Search

An important reason to prefer dependency parsing over classical phrased based methods, especially for languages such as Persian, with the property of being "free word order", is that this particular property has a negative impact on the accuracy of conventional parsing methods. In Persian, some words such as adverbs can freely be moved within a sentence without affecting its correctness or meaning. In this paper, we illustrate the robustness of dependency parsing against this particular problem by training two well-known dependency parsers, namely MST Parser and Malt Parser, using a Persian dependency corpus called Dadegan. We divided the corpus into two separate parts including only projective sentences and only non-projective sentences, which are corelated with the free word order property. As our results show, MST Parsing is not only more tolerant than Malt Parsing against the free word order problem, but it is also in general a more accurate technique.

We propose a hybrid dependency parsing pipeline which combines transition-based parser and graph-based parser, and use segmented treebanks to train transition-based parsers as subparsers in front end, and then propose a constrained Eisner’s algorithm to reparse their outputs. We build the pipeline to investigate the influence on parsing accuracy when training with different segmentations of training data and find a convenient method to obtain parsing reliability score while achieving state-of-the-art parsing accuracy. Our results show that the pipeline with segmented training dataset could improve accuracy through reparsing while providing parsing reliability score.

This chapter investigates which formal properties make a formal grammar suitable for automatic language analysis and which do not. For this purpose, context-free PS Grammar and its parsers will be used as our main example.

Language embedding is a method to implement a new language within the framework of an existing programming language. This method is known to speed up the development process compared to standalone languages using classical compiler technology. On the other hand, embedded languages may not be that convenient for the end-users as standalone ones with own concrete syntax. This paper describes a method that uses the flexibility of language embedding in the experimental phase of the language design process, then, once the language features are mature enough, adds concrete syntax and turns the language to a standalone one. Lessons learnt from a project, run in industry-university cooperation and using the presented method, are discussed. Based on these results, a cost model is established that can be used to estimate the potential benefits of this method in case of future language design projects.

Concept-level text analysis is superior to word-level analysis as it preserves the semantics associated with multi-word expressions. It offers a better understanding of text and helps to significantly increase the accuracy of many text mining tasks. Concept extraction from text is a key step in concept-level text analysis. In this paper, we propose a ConceptNet-based semantic parser that deconstructs natural language text into concepts based on the dependency relation between clauses. Our approach is domain-independent and is able to extract concepts from heterogeneous text. Through this parsing technique, 92.21% accuracy was obtained on a dataset of 3,204 concepts. We also show experimental results on three different text analysis tasks, on which the proposed framework outperformed state-of-the-art parsing techniques.

Parsing plays a significant role in many natural language processing NLP applications as their efficiency relies on having an effective parser. This paper presents Amharic sentence parser developed using base phrase chunker that groups syntactically correlated words at different levels. We use HMM to chunk base phrases where incorrectly chunked phrases are pruned with rules. The task of parsing is then performed by taking chunk results as inputs. Bottom-up approach with transformation algorithm is used to transform the chunker to the parser. Corpus from Amharic news outlets and books was collected for training and testing. The training and testing datasets were prepared using the 10-fold cross validation technique. Test results on the test data showed an average parsing accuracy of 93.75%.

Annotations on structured corpora provide a foundational instrument for emerging linguistic research. To generate annotations automatically, data-driven dependency parsers need a large annotated corpus to learn from. But these annotations are expensive to collect and require a labor intensive task. In order to reduce the costs of annotation, we provide a novel framework in which a committee of dependency parsers collaborate to improve their efficiency using active learning.

The paper presents a new online service for the dependency parsing of Polish. Given raw text as input, the service processes it and visualises output dependency trees. The service applies the parsing system – MaltParser – with a parsing model for Polish trained on the Polish Dependency Bank, and some additional publicly available tools.

International audience; The paper investigates an extension of LR parsing that allows the delay of parsing decisions until a sufficient amount of context has been processed. We provide two characterizations for the resulting class of grammars, one based on grammar transformations, the other on the direct construction of a parser. We also report on experiments with a grammar collection.

