 |
| The on-line AND: user interface and features | |
|
|
Wherever Screen appears in the left hand margin of an item below, you may click on the link to see, in a separate browser window, a screenshot illustrating the feature described. Most of these screenshots were created using Firefox 1.5 on Linux at a screen resolution of 800x600, though in a few cases where it was preferable for demonstration purposes to spread the windows over a larger area, Firefox running on Windows at 1280x 1024 resolution was used. If you find the text in some of these screenshots initially illegible, that will be because your browser is shrinking the image to the size of your current window. You may want to consult your browser help file to discover how to display the image at its native resolution (which can generally be done by a single mouse click in the correct place) |
|
TABLE OF CONTENTS SECTION A The basics
SECTION B More advanced searching
SECTION C Additional features
SECTION D SECTION E |
|
If you wish, you can locate and browse entries in the electronic AND more or less as you would in a paper edition: that is, peruse a list of headwords till you find the form you are looking for, then read the entry associated with that headword. |
|
| Screen | When you first go to the Dictionary interface, you will see on your screen the letters of the alphabet laid out in a horizontal bar. |
| Screen | Put your mouse pointer over the initial letter of the form you want to look up. A list will drop down, showing the first and last forms of the segments into which the headword list is divided. For letters relatively rarely attested at the start of A-N words, there will be only one segment listed, covering the whole range of forms with that initial letter; for some other letters there are six or more segments. Identify the segment where the form you are looking for belongs, pull the mouse slightly towards you to move its pointer on to the list, select the segment you want, and click. |
| Screen | The wordlist segment you chose will now appear on the left of the screen. Scroll to the actual form you are looking for, and if you find it, click on it (if, however, you don't find it, move to part 2 below). You will notice that some of the headwords in this scrolling list are white, while others are yellow. The white items take you directly to an entry whose lemma or lead headword form is identical to the white item in the list. The yellow forms indicate wordlist items which in the printed version of the Dictionary appear as pure cross references, pointing to a substantive entry elsewhere in the Dictionary, with a different spelling for the headword. In the electronic version, clicking a yellow item automatically follows the cross reference for you. This means that the entry actually displayed will not have the form you clicked on as its lemma, but it will be the entry to which the print edition would have referred you. Normally, "your" form will be listed among the variants after the lead headword, and be picked out in red, to help you see how and why you got here. In some instances, such "pure cross reference" AND headwords point to two or more entries. In that case, the server will send you all the entries, one below the other, and will precede them by a message alerting you to the fact that there more than one entry is to be found on your display. (Some pre-release testers have expressed a little irritation that in such cases a sometimes lengthy copyright and acknowledgement notice separates each and every entry in the set, rather than appearing once only at the very end. This is, however, an unavoidable reflection of the somewhat complex intellectual property situation, as explained on the copyright information page.) |
| Screen | Suppose you can't find the form you are looking for in the headword lists produced by the method above. That doesn't mean it isn't in the Dictionary. To reflect the orthographical vagaries of Anglo-Norman scribes (and the variety of practices of editors in different periods) the Dictionary follows many headwords with a number of variant forms. To include all these variants as pure cross references in the scrolling headword list would make those lists unmanageably long for users, so only a selection of such variants figures there. But you can locate any variant given in any entry by using the search field at the top of the Dictionary display. Type the form you are looking for (for example, chao) into the box (ensuring that the small square next to the "Regex" label isn't ticked, as it won't be unless you have previously ticked it yourself, as part of more advanced search methods explained later) then click the Search button. |
| Screen | If the form you typed in is found as a variant in one or more entries, you will see the headwords of those entries listed on the left of the screen. and you can view any of them by clicking the link. Notice that this list, like other lists of selected, probably non-contiguous, headwords you have asked to see, whether by this method or by others described later, has a white background, whereas lists of alphabetically contiguous headwords have a blue background. |
| Screen | If the form you typed in is not found as a variant form in any entry, then the system will attempt to find it in one or more citations. If it succeeds, the citations concerned will be shown in the main Dictionary window, with the matching form highlighted, and a link to the headword for the entry where the citation is found. |
| Screen | Often, if your match has failed because the form was an inflected version of a lexical item listed only by its root form, you will be able to see the appropriate entry among the citations and select it for full display from there. If you do so, the entry concerned will open up in a separate window, not in the main display area (so that you can select more than one link displayed in that area without losing track of your "place"). If, however, you would prefer to have the entry loaded into the main display area in place of the list of citations containing your original search term, then simply type the headword form into the search box and click search again, then choose the headword from the list that will appear on the left-hand side. |
Of course, there is no
reason whatever why you shouldn't make this "type in the
form and click Search" procedure your standard method
for finding any entries, without first selecting and browsing
a headword list. If
you prefer this approach, you can always at any time switch to
browsing by headword, either by using the letter menu bar above
each entry displayed in the main window, or by using the Synch
Wordlist to the Entry feature
described below. The Dictionary interface has been designed to
allow users easily to mix and match these two approaches depending
on the task at hand: they are presented here sequentially and as
apparent alternatives merely to help newcomers orient themselves. |
|
The only way you can hope to find things systematically in a printed dictionary is via looking up headwords. Yet in a scholarly dictionary with a large number of citations, this means that many instances of forms or usages that may be of interest can go unnoticed, because they are located in a citation under a quite different headword. This limitation does not apply to the AND as an electronic dictionary based on underlying XML structural markup (which, to put it somewhat over-simply, means the computer can tell what the various parts of an entry are, and hence focus a search solely on French words in citations, or English words in translations etc). In addition to letting you find variants as well as headwords by the method just described, the on-line AND also lets you locate forms in citations, no matter under what headword(s) the citation(s) occur. |
|
| Screen | To do this, you need to bring up the Concordance of AND Citations interface. You can get there either from the link on the Home Page, or by choosing the appropriate option under "Site Index" to the left of the letters on the headword selection bar. |
| Screen | If you want to locate all instances of a specific
form in the citations, simply type it into the field provided and click
the Fetch List button. |
| Screen | If
the form is found anywhere in a citation, you will see it listed
to the left of the display, followed by a number in square brackets
indicating how often the form occurs in the citations. |
| Screen | Click
the listed item, and in the main area of the display you will see
all the citations concerned, with the form highlighted. |
| Screen | If
there are more than two citations to display, they will be listed
initially in order of the headword under which the citations occur.
If you go to the very end of the displayed citations, you will find
a set of option buttons that let you alter the sort order, then by
clicking the adjacent Re-sort button cause the entire list to be
resorted according to the new criterion of your choice. As well
as the initial sort by entry headword, you can sort by source siglum
(which brings all attestations from a given source together on the
display) or by one of four co-occurrence points surrounding the actual
matched form, namely by the form which is one or two words away either
to the left or right of the actual matched term. |
Of course, given the relative brevity of
AND citations (especially those from the AND1 stratum) it quite often
happens that there is no form in the specified position, in which case
such entries sort to the top. If at the specified co-occurrence point in
the citation there is an ellipsis mark, this citation too will be treated
as though the matched term had nothing before or after it, as the case
may be. |
|
But a more powerful use of the concordancer is to obtain lists of forms which are not identical, but share common affixation, or maybe have differing affixations on a common root. To do this, you will need to go to the top of the screen and select something other than the default "matches" criterion. |
|
| Screen | Suppose we want to see all forms in citations beginning "tres". At the top of the concordancer window, click on the circle next to the "starting with" option to activate it, then type tres into the text entry field, and click Fetch List again. This time the list on the left will tell you how many forms you have matched, followed by a full alphabetical listing of those forms, each with an occurrence count. You can click on any form in the listing to see the actual citations in the main display section (and once again, if you wish, change their sort order) |
| Screen | Or we might want to see all forms ending in "ement". This time, we activate the "ending with" option, type in ement and click Fetch List. Then we can proceed as before to review the list and the counts and select for inspection the citation(s) containing any of the forms listed. |
Finally, if we select the "contains" option, then type in part of a form, and once more click Fetch List we shall get a list of all forms found in citations which contain somewhere within them (but not at the very start or very end) whatever sequence of characters we typed . |
|
That's by no means all we can do with the concordancer interface; but the further possibilities can't be explained until the use of regular expressions has been covered later on. Warning. The citations
concordancer does what it says: it locates, lists, and displays in
context only forms found in citations.
By deliberate design it ignores forms located
in headwords or headword variants, even if they match what you are
searching for. It
is therefore not a means of browsing the dictionary in any conventional
sense and is not intended as a tool of first resort for those with
only basic lookup needs. It will mainly
be of use to those who want to see additional attestations of forms
that are found in the citations of other entries, or who
want to examine the distribution of forms across the sources, or
look for particular co-occurrences. But it may also help users who,
maybe without fully realising it, have come across an inflected form
which consequently has no headword match: they are likely in such
a case to discover that the citations concordancer will locate the
form in a citation that leads them to the appropriate headword and
entry. |
|
| Screen | Occasionally, a user may be unable to track down a form in the Dictionary by any of the methods so far proposed, but may already have a rough idea from the context of what it might mean in English. Or perhaps a user would like to explore a particular semantic field that can be designated by a single concept in English but is spread across several divergent forms in Anglo-French. Either of those cases can be met by using the facility to search English translations and glosses. Like the citations concordancer, this feature can be selected either from the site home page, or from the Site Index drop down menu that is present on all entries in the standard Dictionary interface. |
| Screen |
|
The examples of using the headword/variants freeform search feature in section 2 above all entailed typing in the exact form we wanted to locate. But in any language, and when trying to locate Anglo-Norman forms in particular, we would often like to search while making allowance for various spellings of the same lexical item we are looking for. To make that possible, the search for headwords and variants can be switched into regular expression mode. It is not possible here to give full coverage of what regular expressions are and how to use them, but the following examples should be enough to get most users started who are not already well versed in their notation and application. One word of warning in advance. All the ANOH search facilities do indeed support full regular expressions: which means they do not support what some people confuse with regular expressions, namely the use of "wildcards" such as * to represent any sequence of characters or ? to represent any single character at a given position. Regular expression notation does of course allow matching any sequence of characters, or a single character at a given position, but not using the * and ? characters in the way just referred to. These two symbols are indeed used in regular expression notation, but with significantly different meanings. How dramatically those meanings can differ can be illustrated by a simple example. Suppose you want to see all the headword or variant forms in the AND beginning with the letter "y". There are in fact currently 14 such forms, all but two of them found under headwords not beginning with that letter, so you would expect 14 "hits". If however, you attempted to use the "wildcard" notation y* to match them, you would be in for a surprise. The AND system interprets y* as a regular expression, and that expression actually matches every single headword and variant form in the Dictionary, so you would get over one hundred thousand results (or you would, if the server permitted queries capable of thus swamping its resources and your browser with unwanted data: in fact, such a query would return simply a polite request to reconsider and reformulate). The correct regular expression query corresponding to the wildcard notation y* is ^y (or ^y.+ if the letter "y" on its own is to be excluded from matching). What the symbols in that expression mean, and why in regular expression syntax y* matches absolutely everything, will be explained below. |
|
| Screen | To see the effect of turning on regular expressions, we can compare the effect of searching for the same input term first without regular expression support (as above) then with regular expressions turned on. A search for cas without regular expressions active returns the exact form we searched for, reporting that that form occurs as headword or variant in the six listed entries. |
| Screen | If we now click on the square box to the right of the Regex label to turn on regular expression support, then search again on the same term, we get 86 matches (at the time this document was written). A closer look at the list of matches shows what has happened. The server has found all the headword and variant forms that contain the letters "cas", in that order in any position. So we still see in the list "cas", reported as figuring either as headword or variant in the same 6 entries, but we also find matches on forms beginning with the letters "cas" (casal, cascun etc.), as well as on forms which have the sequence "cas" either somewhere within them (falcastre, publicasioun etc.) or at the very end (becas, carcas etc.). This illustrates that, in regular expression mode, the characters we type in, if they do not match the start of a headword or variant, will be "slid along" the headword and variant forms to see if a match can be found further along. |
| Screen | We can prevent this "sliding" behaviour of the match by "anchoring" it to the start of the target forms. We do this by preceding it with the symbol ^. So if, after ensuring regex mode is still active, we alter the search term to read ^cas, we see that the match count falls from 86 to 67. All the forms that didn't start with "cas" have now no longer been matched, but those that start with the letters "cas" followed by further characters are still in the list, because we have "anchored" only the start of our search term, and so any form beginning with "cas" satisfies our search criterion. |
| Screen | There is also a regular expression symbol that "anchors" our matches to the end of the target words, the dollar sign or $. So if we delete the ^symbol before cas and place instead a $ after it, our regex search on cas$ matches only 5 forms, cas itself, and the four further forms ending with cas that were part of the much longer listing when we applied cas as a regular expression without any anchoring. Predictably enough, if we anchored both ends of our search expression, making it ^cas$, we would get the same results as we saw with regular expressions turned off altogether, i.e. we see the one form cas, with the six entries where it can be found duly listed again. |
| Screen | Supposing we would now like to match forms which differ from one another only by one character at a particular position. In regular expression syntax, the symbol used to match one single character, no matter what that character is, is the period or "full stop". If in regular expression mode, we enter ^b.isser (that's a start of word anchor, a letter "b", a period to stand for any letter, and the letters isser) as our search term, the server will respond with a list of 5 matched forms: baisser, baissere, beisser, boisser, brisser, confirming that that period symbol has here matched, as the second character, an a (twice), along with an e an o and an r. |
| Screen |
But maybe the period symbol is too general for
the match that interests us: perhaps, for example, we are interested
only in those terms where the letter before the "i" is a
vowel. We need to replace our "match-any-single-character" period
symbol with a notation that says "match one, and one only, of
a specified set of characters". To achieve this with our example,
type ^b[aeiou]isser into the search term field. In place of the single
period, we now have a list, enclosed in square brackets, of the letters
that are permissible matches at the second position (in this example,
any one of the vowels). This time the server replies with the forms
baisser, baissere, beisser and boisser. Because we used what is called "character
class notation", indicating by a sequence of characters within
square brackets that we wanted to see only forms where the character
in second position was either a e i o or u, the form brisser is no
longer matched. |
| Screen | The character class expression [aeiou] can be understood as a form of alternation (match either "a" or "e" or... etc), but it is limited to single character alternatives. If any or all of our desired alternative matches contains more than one character, we need a different notation. We separate our alternatives by a vertical bar symbol "|" (this symbol wanders about the keyboard on different machines, but it is there somewhere, above the \ symbol on UK and US PC keyboards), and surround the group of alternatives with parentheses. So to locate all forms ending with either "ant" or "aunt" we query for (a|au)nt$ (Note that we no longer want the ^ start anchor but instead need the $ end anchor because that is where we want to "pin" our matches). The server responds (at the time of writing) with 893 forms matching this pattern, and we can see close to the top of the resulting list that it includes both adamant and adamaunt, confirming that our search has been correctly carried out. |
Instead of viewing the query we just performed as searching for either a or au followed by "nt" at the end of target forms, we could think of it as a request to locate in final position the sequence "aunt" where the "u" need not always be present. Conceived of in this way, we need a notation to denote the optional presence of a single specified character at a given position in the match, in this instance an optional "u" after the "a". This is done by following the optional character immediately by a question mark, so that another way of achieving the same results as (a|au)nt$ is to query instead for au?nt$. The result will be exactly the same list of forms, 893 in the state of the Dictionary at the start of March 2006. |
|
We could combine the alternate group and optional character notation to extend our search to locate forms which end in anz, antz, anz, aunz, auntz or aunz with the single expression au?n(tz|t|z)$ – currently matching 1028 forms – but at this point anyone new to regular expressions would probably do better to pursue the details elsewhere. AND searches support the full repertoire of regular expressions as found in programming languages with advanced text handling, such as Perl, Python and Java. If you want to experiment with regular expressions for AND searches, the section on regular expressions in any primer of one of those languages is as good a place as any to look for information and ideas. |
|
|
Because an expression like y* would
produce somewhere in excess of 100,000 completely useless matches,
the server will not attempt to process it, but will respond with
a message pointing you to this page for an explanation. To reiterate
the answer given above, which may now make rather more sense: the
correct way to create a list of all forms beginning in "y" in
regular expression mode is to query on either ^y if you would regard
the form "y" itself as "beginning with y ", or
on ^y. (where the period after "y" matches
and therefore requires a following letter) if your definition of "beginning
with y " includes at least one letter after the initial "y". |
|
| Screen | Those who have followed this explanation (or who know enough about regular expressions already) will probably have realised that the "starting with", "containing", "matching" and "ending with" options found on the citations concordancer interface are in effect just slightly more user-friendly ways of adding some basic regular expression notations to the form entered into the box. This feature is absent from the main Dictionary interface mainly to avoid clutter there. What may not be so obvious, though, is that the concordancer interface also allows users to employ regular expression notation in the terms they enter. So the original example in Section 3 above of using the "ending with" setting and the term ement to get a list of all forms in citations ending with that character sequence can now be extended: if the setting is left at "ending with" but the search term entered is, say, (au|e|a)ment (note that in the concordancer interface we don't add the dollar end anchor, because the "ending with" setting does that for us behind the scenes, but the server won't mind if you add it yourself by mistake), then our word list will include all citation content forms ending in - aument, -ement and -ament, 2662 of them at the present time. |
| Screen | As well as searching the citations via the wordlists which the concordancer builds for us, we can also locate citations in which two terms of our choice occur, in either order, and within a span of words which we can specify. This is useful if, for instance, we are interested in phrases rather than single lexical items. For such searches, we use the citations term proximity search interface, which is accessible either from the link labelled AND citations: Proximity Search on the home page, or via the drop-down menu on the main Dictionary display. |
| Screen | When the proximity search form first appears, it is set to query terms which are immediately adjacent (the word-distance box is initially set to 1) and where the term we type into the left hand box comes before the one we type into the right-hand box. |
| Screen | Supposing we want to examine instances of the two-word combination "tres bon ". If we type tres into the left-hand box, bon into the right hand box, leave the settings as they are and click on search, we locate (in the state of the Dictionary when this document was written) seven citations, which are displayed sorted by the alphabetical order of their source sigla, so that attestations from the same source are grouped together on screen. The display tells us that there are 347 citations containing the form tres and 1018 citations containing the form bon, but that bon immediately precedes tres in only seven of them, which it displays for us, along with links to the entries in which they are found. |
|
|
| Screen | However, any regular expression notation whatsoever entered by the user into one of the query fields causes the server to leave whatever the user enters into that field exactly as it is, evaluating the regular expression precisely as the user entered it. In other words, if I want to anchor the term bon on its left side only, so that it would match not only bon as a stand-alone form, but also bone bones bonté etc etc, then I can type into the right- hand box ^bon. This time, the response shows that the server has used my own regular expression in the right hand box (while continuing to package the plain term in the left hand box between start and end anchors as before). We notice that the number of citations with a match on our right-hand term has risen to 2623, and that increase has added five citations to the set that satisfies our overall conditions by additionally matching instances of the form "bone". |
| Screen | In the case of the form bon, the regular expression notation needed to extend the matches was fairly elementary. However, the nature of Anglo-Norman orthography means that for many other base terms, a more sophisticated deployment of regular expressions is needed to locate all the co-occurrences that are being sought. The associated screenshot illustrates the sort of notation required to capture the various manifestations of the word which in modern French would be spelled "mauvais(es)". All the symbols used in this example are explained in the section on regular expressions above. |
This interface is of course capable of locating items in much more flexible ways that the examples just given indicate, but the use of the "comes up to" field to alter the required word span and of the "before" and "after" buttons to change the order in which they must occur should require no further elaboration. |
|
|
|
Having brought up an AND entry on screen in the normal interface (that is, the one described and illustrated here: not all the features about to be explained work when an entry is summoned up in a free-standing window of its own) there are a number of things that can be done with it. |
|
| Screen | By double-clicking with the mouse pointer over any siglum following a citation, you can cause the full List of Texts entry for that siglum to pop up in a small window of its own. |
| Screen | That
window will also offer you a link to the DEAF entry for the same
item (if there is one) This is a good
way of accessing such information about the dating of a given citation
as the AND, with its primarily semantic focus, allows. |
| Screen | And
a further link in the List of Texts entry popup window will generate
a full list of all citations from the source in question in the
AND, with further links to the entries in which the citations are
to be found. |
| Screen | Where a citation is from a source which is one of the texts currently on-line elsewhere on the site, a small icon appears immediately after the citation text. If you click on that icon, you will be taken to the part of the text concerned where the citation is located: from there, you can if you wish page backwards and forwards through that text. |
| Screen | If you double-click on any Anglo-Norman form within a citation, the server will send back, in a popup window, the results of looking up the form you have double-clicked on in the Dictionary, followed by a list of all the citations in which that form occurs. You should bear in mind, however, that this feature has no semantic or morphological intelligence whatever: if treats the form you click on simply as a sequence of characters and looks for matches on that sequence, so will sometimes miss relevant items and return spurious ones. Nevertheless, it can be useful as a form of quick reference. Note that any entries retrieved in this way will not be added to the list of entries you have consulted which the server maintains for you (see next item). This feature was designed mainly to give users rapid access to the Dictionary when browsing the source texts on the site (although it will not be activated for the source texts until a later phase in the project, when more sources are on line). However, it has been applied to the Dictionary itself because some early testers found it was sometimes useful. |
| Screen | Provided the server has been able to establish a "session" with your browser (for details of what this means, and some reasons why it may not be possible, please see the "privacy" page) it maintains a list of the entries you have consulted, either by choosing them from a pick list, or by following cross references within entry bodies, for as long as your session lasts. (Your "session" generally comes to an end when you close down all copies of the browser you used to visit the ANOH site: so if you keep the browser open and go off to look at other sites, the session will normally still be active if you return later to the AND, although this does somewhat depend on what scope your browser gives the server for maintaining a "session".) At the top of every entry displayed in the normal interface, you will see a link labelled "List entries visited". Clicking this link will cause whatever wordlist you have displayed to the left of the screen to be replaced by a list of the entries you have viewed so far. You can then click on any item in this list to bring the entry back on to your screen. |
The lists generated by this feature have another special function: you may drag and drop them, or cut and paste them, on to your desktop or into other documents of your own in order to create persistent links to the entries concerned. But no other links (such as those in the normal blue-background scrolling wordlists) should be used in this way: unlike the "entries visited" links, all other links, although they work within the Dictionary interface will either not work at all, or will work only for a strictly limited time, if you try to use them as "shortcuts" or "bookmarks" on your local machine. For further details of this feature and its uses, please see the page on Linking to AND entries. |
|
It may be that you would like to keep the list of entries you have consulted during a session, re-using and adding to it on a future occasion when you consult the on-Line AND. Subject to the same proviso about your client supporting the maintenance of sessions referred to in the previous section, you may do this by taking the link labelled Bookmark this Session at the top of every entry displayed in the standard Dictionary interface, and following the instructions which will then appear on your screen. As those instructions explain, you can also use this technique to transfer a "session" and its list of entries visited, between two different machines (perhaps on in your office and another at home) or indeed between different people (by sending the url created by this facility in an email to the person with whom you wish to share your list). |
|
| Screen | |
You may be working on an activity where the AND is only one of several electronic resources you wish to consult in parallel. In such circumstances, especially if your main need for the AND is simply to look up individual A-N forms occasionally, you may find it more convenient not to go to the ANOH site and work in the full AND interface, but instead to have AND entries appear in a small window on your screen alongside your main work areas. That is what the AND lookup tool is mainly for. The method of fetching the tool and installing it your browser is described on the Lookup Tool page |
|
Every single aspect and phase of the digitisation and electronic publication of the Anglo-Norman Dictionary and its associated materials has been and remains an undertaking purely by scholars for the benefit of scholars, and for anyone else with an interest of any kind in the results of lexicographical research. No professional programmers or interface designers have been involved at any stage. This means that the features described here, both the way they look and the way they work, have been mainly shaped by long and varied experience of using dictionaries, both on paper and on screen, and teaching others to use them, rather than by any more general doctrines of application or interface design. We hope that those features will be assessed primarily by their practical usefulness and efficiency to people who need ready access to an authoritative source of information about Anglo-Norman lexis, whether they be philologists, students of literature, historians or members of the general public with an interest in the areas this site covers. For that reason, we especially welcome comments and suggestions from members of that broadly-defined constituency. If you think the result of our efforts so far is good enough for your needs and interests, please take a moment to send us an email, however brief, telling us so. If you think it isn't good enough, particularly if you have an idea about how it could be made better, we'd like to hear about that, too. You may send your emails to comments@anglo-norman.net, and we shall consider them all carefully and respond as appropriate in due course. Please be patient, however, if the response seems slow in coming. The very fact that we have placed that email link on this page for our readers' convenience means that the address it contains will be harvested by spam robots and so become the target of many hundreds of junk emails and virus payloads daily, and we may not always immediately spot your message among all the rubbish. But spot it we shall, eventually. |
|
 |