view is not limited to syntactic information alone. The present study combines structural and non‐structural information in unified models of word‐by‐word sentence processing difficulty. Using probabilistic minimalist grammars (Stabler, 1997), we extend expectation‐based models to include frequency information about noun phrase animacy. Entropy reductions derived from these grammars faithfully reflect the asymmetry between subject and object relatives (Staub, 2010; Staub, Dillon, & Clifton, 2017), as well as the effect of animacy on the measured difficulty profile (Lowder & Gordon, 2012; Traxler, Morris, & Seely, 2002). Visualizing probability distributions on the remaining alternatives at particular parser states allows us to explore new, linguistically plausible interpretations for the observed processing asymmetries, including the way that expectations about the relativized argument influence the processing of particular types of relative clauses (Wagers & Pendleton, 2016).
Author investigates sets of strings produced from regular sets of trees. The first result exhibits some semi-Thur systems which generate a special subclass of the context-free sets. These context-free sets have properties similar to the regular sets. The second result is that trees cannot naturally be written as strings so that regular sets of trees become regular sets of strings. The method of proof of nonregularity is what is of interest, since the result is definitely expected.
This paper reviews the definition of a new form of tree grammar, the tree adjunct grammar [1]. Using tree adjunct grammars, we obtain two results:
This paper extends a formal theory of language learning to transformational components. Learning procedures which are psychologically more suggestive than those previously studied are shown to yield positive results under formally specified conditions. Part 1 introduces the general class of problems to be studied; Part 2 states and discusses various possible assumptions; and Part 3 sketches the proof for a particular case of interest. For a complete proof, see Hamburger (1971).
Let Φ be any abstract measure of computational complexity, and let L denote the specific measure of memory resource (tape) on one tape Turing machines. Denote by Rt( )Φ the class of all total functions whose Φ-complexity is bounded by ...
In this paper [and in Joshi et al. (1972) which is a sequel to this paper] a new style of formal grammar called String Adjunct Grammars (AG) has been studied. The rules in an AG have a character essentially different from the “rewrite rule” in a Phrase Structure Grammar (PSG). Such a study of formal grammars of different styles is of great interest because each style is well suited for characterizing certain aspects of natural language structure but has inherent difficulty in characterizing certain other aspects. Several subclasses of AG's motivated by strong linguistic considerations have been studied. Linguistic relevance of these grammars and other grammars suggested by this study has also been discussed.
In this paper, we continue the study of String Adjunct Grammars (AG) introduced in Joshi et al. (1972). In particular, equational representations of LAG's, LAG's with null symbols, and some special cases of LAG's are studied. Linguistic relevance of these grammars is also discussed in some detail.
This paper is intended for computer programmers and other scientists who may be interested in linguistics. We assume that the reader has no previous knowledge of linguistics, but some experience with logical or mathematical reasoning. * *This paper constitutes the substance of a talk given for members of SLANG, a special committee of the Association for Computing Machinery, at Atkins Laboratory at Harvard University, Cambridge, Mass., on July 6, 1967. I am grateful to Guy Carden for extensive discussion of this paper and for many editorial comments and improvements. View all notes
The following three results concerning tree automata are presented in this paper. (1) Rounds has presented the following open problem: For every recognizable setR, can we construct a deterministic finite-state transformation recognizingR? We show that this is not possible, in fact, even for a local set. However, the following is true: For every recognizable setR there is an inverse projectionR effectively obtained such thatR is recognized by a deterministic finite-state transformation. (2) Martin and Vere in their study of tree automata leave open the question of whether Generalized Syntax Directed Transductions (GSDT's) are closed under Arden's transformation or Greibach's transformation, and conjecture that they are not. We prove that this conjecture is true. It is also shown that GSDT's are not closed under transformation to LR(k) grammars. (3) Peters and Ritchie have shown that if, in a grammar where the generative rules are context-free, there are recognition rules which are context-sensitive, the language recognized is still context-free. A tree-automata-oriented proof is given by Rounds. We show that a similar result holds also for right linear grammars, i.e., if the generative rules are right linear, then using context-sensitive rules for recognition, one can still recognize only regular languages. Some other related results concerning context-sensitive extensions of subclasses of context-free languages are also presented.