Combinatory Categorial Grammar CCG is a grammar formalism used for natural language parsing. CCG assigns structured lexical categories to words and uses a small set of combinatory rules to combine these categories in order to parse sentences. In this work we describe and implement a new approach to CCG parsing that relies on Answer Set Programming ASP -- a declarative programming paradigm.Different from previous work, we present an encoding that is inspired by the algorithm due to Cocke, Younger, and Kasami CYK. We also show encoding extensions for parse tree normalization and best-effort parsing and outline possible future extensions which are possible due to the usage of ASP as computational mechanism. We analyze performance of our approach on a part of the Brown corpus and discuss lessons learned during experiments with the ASP tools dlv, gringo, and clasp. The new approach is available in the open source CCG parsing toolkit AspCcgTk which uses the C&C supertagger as a preprocessor to achieve wide-coverage natural language parsing.

The theory of context-free languages is well-understood and context-free parsers can be used as off-the-shelf tools in practice. In particular, to use a context-free parser framework, a user does not need to understand its internals but can specify a language declaratively as a grammar. However, many languages in practice are not context-free. One particularly important class of such languages is layout-sensitive languages, in which the structure of code depends on indentation and whitespace. For example, Python, Haskell, F#, and Markdown use indentation instead of curly braces to determine the block structure of code. Their parsers (and lexers) are not declaratively specified but hand-tuned to account for layout-sensitivity.

This paper investigates two subjects in push-down automata (PDAs) and linear indexed grammars (LIGs), which are extended PDAs, focusing on eliminating the stack symbols. One of the subjects is concerned with PI- (push-input-) PDA and PI-LIG without e-transition rule, in which only input symbols are pushed down to the stack. It is shown that the class of languages of PI-LIGs is incomparable with that of PDAs, which is the class of context-free languages (CFLs). The other subject is a simple bottom-up parsing method for LIGs, in which the stack symbols are eliminated at the first step of the parsing. The paper shows several PI-LIGs, including PI-PDAs for fundamental context-free and context-sensitive languages, which are synthesized by a grammatical inference system LIG Learner.

If all we want from a syntactic parser is a dependency tree, what do we gain by first computing a different representation such as a phrase structure tree? The principle of parsimony suggests that a simpler model should be preferred over a more complex model, all other things being equal, and the simplest model is arguably one that maps a sentence directly to a dependency tree --- a bare-bones dependency parser. In this paper, I characterize the parsing problem faced by such a system, survey the major parsing techniques currently in use, and begin to examine whether the simpler model can in fact rival the performance of more complex systems. Although the empirical evidence is still limited, I conclude that bare-bones dependency parsers can achieve state-of-the-art parsing accuracy and often excel in terms of efficiency.

The fastest parsers currently can parse an average sentence in up to 2.5ms, a considerable improvement, since most of the older accuracy-oriented parsers parse only few sentences per second. It is generally accepted that the complexity of a parsing algorithm is decisive for the performance of a parser. However, we show that the most time consuming part of processing is feature extraction and therefore an algorithm which allows efficient feature extraction can outperform a less complex algorithm which does not. Our system based on quadratic Covington's parsing strategy with efficient feature extraction is able to parse an average English sentence in only 0.8ms without any parallelisation.

Combinatory categorial grammar (CCG) is a grammar formalism used for natural language parsing. CCG assigns structured lexical categories to words and uses combinatory rules to combine these categories to parse a sentence. In this work we propose and implement a new approach to CCG parsing that relies on a prominent knowledge representation formalism, answer set programming (ASP) - a declarative programming paradigm. We formulate the task of CCG parsing as a planning problem and use an ASP computational tool to compute solutions that correspond to valid parses. Compared to other approaches, there is no need to implement a specific parsing algorithm using such a declarative method. Our approach aims at producing all semantically distinct parse trees for a given sentence. From this goal, normalization and efficiency issues arise, and we deal with them by combining and extending existing strategies.We have implemented a CCG parsing tool kit-AspCcgTk-that uses ASP as its main computational means. The C&C supertagger can be used as a preprocessor within AspCcgTk, which allows us to achieve wide-coverage natural language parsing.

Semantic parsing is the task of mapping a natural sentence to a meaning representation. The limitation of semantic parsing is that it is very difficult to obtain annotated training data in which a sentence is paired with a semantic representation. To deal with this problem, we introduce a semi supervised learning model for semantic parsing with ambiguous supervision. The main idea of our method is to utilize a large amount of data, to enrich feature space with the maximum entropy model using our semantic learner. We evaluate the proposed models on standard corpora to show that our methods are suitable for semantic parsing problem. Experimental results show that the proposed methods work efficiently and well on ambiguous data and it is comparable to the state of the art method.

