September 1, 2009
Google Mobile App: What Might It Mean for Legal Research?
Voice search may soon become a popular tool for legal research in the U.S. The reason? The Google search engine’s smartphone application, called Google Mobile Application or Google Mobile App (“GMA,” currently available for BlackBerry, iPhone, and Android phones), includes a “search by voice” feature that executes voice searches with surprising accuracy, at least for known legal resources. (Try it on a search for: Mapp v. Ohio; or for: 42 USC section 2000e.) Not only is GMA’s voice search feature quite good already; it is likely to improve over time, because, as Tim O’Reilly and John Battelle note GMA, which is cloud-based, utilizes machine-learning on a very large scale. The legal materials that GMA searches consist of a large (and growing) share of U.S. primary legal resources that are freely available on the Web and are indexed by Google. These include much U.S. federal case, statutory, and administrative law, as well as some recent state primary law, and certain well-known older U.S. state cases. Google also indexes a great deal of metadata describing primary and secondary legal resources, not freely available on the Web, from the U.S. and other jurisdictions. The result: GMA makes a substantial body of legal information accessible via voice search on smartphones.
How might GMA affect legal research in law firms and law schools? To explore answers to this question, this post discusses several types of information. First, this post examines the history of voice recognition in connection with commercial computer-assisted legal research (CALR) services. The post next considers factors suggesting the willingness of practicing lawyers, law professors, and law students to use GMA for legal research, including data from the 2009 ABA Legal Technology Survey Report (the “ABA Survey”). Finally, this post discusses how U.S. commercial CALR services might deal with a potential challenge from GMA. The post presents results from an informal survey of major CALR providers respecting whether they expose their metadata to Internet search engines, and whether those providers offer, or plan to offer, voice search.
In the U.S., the major CALR vendors appear to have first offered voice search in the mid-1990s. At the August 1994 ABA Annual Meeting, Westlaw premiered a voice search feature, using Kolvox Communications, Inc.’s speech recognition software, LawTALK. See James Milles, “Law Library Hi-Tech,” 25 AALL Newsl. 296, 297 (1994); “Southern Exposure, Annual Meeting Exhibition in New Orleans Will Showcase Latest in Legal Technology and Services,” ABA J., July 1994, at 66; David P. Vandigriff, “Talking Heads: Voice-Activated Legal Research Turns the Keyboard-Shy into Bold Pros,” ABA J., Sept. 1994, at 76. Shortly afterwards, Lexis/Nexis introduced a similar speech recognition feature for its CALR system, though I haven’t been able to identify the developer. See “Of First Impressions: New Products for Attorneys,” ABA J., Mar. 1995, at 89, 90; “Lexis Offers Voice Recognition Interface,” Law Prac. Mgmt., Jan./Feb. 1995, at 9, 56. By 1996, LawTALK and a competing speech recognition application, Dragon Systems, Inc.’s DragonLaw for Windows, were configurable for use with both Lexis/Nexis and Westlaw. See Virginia C. Thomas, “Digital Information Access for Law Library Patrons with Disabilities: Opportunities and Challenges,” AALL Spectrum, Oct. 1996, at 14, 15.
According to spokespersons for Nuance, Inc. (now the owner of the Dragon software, today called Dragon NaturallySpeaking 10 Legal, as well as the vendor for the number 3 U.S. legal speech recognition system [according to the ABA Survey], IBM ViaVoice , and MacSpeech (the developer of MacSpeech Dictate Legal the number 2 legal voice recognition system in the U.S. market), today none of the most popular legal speech recognition applications is specifically configurable for use with Westlaw, Lexis.com, or any other CALR system.
Nonetheless, the Dragon, MacSpeech, and ViaVoice legal programs can be used to interact by voice with most Web browsers; spoken key-control commands move the cursor, execute clicks, and perform other actions. (Dragon works with Internet Explorer, Mozilla, Firefox, and AOL; MacSpeech works with any browser that can process the Apple key commands; and ViaVoice works with IE and AOL.) Moreover, some computer operating systems, such as Windows Vista have built in speech recognition programs that enable voice search of Web browsers. In addition certain assistive software tools for visually-impaired persons, such as Code Factory S.L.’s Mobile Speak Pocket for Windows Mobile also enable voice search of Web browsers. (See, e.g., Bradley Hodges’s 2007 AccessWorld review.) Therefore, a number of speech recognition systems can currently be used for legal research by voice, in connection with the Web-based user interfaces of the leading U.S. CALR services.
Yet we don’t seem to know the extent to which these voice recognition tools are actually being used for legal research, whether in legal practice or law schools. For example, the ABA Survey doesn’t appear to report on the specific uses to which lawyers put voice recognition programs, and a recent search of the legal and library scholarly literature did not turn up any empirical studies of voice search use in the firm or law school environments. Empirical data on the use of voice recognition for legal research seems highly desirable.
Factors Suggesting the Potential of GMA for Legal Research Use
Several factors suggest that lawyers, law professors, law students, paralegals, and law librarians might be inclined to use GMA for legal research:
GMA appears to be extremely easy to use; quite accurate in interpreting spoken search statements, and likely to improve; and surprisingly effective for legal research, at least for known item searches respecting jurisdictions with substantial portions of their primary materials available on the open Web. The usefulness of GMA will likely increase as the amount of primary law available on the open Web grows. Further, the open access movement and the engagement of lawyers with social media will likely result in the availability on the free Web of more quality, free-of-charge secondary legal resources, such as the Cornell Legal Information Institute’s Wex legal encyclopedia, and legal articles freely available through scholarly repositories.
