Customer Reviews

Speech and Language Processing: An Introduction to Natural Language Processing, Computational Linguistics and Speech Recognition (Practical Resources for the Mental Health Professionals)

5 star:		(17)
4 star:		(7)
3 star:		(3)
2 star:		(0)
1 star:		(2)

 

 

The previous best book on NLP was James Allen's (1995), which was considered ambitious at the time because it covered syntax, semantics and some pragmatics. But Martin and Jurafsky is far more ambitious, because it covers speech recognition as well, and has far expanded coverage of language generation and translation. It also covers the great advances in statistical techniques that have marked the last decade. It is a beautiful synthesis that will reward the experienced expert in the field with new insights and new connections in the form of historical notes that are not well known. And it is well-written and clear enough that even the beginning student can follow it through. Before this book, you would have had to read Allen's book, Charniak's short book on statistical NLP, something on speech recognition, and something else on generation and translation. Like squeezing clowns into a circus car, Jurafsky and Martin somehow, improbably, manage to squeeze this all into one book, but in a way that is elegant and holds together perfectly; not at all the hodge-podge that one might expect. I expect that this book will be seen as one of the landmarks that pushes the field forward.

It's worth comparing this book to the other recent NLP text: Manning and Shutze. Jurafsky and Martin cover much more ground, including many aspects that are ignored by Manning and Schutze. So if you want a general overview of natural language, if you want to know about the syntax of English, or the intricacies of dialog, if you are teaching or taking a general NLP course, then Jurafsky and Martin is the one for you. But if your needs are more focused on the algorithms for lower-level text processing with statistical techniques, or if you want to build a specific practical application, then Manning and Schutze is far more comprehensive and likely to have your answer. If you're a serious student or professional in NLP, you just have to have both.

This book is by now an accepted classic in the field. It is basically the only textbook that covers so much of computational linguistics, so I have had no choice but to use it for the past several years. Just the same, I'd rather not use it for teaching linguistics students. While the book has much to offer the professional, including a broad range of topics extensively researched, it is much more useful in this "handbook" capacity than as a textbook for the uninitiated. The chief reasons for this are: 1) It is pedagogically very poor; the majority of concepts are either explained in a confusing and obfuscatory manner or are not explained and are simply left in algorithmic form. This is not usually edifying to the linguistics student with no computer science background. 2) There are too many mistakes in its algorithms and method overviews. So far as I can see, even the famed Earley parsing algorithm is wrong here, it will not yield the correct output. 3) It is not written in a language that linguistics students can understand. With no background in mathematics, computer science, or pseudocode, such students need much more coddling than is provided by this book, and they are virtually unable to read it. Basically, as the title to this review states, what is called for now is a book to explain the contents of this book. Perhaps if my students keep encouraging me to write it. . .

This book is a great general introduction to NLP, covering a broad range of topics. Unfortunately there are many errors in the mathematical formulae and the algorithm descriptions, so do make sure to download the errata list from the book's home page.

The authors have the challenge of covering a vast area, and they do a good job of highlighting the hard problems within individual sub-fields, such as machine translation. The availability of an accompanying Web site is a strong plus, as is the extensive bibliography, which also includes links to freely available software and resources.

Now for the negatives.

While I would still buy and recommend this book, you will need to supplement it with other material; in addition to the accurate "broad and shallow" comment made by another reviewer, I would add that much of the material, as presented, is aimed at the comprehension level of a computer-science PhD and doesn't really meet the definition of a textbook for either undergraduate or graduate students. It is not that the material is intrinsically difficult: one recurring problem in the book is the vast number of forward references, where a topic is introduced very briefly but not explained until 20-50 pages later. In most cases, if you don't understand a passage in the text, I would advise that you keep skimming ahead - you may be rewarded because in several cases, the book covers a particular approach for 2-3 pages before telling you that its underlying assumptions are flawed, and that modern methods for addressing the problem use alternative approaches.

In other cases, the authors try to explain topics that might deserve entire chapters in about ten lines - a poster child is the explanation on page 736 of how Support Vector Machines can be used for multiclass problems. To someone who is familiar with SVMs, this material is unnecessary, while those who are not will not be enlightened by knowing that SVMS are "binary approaches based on the discovery of separating hyperplanes". I understand that this is not a text on machine learning approaches, even though machine-learning approaches have revolutionized NLP, but if the authors are clearly in no position to do justice to a particular topic in limited space, I would have preferred that they do the reader the courtesy of acknowledging the same, and simply point to a useful source, preferably online. (While the Wikipedia entry on SVMs is, as of this writing,incomprehensible to non-Math PhDs, the 2nd Google link, at www.dtreg.com, provides a reasonable overview.)

On the other hand, in a book that has to cover a vast area in limited space, there is a surprising amount of repetition. The page-long explanation of F-measure, a statistic used to evaluate the accuracy of a method, is repeated in three places almost verbatim, on pg. 455, 479 and 733; the repetition 24 pages apart (in chapters 13 and 14) should be considered astonishing given that the same author in the two-author collaboration clearly wrote both passages.

Finally, given the way algorithms are described - some reviewers point to errors in some of the descriptions, but I can't verify this - you would be hard-pressed to complete many of the exercises that follow each chapter, in terms of being able to implement a working program.

A final word of advice to the authors: I really do want to see a Third Edition, but I would recommend that you beta-test your material on a sample of your target audience, and incorporate their feedback. When you write a textbook, you really need to make a serious effort to communicate: if smart undergraduates or grad students tell you certain material is hard to follow, the fault almost certainly lies with you and not them.

GENERAL IDEA: Broad coverage, it lacks depth and details - particularly practical details. That is, the presentation is often sketchy, mainly because it approaches too many subjects for its available space. I would not say that this book is strong on theory either. It is quite obvious that it avoids getting too formal and precise, probably to remain attractive for non-specialists too.

CASE STUDY: One specific problem I had with the Hidden Markov Models, that are supperficially presented (or spread I could say) in several separate sections of the book, so it's not been a pleasure trying to actually understand them properly and completely as a fundamental concept, to make them work in my particular application.

TITLE: The book's title IS misleading because it starts with "Speeech" and this book's main subject is not speech but (written) language. Actually there are only a few chapters on speech.

CONCLUSION: Get this book if you are looking for a good overview of the field. The book will introduce you to a thousand of topics. As soon as you need in-depth coverage of some particular topic, you will look for additional resources.

I started reading James Allen's Natural Language Understanding to get background information on an NLP indepedent study project. The book was good, but I still found some points unclear and turned to Jurafsky/Martin for more information. In the end I found Jurafsky very comprehensive and much more down to earth than Allen. (They make useful references to popular movies and culture without sacrificing their academic reputation.) The work introduces basic NLP concepts as Allen does, but then presents applications that continually refer back to the methods. For example, Allen explains the Viterbi algorithm as a method for tagging sentences. Jurafsky/Martin present it, then refer to it in applications such as spell checking, voice recognition, and sentence tagging. The book also serves as a useful guide to finding the more significant NLP papers for further research. If you're interested in NLP this is an excellent place to start!

This book is a feat for anybody interested in Natural Language Processing and probably the most comprehensive book on this subject. It provides an in-depth overview of the most important aspects of NLP from regular expressions to sense disambiguation, discourse, and machine translation. I particularly like the bibliographical and historical notes in each chapter, which provide additional historical context and lots of references.

The book is well written and carefully structured. However, it contains several silly typos (real-word errors) that are a bit embarrassing, considering the topic of the book.

This book does not cover the hardware components of speech recognition. It only provides an introduction to the computational aspects. Nevertheless, I don't think the title is misleading (as other reviewers claim), but the back-cover should mention that it doesn't cover the electronic and signal processing components of speech recognition.

I read the first edition of that book and it is terrific. The second edition is much more adapted to current research. Statistical methods in NLP are more detailed and some syntax-based approaches are presented. My specific interest is in machine translation and dialogue systems. Both chapters are extensively rewritten and much more elaborated. I believe this book is perfect for everyone who starts in speech and language processing. With precision, coherent examples and some humor, this book give a great introduction into this topic as well as material for already experienced readers.

This book helped me accomplish what I set out to do; namely to obtain an overview of the field of natural language processing, with an emphasis on language understanding (as opposed to recognition). And I can recommend it on that level. The weakness of the book however is that it left me asking, "OK, now what?". The book started off strong with a number of dynamic-programming algorithms, finite automaton models, and N-grams that one could sink his/her teeth into from an algorithmic point-of-view. But when it came to actual techniques for natural-language understanding (chapters 14-17) the goods were not delivered. The algorithms disappeared, and the best I could find was in Chapter 15 an incomplete, and unconvincing treatment of Hiyan Alshawi's semantic parsing techniques which fueled the Core Language Engine last decade. Chapter 16 dealt with lexical semantics and was almost entirely devoid of algorithms.

My gut feeling after reading this text is that parsing techniques will likely give way to statistical and probabilistic learning methods that will in some sense bypass the need to correctly or accurately parse language. I cannot fault the authors for not exploring this in more depth,as this represents the cutting edge for both NLP and artificial intelligence. In any case, I'm off to read Schutze and Manning's book which will hopefully provide a bit more focus on that perspective. What intrigues me is that most people can understand some language, but very few people understand the grammar of their own language, especially if they have been deprived of a formal education. So why should computers need to know all about grammar rules and parsing? Could they instead be trained by simply being exposed to enough interactions between language and objects? I teach in a department dominated by both foreign and immigrant students. I understand them most of the time, but I would estimate that half the time their sentences or utterances would not fail to be parsed correctly.

As its lengthy subtitle suggests, this is a big book (just under a thousand pages) and unbelievably comprehensive. On the whole, the book is a major improvement over its predecessor. The first edition was plagued with typos on seemingly every page, and was also way too thin in certain places. I seem to remember them rushing through phonetics in a single page or two, and then describing optimality theory in just a couple sentences! The second edition's coverage of the field is significantly broader and deeper. Phonetics now gets a good 15 pages. The typos are gone and the appearance of the book is also much improved, with nice-looking black-and-white diagrams on nearly every page.

I have one pedagogical quibble with the new edition. The first edition introduced readers to the Bayesian noisy channel model by applying it to the problem of spelling correction, as implemented in the classic paper by Kernighan et al. Because noisy channel spelling correction is so fiendishly simple, and the paper is so readable, this was the perfect way to introduce a student to Bayesian models of language. In the second edition, however, the authors decided to jump straight into noisy channel POS tagging, a much more challenging topic, and to relegate spelling correction to an "Advanced" (?) section at the end of Chapter 5. They really should have started with spelling correction and then moved to tagging.

Quibbles aside, this book is a spectacular achievement. The first edition of Speech and Language Processing was a breathtaking synthesis of material, and it helped to unify the field of language technology, despite its flaws. This greatly updated second edition is a big improvement and will be the standard text in the field for years to come.