The recognizable sets of value trees (pseudoterms) are shown to be exactly projections of sets of derivation trees of (extended) context-free grammars.
The introduction of syntax directed translations and transformations into formallanguage theory presents a very interesting area with considerable promise of application to questions of syntax and semantics of programming languages. The concept of generalized sequential machine (gsm) mapping (already of importance in language theory) is developed here in its natural extension to trees (or expressions). That generalized concept of gsm mapping encompasses most of the previously defined concepts relating to translations and transformations.
Trees are defined as mappings from tree structures (in the graph-theoretic sense) into sets of symbols.Regular systems are defined in which the production rules are of the form Φ → ψ, where Φ and ψ are trees. An application of a rule involves replacing a subtree Φ by the tree ψ.The main result is that the sets of trees generated by regular systems are exactly those that are accepted by tree automata. This generalizes a theorem of BÜchi, proved for strings.
A new type of grammar for generating formal languages, called an indexed grammar, is presented. An indexed grammar is an extension of a context-free grammar, and the class of languages generated by indexed grammars has closure properties and decidability results similar to those for context-free languages. The class of languages generated by indexed grammars properly includes all context-free languages and is a proper subset of the class of context-sensitive languages. Several subclasses of indexed grammars generate interesting classes of languages.
In this report, certain properties of context-free (CF or type 2) grammars are investigated, like that of Chomsky. In particular, questions regarding structure, possible ambiguity and relationship to finite automata are considered. The following results are presented:
The language generated by a context-free grammmar is linear in a sense that is defined precisely.The requirement of unambiguity—that every sentence has a unique phrase structure—weakens the grammar in the sense that there exists a CF language that cannot be generated unambiguously by a CF grammar.The result that not every CF language is a finite automaton (FA) language is improved in the following way. There exists a CF language L such that for any L′ ⊆ L, if L′ is FA, an L″ ⊆ L can be found such that L″ is also FA, L′ ⊆ L″ and L″ contains infinitely many sentences not in L′.A type of grammar is defined that is intermediate between type 1 and type 2 grammars. It is shown that this type of grammar is essentially stronger than type 2 grammars and has the advantage over type 1 grammars that the phrase structure of a grammatical sentence is unique, once the derivation is given.
Noam Chomsky's first book on syntactic structures is one of the first serious attempts on the part of a linguist to construct within the tradition of scientific theory-construction a comprehensive theory of language which may be understood in the same sense that a chemical, biological theory is understood by experts in those fields. It is not a mere reorganization of the data into a new kind of library catalogue, nor another specualtive philosophy about the nature of man and language, but rather a rigorus explication of our intuitions about our language in terms of an overt axiom system, the theorems derivable from it, explicit results which may be compared with new data and other intuitions, all based plainly on an overt theory of the internal structure of languages; and it may well provide an opportunity for the application of explicity measures of simplicity to decide preference of one form over another form of grammar.
In this paper, we have introduced a new style of formal grammars called String Adjunct Grammars (AG). The rules in an AG have a considerably different formal character as compared to the 'rewrite rule' in a Phrase Structure Grammar (PSG). Such a study of formal grammars of different styles (i.e., formal character of rules) is of great interest because each style is well suited for characterizing certain aspects of natural language structure and is awkward for characterizing certain other aspects. Several subclasses of AG's motivated by strong linguistic considerations have been studied, comparing them with PSG's. Linguistic relevance of these grammars (and other gram mars suggested by this study) has been discussed at the end.
How much hierarchical structures is necessary for sentence description?
- A.K. Joshi
- A.K. Joshi
A characterization of the derivation trees of a context-free grammar and an intercalation theorem
- Joshi
A. K. JOSHI AND M. TAKAHASHI, "A characterization of the derivation trees of a context-free grammar and an intercalation theorem," Tech. Report, The Moore School of Electrical Engineering, University of Pennsylvania, 1971.
Generalized sequential transducer
- J W Thatcljer
J. W. THATClJER, Generalized sequential transducer, ]. Comput. System Sci. 4 (1970), 339-367.
String adjunct grammars: Parts I and II
- Joshi
Generalized sequential transducer
- Thatcher