Searching EEBO


Last semester, when we explored ECCO, one of the concerns addressed was getting students acclimated to the product.  Even before students call up a digital facsimile of an eighteenth-century printed text and find themselves deciphering early modern fonts, they need to know how best to navigate an ECCO search.   Now that we have a free trial of EEBO, it might be useful to begin by discussing how best to help students become familiar with efficient ways to search EEBO.

A useful first step is to share with students Alice Eardley’s “video guide to using EEBO” on the University of Warwick’s web site.  The guide takes only seven minutes; at its conclusion, students will understand that EEBO holds both digital facsimiles and typed transcripts for some of those facsimiles, and they will have a working understanding of the icons associated with a given entry.  It’s the best introduction I have seen, and it calls attention to the value of video guides for presenting usage guides to electronic products.

I would be interested in hearing other responses to students’ initial contact with EEBO and to strategies they find helpful in beginning to make use of EEBO.


19 Responses to “Searching EEBO”

  1. Nick Says:

    I seem to remember the thing I found most frustrating when first encountering it was how its search engine worked. For anyone used to Google adding “and” automatically between search terms, getting lots of “No results” returns for a string of words when you know full well there are titles of that name is a bit frustrating until you figure out how it works…

    That is a very good video, and should really be required viewing for anyone starting out using EEBO.


  2. Eleanor Shevlin Says:

    Let me second the recommendation of Eardley’s video for a very fine introduction to EEBO.

    I will note that the first icon of which she speaks, the one that brings you the records tied to Philips’ name on Eardley’s screen, has a link to “Author page in Literature online.” Obviously, Warwick University has a subscription to Proquest’s LION (Literature Online), but not all institutions that have EEBO will also have this Proquest database. In addition, the transcription link that she later mentions does not appear for this entry when I access the site from my library subscription site.

    In other words, we see that not all subscriptions will necessarily have the same features, and these differences will affect different capabilities. For instance the lack of a transcription file will affect one’s ability to search.

    If one were to use Eardley’s video as an introduction (which generally seems worth doing), I would be prepared to point out the differences with the features offered at Warwick and those present in one’s home institution.


  3. Anna Battigelli Says:

    The full-text icon only appears for pieces that the TCP has transcribed. And if a school does not subscribe to TCP, the icon does not appear. In our trial version, accessible through this site, the full-text icon does appear.


  4. Eleanor Shevlin Says:

    Thanks, Anna. I should have specified that the TCP efforts are responsible for the transcribed texts and that our trial does offer access to the TCP. The advantages of having the TCP transcriptions and the search capabilities they afford are clear. While lack of access to the transcribed texts alters somewhat how one might employ EEBO in an undergraduate classroom, EEBO still can significantly enrich the undergraduate classroom experience (not to mention the graduate course as well).

    In earlier postings I have noted the ways I have used EEBO to discuss representations of the “Orient” by English writers to broaden an understanding of Montagu’s Turkish Embassy Letters. It has also proven useful in examining Dryden’s Absalom and Achitophel as part of a larger exchange about the succession crisis. So, too, has it been invaluable in teaching Dryden’s “MacFlecknoe.”


  5. Anna Battigelli Says:

    Perhaps it’s obvious to mention this, but to show the value of Eleanor’s example of using EEBO to teach Absalom and Achitophel, all one has to do is type in “Absalom and Achitophel” under “title” and up come 40 records, and though many of these titles are to editions of Absalom and Achitophel and to other Exclusion Crisis poems by John Dryden, others are titles of vituperative responses that Dryden’s poem elicited. A beginning bibliography of responses to the poem can be built from this search. (This might be a useful moment to introduce the value of the ESTC, to be explored more fully during a subsequent class.)

    The benefits of calling this search to students’ attention include introducing them to the competitive political/literary sphere of Restoration culture and specifically to the high stakes game Dryden engaged in when he published his most famous satire. The poem was not simply a literary event, as the responses demonstrate. Neither can the political moment be understand apart from its literary components.


  6. Eleanor Shevlin Says:

    Absolutely, Anna.

    The differences in exposing students to this broader context in the pre- and post-EEBO environment are perhaps also fairly obvious but may be worth recounting. Before EEBO my students used ESTC (on CD-Rom at the time) and the microfilms of Early English books. I usually had this assignment as part of a series of group presentations, and thus just three or so students engaged in the process–those assigned to the presentation on Dryden’s Absalom and Achitophel. With access to EEBO I at first continued to use it as a presentation assignment, but unlike the time when we were relying on microfilm, a number of students who were not involved in the presentation were inspired to seek and explore these exchanges on EEBO on their own for their formal papers. I have since conducted this assignment as an activity involving the whole class–with good success.