Smartphone use by U.S. lawyers is widespread. In the latest ABA Survey, 64% of responding lawyers reported using a smartphone or BlackBerry for legal work. Of these lawyers, almost two thirds reported using a BlackBerry, and 14% an iPhone. These lawyers seem likely to use GMA. Further, in the ABA Survey almost one fifth of lawyers reported regularly using smartphones to conduct legal research outside the office. (However, in another portion of the ABA Survey, only 2% of lawyers reported using PDAs, smartphones, or BlackBerrys for legal research outside of the office; the reason for this discrepancy, which ABA Legal Technology Resource Center is reviewing, is unclear.) Although I haven’t found current data on law professors’ or law students’ use of smartphones, recent statistics on mobile device use on U.S. university networks indicate very high rates of iPhone use, which may suggest similarly high rates of use among U.S. law students and faculty. Though the ABA Survey did not appear to measure smartphone use by paralegals or law librarians, anecdotal evidence suggests that such use is common.
In the ABA Survey more than 90% of lawyers reported using free-of-charge Internet resources in their legal research. Although problems with the accuracy and authenticity of free online legal information are well known (as discussed, for example, in the 2009 AALL Annual Meeting program on “Advocating for Authentication”; in the 2009 ABA Annual Meeting program “When Is a Law the Law?”; by John Joergensen in his recent VoxPopuLII post; and in the 2007 AALL summit and report on authentication of online legal information), for certain kinds of legal research, many U.S. lawyers appear to consider at least some legal resources available on the free Web “good enough” for the task at hand. Similarly, in many law school teaching and learning contexts, where the authenticity of legal information is often less important than in legal practice, one might expect substantial use of legal resources available on the free Web.
In the ABA Survey almost one quarter of lawyers reported beginning their research with an Internet search engine, and almost one third identified Google search as the free website they used most often for legal research. Although I was unable to locate current empirical data on U.S. law professors’ or law students’ use of Internet search engines for legal research, Dr. Stephann Makri and colleagues’ 2007 paper on UK law teachers’ and law students’ use of Google for legal research and OCLC research reporting very large percentages of all persons, and especially of college students, beginning their research with Internet search engines, lead one to expect quite high rates of Internet search engine use for legal research among U.S. law professors and law students.
- U.S. lawyers are already familiar with voice recognition software. In the ABA Survey, more than one fifth of lawyers reported that their firms had such software, and about 10% of lawyers reported using such software, though usage was higher among solo practitioners (almost 15%) and lawyers in firms of 2 to 9 members (12%). (See, e.g., JoAnn L. Hathaway’s new review of dictation programs; Mark Tamminga’s recent review of voice software for managing e-mail; Tonya L. Johnson’s recent article “Dictation Made Easy”; and Stephen Stine’s recent article on voice transcription services [scroll down to “Tech Corner”]).
Other factors suggest infrequent use of voice search for legal research in the law firm, however. For example, in response to one question on the ABA Survey, lawyers reported little use of their smartphones for legal research (though this figure appears to be contradicted by other ABA Survey findings, as noted above). In addition, although paralegals’ and law librarians’ smartphone use may be common, the ABA Survey reported infrequent assignment of legal research to paralegals or law librarians, with only about 6% of lawyers reporting that they regularly asked either paralegals or law librarians to perform legal research.
While no one can predict with certainty whether legal research via voice search using GMA will catch on with practicing lawyers, law professors, law students, law librarians, or law firm support staff, the factors identified above suggest that adoption of voice search via GMA for legal research is a substantial possibility. If GMA voice search takes hold as a popular legal research method, GMA’s success is likely to stem primarily from two factors: high quality voice recognition, and the exposure to Google’s search engine crawlers of very large amounts of metadata describing legal resources.
How are U.S. commercial CALR vendors likely to respond to the GMA challenge? Of many possible responses, two seem most probable. First, CALR vendors could allow Google to index the metadata describing the content of their databases, enabling that metadata to appear in Google results lists stemming from GMA searches for legal resources. (Since Microsoft might eventually offer a mobile voice search feature to compete with GMA, CALR vendors might also choose to expose their metadata to Yahoo! and Bing.) In this way, the CALR vendors could turn GMA to their advantage, as means of attracting new customers for legal information available for a fee. Another approach for CALR vendors would be to compete with GMA directly, by offering a voice search tool customized to work with the vendors’ CALR services, as Westlaw and LexisNexis did in the mid-1990s.
In part two of this post, we’ll explore the extent to which the major U.S. CALR vendors are pursuing these approaches.
Robert C. Richards
Respecting the two questions in the 2009 ABA Legal Technology Survey Report, about the use of mobile devices for legal research: Stephen Stine of the ABA Legal Technology Resource Center has kindly provided an explanation of the behaviors intended to be measured by each question, and an interpretation of the results for each question, in a new post at the Legal Informatics blog at http://bit.ly/tIXoq .
Posted by: Robert Richards | Sep 1, 2009 4:57:31 PM