Humans use semantics during parsing; so should computers. In contrast to phrase structure-based parsers, COGPARSE seeks to determine which meaning-bearing components are present in a text, using world knowledge and lexical semantics for construction grammar form selection, syntactic overlap processing, disambiguation, and confidence calculation. In a brain-inspired way, COGPARSE aligns parsing with the structure of the lexicon, providing a linguistic representation, parsing algorithm, associated linguistic theory, and preliminary metrics for evaluating parse quality. Given sufficient information on nuanced word and construction semantics, COGPARSE can also assemble detailed full-semantics meaning representations of input texts. Beyond the ability to determine which parses are most likely to be intended and to use knowledge in disambiguation, full-semantics parsing enables nuanced meaning representation, learning, summarization, natural language user interfaces, and the taking of action based on natural language input.

Dependent grammars extend context-free grammars by allowing semantic values to be bound to variables and used to constrain parsing. Dependent grammars can cleanly specify common features that cannot be handled by context-free grammars, such as length fields in data formats and significant indentation in programming languages. Few parser generators support dependent parsing, however. To address this shortcoming, we have developed a new method for implementing dependent parsers by extending existing parsing algorithms. Our method proposes a point-free language of dependent grammars, which we believe closely corresponds to existing context-free parsing algorithms, and gives a novel transformation from conventional dependent grammars to point-free ones. To validate our technique, we have specified the semantics of both source and target dependent grammar languages, and proven our transformation sound and complete with respect to those semantics. Furthermore, we have empirically validated the suitability of our point-free language by adapting four parsing engines to support it: an Earley parsing engine; a GLR parsing engine; memoizing, arrow-style parser combinators; and PEG parser combinators.

Discourse parsing of complex text types such as scientific research articles requires the analysis of an input document on linguistic and structural levels that go beyond traditionally employed lexical discourse markers. This chapter describes a text-technological approach to discourse parsing. Discourse parsing with the aim of providing a discourse structure is seen as the addition of a new annotation layer for input documents marked up on several linguistic annotation levels. The discourse parser generates discourse structures according to the Rhetorical Structure Theory. An overview of the knowledge sources and components for parsing scientific journal articles is given. The parser’s core consists of cascaded applications of the GAP, a Generic Annotation Parser. Details of the chart parsing algorithm are provided, as well as a short evaluation in terms of comparisons with reference annotations from our corpus and with recently developed systems with a similar task.

Word alignment and parsing are two important components for syntax based machine translation. The inconsistent models for alignment and parsing caused problems during translation pair extraction. In this paper, we do word alignment and dependency parsing in a multi-task learning framework, in which word alignment and dependency parsing are consistent and assisted with each other. Our experiments show significant improvement not only for both word alignment and dependency parsing, but also the final translation performance.

In the following, we will see different parsing algorithms for RCGs. Let us consider a variant of the RCG for the MIX language from Fig. 8.4 as a running example. This variant is given in Fig. 9.1.

In this chapter, we complete our study of regular dependency languages by investigating their string-generative capacity and parsing complexity. Specifically, we study the connection between these two measures and the structural constraints discussed in the first part of this book.

Maltparser is a contemporary dependency parsing machine learning-based system that shows great accuracy. However 90% for Labelled Attachment Score (LAS) seems to be a de facto limit for such kinds of parsers. Since generally such systems can not be modified, previous works have been developed to study what can be done with the training corpora in order to improve parsing accuracy. High level techniques, such as controlling sentences' length or corpora's size, seem useless for these purposes. But low level techniques, based on an in-depth study of the errors produced by the parser at the word level, seem promising. Prospective low level studies suggested the development of n-version parsers. Each one of these n versions should be able to tackle a specific kind of dependency parsing at the word level and the combined action of all them should reach more accurate parsings. In this paper we present an extensive study on the usefulness and the expected limits for n-version parser to improve parsing accuracy. This work has been developed specifically for Spanish using Maltparser.

One obstacle to the implementation of modular extensions to programming languages lies in the problem of parsing extended languages. Specifically, the parse tables at the heart of traditional LALR(1) parsers are so monolithic and tightly constructed that, in the general case, it is impossible to extend them without regenerating them from the source grammar. Current extensible frameworks employ a variety of solutions, ranging from a full regeneration to using pluggable binary modules for each different extension. But recompilation is time-consuming, while the pluggable modules in many cases cannot support the addition of more than one extension, or use backtracking or non-deterministic parsing techniques. We present here a middle-ground approach that allows an extension, if it meets certain restrictions, to be compiled into a parse table fragment. The host language parse table and fragments from multiple extensions can then always be efficiently composed to produce a conflict-free parse table for the extended language. This allows for the distribution of deterministic parsers for extensible languages in a pre-compiled format, eliminating the need for the “source code” grammar to be distributed. In practice, we have found these restrictions to be reasonable and admit many useful language extensions.

The code generator in a compiler attempts to match a subject tree against a collection of tree-shaped patterns for generating instructions. Tree-pattern matching may be considered as a generalization of string parsing. We propose a new generalized LR (GLR) parser, which extends the LR parser stack with a parser cactus. GLR explores all plausible parsing steps to find the least-cost matching. GLR is fast due to two properties: (1) duplicate parsing steps are eliminated and (2) partial parse trees that will not lead to a least-cost matching are discarded as early as possible.

Despite the popularity of lexicalized parsing models, practical concerns such as data sparseness and applicability to domains of different vocabularies make unlexicalized models that do not refer to word tokens themselves deserve more attention. A classifier-based parser using an unlexicalized parsing model has been developed. Most importantly, to enhance the accuracy of these tasks, we investigated the notion of distituency (the possibility that two parts of speech cannot remain in the same constituent or phrase) and incorporated it as attributes using various statistic measures. A machine learning method integrates linguistic attributes and information-theoretic attributes in two tasks, namely sentence chunking and phrase recognition. The parser was applied to parsing English and Chinese sentences in the Penn Treebank and the Tsinghua Chinese Treebank. It achieved a parsing performance of F-Score 80.3% in English and 82.4% in Chinese.

In this paper we assess to what extent the available Portuguese treebanks and available probabilistic parsers are suitable for out-of-the-box robust parsing of Portuguese. We also announce the release of the best parser coming out of this exercise, which is, to the best of our knowledge, the first robust parser widely available for Portuguese.

Dependency parsing can be cast as a classification problem over strings of observations. Compared to shallow processing tasks like tagging, parsing is a dynamic classification problem as no statically predefined set of classes exists and any class to be distinguished is composed of pairs from a given label set (syntactic function) and the available attachment points in the sentence, so that even the number of “classes” varies with the length of the input sentence. A number of fundamentally different approaches have been pursued to solve this classification task. They differ in the way they consider the context, in whether they apply machine learning approaches or not, and in the means they use to enforce the tree property of the resulting sentence structure. These differences eventually result in a different behavior on the same data making the paradigm an ideal testbed to apply different information fusion schemes for combined decision making.

Property Grammars , or PGs, belong to a new family of linguistic formalisms which view a grammar as a set of linguistic constraints, and parsing as a constraint satisfaction problem. Rigid hierarchical parsing gives way to flexible mechanisms which can handle incomplete, ambiguous or erroneous text, and are thus more adequate for new applications such as speech recognition, internet mining, controlled languages and biomedical information. The present work contributes a) a new parsing methodology for PGs in terms of Hyprolog --- an extension of Prolog with linear and intuitionistic logic and with abduction; and b) a customisable extension of PGs that lets us model also concepts and relations to some degree. We exemplify within the domain of extracting concepts from biomedical text.

Shallow parsing has been proposed as a means of arriving at practically useful structures while avoiding the difficulties of full syntactic analysis. According to Abney's principles, it is preferred to leave an ambiguity pending than to make a likely wrong decision. We show that continuous phrase chunking as well as shallow constituency parsing display evident drawbacks when faced with freer word order languages. Those drawbacks may lead to unnecessary data loss as a result of decisions forced by the formalism and therefore diminish practical value of shallow parsers for Slavic languages. We present an alternate approach to shallow parsing of noun phrases for Slavic languages which follows the original Abney's principles. The proposed approach to parsing is decomposed into several stages, some of which allow for marking discontinuous phrases.

Analysis and renovation of large software portfolios requires syntax analysis of multiple, usually embedded, languages and this is beyond the capabilities of many standard parsing techniques. The traditional separation between lexer and parser falls short due to the limitations of tokenization based on regular expressions when handling multiple lexical grammars. In such cases scannerless parsing provides a viable solution. It uses the power of context-free grammars to be able to deal with a wide variety of issues in parsing lexical syntax. However, it comes at the price of less efficiency. The structure of tokens is obtained using a more powerful but more time and memory intensive parsing algorithm. Scannerless grammars are also more non-deterministic than their tokenized counterparts, increasing the burden on the parsing algorithm even further. In this paper we investigate the application of the Right-Nulled Generalized LR parsing algorithm (RNGLR) to scannerless parsing. We adapt the Scannerless Generalized LR parsing and filtering algorithm (SGLR) to implement the optimizations of RNGLR. We present an updated parsing and filtering algorithm, called SRNGLR, and analyze its performance in comparison to SGLR on ambiguous grammars for the programming languages C, Java, Python, SASL, and C++. Measurements show that SRNGLR is on average 33% faster than SGLR, but is 95% faster on the highly ambiguous SASL grammar. For the mainstream languages C, C++, Java and Python the average speedup is 16%.

The impact of clause and intraclausal coordination detection to dependency parsing of Slovene is examined. New methods based on machine learning and heuristic rules are proposed for clause and intraclausal coordination detection. They were included in a new dependency parsing algorithm, PACID. For evaluation, Slovene dependency treebank was used. At parsing, 6.4% and 9.2 % relative error reduction was achieved, compared to the dependency parsers MSTP and Malt, respectively.

There are numerous ways to implement a parser for a given syntax; using parser combinators is a powerful approach to parsing which derives much of its power and expressiveness from the type system and semantics of the host programming language. This tutorial begins with the construction of a small library of parsing combinators. This library introduces the basics of combinator parsing and, more generally, demonstrates how domain specific embedded languages are able to leverage the facilities of the host language. After having constructed our small combinator library, we investigate some shortcomings of the naive implementation introduced in the first part, and incrementally develop an implementation without these problems. Finally we discuss some further extensions of the presented library and compare our approach with similar libraries.

Natural language parsing, as one of the central tasks in natural language processing, is widely used in many AI fields. In this paper, we address an issue of parser performance evaluation, particularly its variation across datasets. We propose three simple statistical measures to characterize the datasets and also evaluate their correlation to the parser performance. The results clearly show that different parsers have different performance variation and sensitivity against these measures. The method can be used to guide the choice of natural language parsers for new domain applications, as well as systematic combination for better parsing accuracy.

This article presents a formalism and a beta version of a new tool for simultaneous morphosyntactic disambiguation and shallow parsing. Unlike in the case of other shallow parsing formalisms, the rules of the grammar allow for explicit morphosyntactic disambiguation statements, independently of structure-building statements, which facilitates the task of the shallow parsing of morphosyntactically ambiguous or erroneously disambiguated input.

This paper presents the design of the text analysis component of a TTS system for the Romanian language. Our text analysis is performed in two steps: document structure detection and text normalization. The output is a tree-based representation of the processed data. Parsing is made efficient with the help of the Boost Spirit LL parser [1], the usage of this tool allowing for a greater flexibility in the source code and in the output representation.

Interactive development environments (IDEs) increase programmer productivity, but unfortunately also the burden on language implementors since sophisticated tool support is expected even for small domain-specific languages. Our goal is to alleviate that burden, by generating IDEs from high-level language specifications using the JastAdd meta-compiler system. This puts increased tension on scope recovery in parsers, since at least a partial AST is required by the system to perform static analysis, such as name completion and context sensitive search. In this paper we present a novel recovery algorithm called bridge parsing, which provides a light-weight recovery mechanism that complements existing parsing recovery techniques. An initial phase recovers nesting structure in source files making them easier to process by existing parsers. This enables batch parser generators with existing grammars to be used in an interactive setting with minor or no modifications. We have implemented bridge parsing in a generic extensible IDE for JastAdd based compilers. It is independent of parsing technology, which we validate by showing how it improves recovery in a set of typical interactive editing scenarios for three parser generators: ANTLR (LL(variable lookahead) parsers), LPG (LALR(k) parsers), and Beaver (LALR(1) parsers). ANTLR and LPG both contain sophisticated support for error recovery, while Beaver requires manual error productions. Bridge parsing complements these techniques and yields better recovery for all these tools with only minimal changes to existing grammars.

Statistical parsers that simultaneously generate both phrase-structure and lexical dependency trees have been limited to date in two important ways: detecting non-projective dependencies has not been integrated with other parsing decisions, and/or the constraints between phrase-structure and dependency structure have been overly strict. We introduce context-free filtering grammar as a generalization of a lexicalized factored parsing model, and develop a scoring model to resolve parsing ambiguities for this new grammar formalism. We demonstrate the new model's flexibility by implementing a statistical parser for German, a freer-word-order language exhibiting a mixture of projective and non-projective syntax, using the TuBa-D/Z treebank [1].

This paper discusses some issues of developing a parser for spoken Estonian which is based on an already existing parser for written language, and employs the Constraint Grammar framework. When we used a corpus of face-to-face everyday conversations as the training and testing material, the parser gained the recall 97.6% and the precision 91.8%. The parsing of institutional phone calls turned out to be a more complicated task, with the recall dropping by 3%. In this paper, we will focus on parsing nonfluent speech using a rule-based parser. We will give an overview of parsing errors and ways to overcome them.

Parser combinators, often monadic, are a venerable and widely-used solution to read data from some external format. However, the capability to return a partial parse has, until now, been largely missing. When only a small portion of the entire data is desired, it has been necessary either to parse the entire input in any case, or to break up the grammar into smaller pieces and move some work outside the world of combinators. This paper presents a technique for mixing lazy, demand-driven, parsing with strict parsing, all within the same set of combinators. The grammar specification remains complete and unbroken, yet only sufficient input is consumed to satisfy the result demanded. It is built on a combinationof applicativeand monadicparsers. Monadic parsing alone is insufficient to allow a choice operator to coexist with the early commitment needed for lazy results. Applicative parsing alone can give partial results, but does not permit context-sensitive grammars. But used together, we gain both partiality and a flexible ease of use. Performance results demonstrate that partial parsing is often faster and more space-efficient than strict parsing, but never worse. The trade-off is that partiality has consequences when dealing with ill-formed input.

A rough set-based semi-naive Bayesian classification method is applied to dependency parsing, which is an important task in syntactic structure analysis of natural language processing. Many parsing algorithms have emerged combined with statistical machine learning techniques. The rough set-based classifier is embedded with Nivre's deterministic parsing algorithm to conduct dependency parsing task on a Chinese corpus. Experimental results show that the method has a good performance on dependency parsing task. Moreover, the experiments have justified the effectiveness of the classification influence.

This study presents new language and treebank independent graph transformations that improve accuracy in data-driven dependency parsing. We show that individual generic graph transformations can increase accuracy across treebanks, but especially when they are combined using established parser combination techniques. The combination experiments also indicate that the presumed best way to combine parsers, using the highest scoring parsers, is not necessarily the best approach.

Parsing models have many applications in AI, ranging from natural language processing (NLP) and computational music analysis to logic programming and computational learning. Broadly conceived, a parsing model seeks to uncover the underlying structure of an input, that is, the various ways in which elements of the input combine to form phrases or constituents and how those phrases recursively combine to form a tree structure for the whole input. During the last fifteen years, a major shift has taken place from rule-based, deterministic parsing to corpus-based, probabilistic parsing. A quick glance over the NLP literature from the last ten years, for example, indicates that virtually all natural language parsing systems are currently probabilistic. The same development can be observed in (stochastic) logic programming and (statistical) relational learning. This trend towards probabilistic parsing is not surprising: the increasing availability of very large collections of text, music, images and the like allow for inducing statistically motivated parsing systems from actual data.

International audience; Range Concatenation Grammar (RCG) is a formalism with interesting formal properties (it has a polynomial parsing time while being more powerful than Linear Context-Free Rewriting Systems). In this context, we present a constraint-based extension of the state-of-the-art RCG parsing algorithm of (Boullier, 2000), which has been used for the implementation of an open-source parsing architecture.

Packrat parsing is a newly popular technique for efficiently implementing recursive descent parsers. Packrat parsing avoids the potential exponential costs of recursive descent parsing with backtracking by ensuring that each production rule in the grammar is tested at most once against each position in the input stream. This paper argues that (a) packrat parsers can be trivially implemented using a combination of definite clause grammar rules and memoing, and that (b) packrat parsing may actually be significantly less efficient than plain recursive descent with backtracking, but (c) memoing the recognizers of just one or two nonterminals, selected in accordance with Amdahl's law, can sometimes yield speedups. We present experimental evidence to support these claims.