  7. sng6 Says:

    Forgive me for breaking into this discussion of EEBO with a comment about ECCO.

    This is a general posting to the list for anyone who has access to ECCO. Something peculiar has happened in my searches on ECCO recently: the results are not what they used to be. Indeed, the results on any search are one-half to one-sixth of what they used to be. To give an example from my research on quotations of Hamlet, on Feb. 17th, 2010, I searched in the “entire text” mode for /rotten N4 state/ for 1701-1750 (all boxes including texts with no date were checked) and received 54 ECCO hits, some of which I recorded because they were of interest to me. So today, Feb. 28th, I did the exact same search and received only 20 ECCO hits for that period (and only 76 hits for the entire 1701-1800 database). I checked through the particular 20 ECCO results for 1701-1750 and noted that two of the 8 specific texts I had recorded from the Feb. 17th search were now missing: The London Magazine and monthly chronologer, for 1736-1746, and A letter to a member of this new Parliament, from a true lover of the liberties of the people, from 1742. Nor, when I searched for these titles in a separate title search, did I get these volumes: “London magazine” searched for 1701-1750 (in “title” mode) received 0 hits, and, while “new Parliament” for 1701-1750 received 16 hits, that “Letter to a member” recorded earlier was not among the results.

    In fact, every search I have done now gets far fewer results than before. Last September, when running a test for /antic* disposition/ I received 24 texts; now I get 4. I used to get 66 texts for /prophetick soul/ (the preferred 18th-century spelling), and now I get 18 texts. For a broader view of the problem, I went back to data from that EC/ASECS presidential address I gave in 2006 about ECCO: in doing the research for this, I put in dates for each decade of the early 18th century with no words at all in the search box to see how many texts were in each decade. That search does not now get the same results either. That number should have grown because of the addition of ECCO II to ECCO, but it has shrunk. 1701-1710 (and nothing put into the search boxes) got me 10438 hits then; now it gets 6557 hits. 1711-1720 got 11776 hits in August 2006; now it receives 6164 hits.

    The problem does not seem to be with the University of Pittsburgh interface (which does sometimes give me trouble), as Anna tried these searches and got my diminished results. The problem is with ECCO itself, it seems. Does the database actually consist of multiple servers and have some of these, or the links among them, broken down?

    If anyone has suggestions, they would be appreciated, as the time draws near for the Albuquerque panel, any my presentation very much depends on having a stable database to analyze.

    Thanks for your help.

    Sayre Greenfield
    Professor of English
    University of Pittsburgh at Greensburg


  8. Anna Battigelli Says:

    I am reprinting Scott Dawson’s response to Sayre below. Others have confirmed Sayre’s recent findings. Scott’s response is interesting:

    Gale did do a release of the ECCO database late last week to add the Library of Congress subject headings (and ability to search them) and the fully expanded title fields. We will look into what this release may have inadvertently done to the indexing to cause such a change in results that are returned, which seems to be related to the date field. I’ll get back with details as soon as I can. If others experience similar issues, please let me know.

    Scott Dawson
    Product Manager – Gale Digital Collections


  9. Eleanor Shevlin Says:

    Sayre’s query reminds me that we never cleared up questions about variations in search results that cropped up this past fall. At one point, we believed the diffferences stemmed from those using the new interface and those not.

    We do not have ECCO II at my institution, so that may explain why I received just 19 hits and not 20 for Sayre’s search for / rotten N4 state/ for 1701-1750.

    My search (entire document, no date restrictions) for /prophetick soul/ yielded 74 hits, and when I repeated the search with limits, I received 17.

    My search (entire document, no date restrictions) for /antic* disposition/
    resulted in 91 hits, but with the 1701-1750 date restrictions, I do receive 4.
    This past fall Sayre and I (and others) ran the same searches yet received different results. Scott Dawson has obviously been alerted to this situation, but I will also email the technical person with whom we were in touch, both to inquire about this problem and to see what response our last queried generated.



  10. Anna Battigelli Says:


    If you run Sayre’s initial search and click “text with no dates,” you’ll probably come up with his current number of 20 hits, rather than 19. But that raises the question of the stability of searching in ECCO. Scott is working on that question.

    Beyond the question of a given search’s stability, these questions also point to the bibliographical integrity of such searches. I’m hoping that 18thConnect will offer some help with this question. If I understand Laura Mandell’s plan correctly, her team will transcribe ECCO texts (just as TCP is transcribing texts for EEBO). By transcribing texts, 18thConnect will help ECCO offer true (as opposed to approximate) full-text searching. This will immeasurably enrich ECCO as a scholarly tool.


  11. Eleanor Shevlin Says:


    I did run the search with “text with no dates” on; nonetheless, I received 19. I did so twice. Last fall we had some discrepancies with results, too.

    Because Scott had suggested in the fall that we contact the tech person, I went ahead and did so with this current issue, noting that Scott might have already been in touch about it.

    And, yes, these issues do underscore issues involving bibliographic integrity. TCP, I believe, is working on ECCO as well as EEBO, but the work on EEBO is much further along. As part of its plans for transcriptions, 18thConnect has partnered with National Center for Supercomputing Applications (NCSA), Gale Group and the IBM World Community Grid to develop OCR technology that can read early texts far more effectively, with the aim of easing the burden of retypying manually.


  12. Anna Battigelli Says:

    This is becoming odder and odder. I’m hoping the Gale tech.
    person will provide some answers.


  13. sng6 Says:

    Eleanor, Anna–
    Eleanor is correct: her 19 hits for /rotten N4 state/ (1701-1750) as opposed to Anna’s and my 20 results, comes because she does not have ECCO II: I checked through the 20 results, and indeed one and only one of these was from ECCO II.
    Other different results, such as those for /prophetick soul/ may come because Eleanor is searching the entire database (wherein I now get 81 hits with ECCO I & II) whereas I may not have made clear my lesser number was for the first half of the century.
    In any case, I’m temporarily suspending my Shakespeare research in ECCO until the mystery of the incredible shrinking database is cleared up.
    I’m also trying to go back through the December posts on TCP, but I’m not quite sure I can wrap my mind around this project and what it entails for ECCO.
    Certainly one would wish for more accurate OCR in the ECCO database, though in Albuquerque I’ll present some partial fixes for misread letter combinations in ECCO that my undergraduate research assistant has been working on. She was the one who first noticed something rotten in the data of ECCO, by the way. However, I can’t quite complete that research until the database is back to normal operations.


  14. Eleanor Shevlin Says:

    I have just heard from Gale’s technical investigator, and he is working on sorting out this issue and resolving the problem. He hopes to have a response for us soon.

    In the meantime, the previous emob discussions on search issues for ECCO can be found here.


  15. Anna Battigelli Says:

    It would be great if Scott were prepared to address these issues at the ASECS session. We need to hear how an ECCO search operates and what variables affect it (interface, ECCO I vs. ECCO I & II), etc. So far, we have raised questions using very specific examples, thanks largely to Sayre and Eleanor. I’m hoping that Scott will be willing to provide an insider’s perspective to these questions and others.


  16. Eleanor Shevlin Says:

    Yes, I am actually grateful that these issues have come up now, so we can address them at our ASECS session. In preparing my remarks, I had intended to see a conclusive answer to our fall dificulties, so these new developments have prompted me to write the tech person sooner.



  17. Eleanor Shevlin Says:

    Scott Dawson Says:

    March 9, 2010 at 8:13 am

    The issue with date searching in ECCO has been fixed in a more timely manner than previously anticipated and has been released so you should no longer encounter that issue. One exception are items with no dates (~37 items) are being picked up in a date range search even when you do not check the box. We will continue to work on that issue but wanted to get the primary issue addressed as quickly as possible. Again, sorry for the inconvenience.

    Scott Dawson
    Product Manager


  18. sng6 Says:

    Dear Everyone (forgive me for posting this twice, but this chain has gotten onto two separate headings):

    ECCO is back, and I’m very happy about that, but the results compared to earlier searches are not 100% the same. They are “close enough for government work,” as they say, but not identical to those before the new installation. Specifically, let me use the searches mentioned in my long post where I announced the problem. All of the searches below are for 1701-1750:
    Feb. 17th, 2010: rotten N4 state (no dates box checked) had 54 hits;
    March 9th, 2010: now it gets 55

    Sept. 2009: antic* disposition (no dates box checked) had 24 hits;
    March 9th, 2010: now it gets 21 hits

    Sept. 2009: prophetick soul (no dates box checked) had 66 hits;
    March 9th, 2010: now it gets 56 hits

    Let me add one other, a more massive search. On Jan. 15th, 2010, a search for passion (set for just English language) received 19386 texts but now, on March 9th, it gets 18883 texts.

    So this fix is pretty good, but not exactly the same as before.


  19. Eleanor Shevlin Says:


    Many thanks. The shifting results you have received are important to know. For some projects especially, the differences in hits may not be “close enough for government work”. For example, in the case of the “prophetick soul” search, the results vary by 15%.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: