root / Portal / configurations / html / Help_SrcmfProject.jsp @ 3
Historique | Voir | Annoter | Télécharger (50,69 ko)
| 1 |
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> |
|---|---|
| 2 |
<html xmlns="http://www.w3.org/1999/xhtml"> |
| 3 |
<head> |
| 4 |
<meta http-equiv="Content-Type" content="text/html; charset=UTF-8" /> |
| 5 |
<title></title> |
| 6 |
</head> |
| 7 |
<body> |
| 8 |
<div style="padding-left:25px;"> |
| 9 |
<h1>SRCMF corpus: TIGERSearch web interface</h1> |
| 10 |
<h2>Contents</h2> |
| 11 |
<ul> |
| 12 |
<li><a href="#interface">Using the TIGERSearch web interface</a></li> |
| 13 |
<li><a href="#query">Writing a simple query</a></li> |
| 14 |
<li><a href="#concordances">Exporting a concordance</a></li> |
| 15 |
<li><a href="#tags">Tagset used</a></li> |
| 16 |
<li><a href="#sample">Sample queries</a></li> |
| 17 |
</ul> |
| 18 |
<h2><a name="interface"></a>Using the TIGERSearch web interface</h2> |
| 19 |
<h3>Writing a query and browsing the results</h3> |
| 20 |
<p>In the TigerSearch tab, queries are entered in the top panel, and matching sentences |
| 21 |
are shown in tree form in the bottom panel. A tutorial on TigerSearch queries may be |
| 22 |
found in the section “<a href="#query">Writing a simple query</a>”.</p> |
| 23 |
<ul> |
| 24 |
<li>Type your query in the top panel (e.g. <tt>#pivot:[word = "Tristran"])</tt></li> |
| 25 |
<li>Click on the ‘Search’ button at the bottom right of the panel.</li> |
| 26 |
</ul> |
| 27 |
<p>If the query is well-formed, and if there are matching results in the corpus, the |
| 28 |
first tree in the forest will appear in the bottom panel.</p> |
| 29 |
<p>The central bar gives the number of matches and the position of the sentence in the |
| 30 |
corpus, in the form <i>sent: [sentence number] [match number] / [total matching |
| 31 |
sentences].</i> Note that subgraph navigation is not yet implemented, and the |
| 32 |
interface does <strong>not</strong> show the total number of matches, only the |
| 33 |
number of matching sentences. You can navigate through the forest of matches using |
| 34 |
the forward and back arrows on this bar. The ‘Export’ button displays the current |
| 35 |
tree as an .SVG file in the browser, which can be saved and downloaded. The ‘Export |
| 36 |
Concordance’ button allows matching sentences to be exported in <a |
| 37 |
href="#concordances">concordance form</a>.</p> |
| 38 |
<h3>Exporting the results</h3> |
| 39 |
<p>To export the results of your query, click the ‘Export Concordance’ button. An export |
| 40 |
window will appear, with the following options:</p> |
| 41 |
<ul> |
| 42 |
<li><p><strong>Type</strong></p> |
| 43 |
<p>Three concordances are currently implemented:</p> |
| 44 |
<ul> |
| 45 |
<li>basic concordance</li> |
| 46 |
<li>single word pivot concordance</li> |
| 47 |
<li>pivot and block concordance</li> |
| 48 |
</ul> |
| 49 |
<p>It is important to note that these concordances use the names of TigerSearch |
| 50 |
variables from the query to structure the concordance. <strong>No |
| 51 |
concordance will be produced if your query does not contain a |
| 52 |
<tt>#pivot</tt> variable.</strong> The pivot and block concordance |
| 53 |
requires at least one additional <tt>#blockXX</tt> variable.</p> |
| 54 |
<p>Further documentation for these concordances may be found in the section “<a |
| 55 |
href="#concordances">Exporting a concordance</a>”.</p></li> |
| 56 |
<li><p><strong>Context (number of words)</strong></p> |
| 57 |
<p>Sets the size of the context preceding and following the pivot.</p></li> |
| 58 |
<li><p><strong>Restore punctuation</strong></p> |
| 59 |
<p>Adds punctuation from the BFM’s digitized edition to the exported |
| 60 |
concordance. It will also restore words excluded from the TIGERSearch corpus |
| 61 |
(e.g. lacunae, AOI in the <i>Chanson de Roland</i>).</p></li> |
| 62 |
<li><p><strong>Properties to show in concordance</strong></p> |
| 63 |
<p>Select which features of terminal and non-terminal nodes should be shown in |
| 64 |
the concordance. This function is only active for the ‘pivot and block |
| 65 |
concordance’.</p></li> |
| 66 |
</ul> |
| 67 |
<p>When you have filled in the form:</p> |
| 68 |
<ul> |
| 69 |
<li>Click the ‘OK’ button.</li> |
| 70 |
</ul> |
| 71 |
<p>After a short delay, a new tab will open in your browser, containing the concordance |
| 72 |
in plain text tabular format (.csv).</p> |
| 73 |
<ul> |
| 74 |
<li>Save this file to disk using the ‘File > Save As...’ menu in your |
| 75 |
browser.</li> |
| 76 |
</ul> |
| 77 |
<h3>Viewing the concordance</h3> |
| 78 |
<p>To view and manipulate the concordance, you will need to use a spreadsheet |
| 79 |
package.</p> |
| 80 |
<ul> |
| 81 |
<li>Open the spreadsheet application.</li> |
| 82 |
<li>Select ‘File > Open...’ from the toolbar.</li> |
| 83 |
<li>Ensure that the file list is showing either ‘All files’ or ‘CSV text |
| 84 |
files’.</li> |
| 85 |
<li>Select the saved .csv file.</li> |
| 86 |
</ul> |
| 87 |
<p>You will need to correctly configure your spreadsheet software to read the file. We |
| 88 |
recommend using LibreOffice or OpenOffice Calc, which will prompt the user for |
| 89 |
settings whenever a .csv file is opened. The following settings are required for the |
| 90 |
import to function:</p> |
| 91 |
<ul> |
| 92 |
<li>Character set: Unicode (UTF-8);</li> |
| 93 |
<li>Separated by Tab (ONLY);</li> |
| 94 |
<li>Merge delimiters OFF;</li> |
| 95 |
<li>Text delimiter: NONE (empty box)</li> |
| 96 |
</ul> |
| 97 |
<p>Troubleshooting likely problems:</p> |
| 98 |
<ul> |
| 99 |
<li>If accented characters do not appear correctly > check the character set is |
| 100 |
UTF-8;</li> |
| 101 |
<li>If some rows do not seem to have the correct number of columns > check that |
| 102 |
Text Delimiter is set to nothing (the default is usually double quote, which |
| 103 |
will cause an error where the text contains double quotes), merge delimiters is |
| 104 |
OFF, and TAB is the only separator selected.</li> |
| 105 |
<li>If zeros appear rather than punctuation (unlikely) > use the ‘Fields’ section |
| 106 |
of the import window to set every column type to ‘Text’ rather than |
| 107 |
‘Standard’.</li> |
| 108 |
</ul> |
| 109 |
<h2><a name="query"></a>Writing a simple query</h2> |
| 110 |
<p>The following section will enable you to write simple TIGERSearch queries for the |
| 111 |
SRCMF corpus. It is not comprehensive, and must be read in conjunction with:</p> |
| 112 |
<ul> |
| 113 |
<li>chapter III of the <a target="_blank" |
| 114 |
href="http://www.ims.uni-stuttgart.de/projekte/TIGER/TIGERSearch/manual_html.html" |
| 115 |
>TIGERSearch user’s guide</a></li> |
| 116 |
</ul> |
| 117 |
<h3>Nodes in the TS graph</h3> |
| 118 |
<p>A TigerSearch graph is made up of two types of nodes: terminal and non-terminal |
| 119 |
nodes. In the graph viewer, terminal nodes appear at the bottom of the graph, while |
| 120 |
non-terminal nodes are represented by labelled white ovals, as shown in the example |
| 121 |
<i>je puis dire</i>.</p> |
| 122 |
<img src="images/jepuisdire.png" alt="Example TIGERSearch tree" /> |
| 123 |
<p>Each node has a number of features (see section “<a href="#tags">Tagset used</a>”</p> |
| 124 |
<h4>SRCMF: ‘split’ nodes</h4> |
| 125 |
<p>In a true dependency graph, words form the only nodes.</p> |
| 126 |
<p>In the TigerXML SRCMF corpus, each ‘word’ in the dependency structure is in fact |
| 127 |
split between a terminal node (which contains the lexical form and the PoS tag of |
| 128 |
the word itself) and a non-terminal node (which contains the syntactic features of |
| 129 |
the structure headed by the word). The non-terminal node and the terminal node are |
| 130 |
linked by an edge labelled ‘L’ (for lexical realization).</p> |
| 131 |
<p>In the example tree, an ‘L’ edge links:</p> |
| 132 |
<ul> |
| 133 |
<li>the terminal node <i>puis</i> to the non-terminal node ‘Snt’: these nodes |
| 134 |
represent the finite verb which heads the sentence;</li> |
| 135 |
<li>the terminal node <i>je</i> to the non-terminal node ‘SjPer’: these nodes |
| 136 |
represent the subject of the sentence <i>je</i>;</li> |
| 137 |
<li>the terminal node <i>dire</i> to the non-terminal node ‘AuxA’: these nodes |
| 138 |
represent the infinitive verb <i>dire</i>.</li> |
| 139 |
</ul> |
| 140 |
<p>A ‘D’ edge links the ‘Snt’ node to the non-terminal nodes ‘SjPer’ and ‘AuxA’: this |
| 141 |
indicates that the subject <i>je</i> and the ‘auxiliated’ infinitive <i>dire</i> |
| 142 |
depend on the main verb <i>puis</i>.</p> |
| 143 |
<h4>SRCMF corpus node features</h4> |
| 144 |
<p>The SRCMF corpus has the following node features:</p> |
| 145 |
<p><i>Terminal nodes:</i></p> |
| 146 |
<ul> |
| 147 |
<li><tt>word</tt>: the word form</li> |
| 148 |
<li><tt>pos</tt>: part-of-speech tag (Cattex)</li> |
| 149 |
<li><tt>form</tt>: whether the text is verse or prose, and position of the word in |
| 150 |
the line of verse.</li> |
| 151 |
</ul> |
| 152 |
<p><i>Non-terminal nodes:</i></p> |
| 153 |
<ul> |
| 154 |
<li><tt>cat</tt>: function of the structure headed by the node</li> |
| 155 |
<li><tt>type</tt>: morpho-syntactic category of the node (VFin, VPar, VInf, NV)</li> |
| 156 |
<li><tt>headpos</tt>: part-of-speech tag of the head word</li> |
| 157 |
<li><tt>coord</tt>: set to ‘y’ if the structure forms part of a coordination</li> |
| 158 |
<li><tt>dom</tt>: underscore-separated list of all functions dominated by the node |
| 159 |
(e.g. for the ‘Snt’ node above ‘AuxA_SjPer’)</li> |
| 160 |
</ul> |
| 161 |
<p>For simple queries, we will focus mainly on the <tt>word</tt>, <tt>pos</tt> and |
| 162 |
<tt>cat</tt> features.</p> |
| 163 |
<h4>Defining the feature specifications of a node</h4> |
| 164 |
<p>Node feature specifications are written between [square brackets] and take the |
| 165 |
following form:</p> |
| 166 |
<ul> |
| 167 |
<li><tt>[feature operator "value"]</tt></li> |
| 168 |
</ul> |
| 169 |
<p>where <i>value</i> is a string or</p> |
| 170 |
<ul> |
| 171 |
<li><tt>[feature operator /value/]</tt></li> |
| 172 |
</ul> |
| 173 |
<p>where <i>value</i> is a regular expression. Permitted <i>operators</i> are ‘=’ |
| 174 |
(equals) and ‘!=’ (does not equal). For example, the following expression identifies |
| 175 |
all nodes where <tt>cat</tt> is "SjPer" (personal subject):</p> |
| 176 |
<ul> |
| 177 |
<li><tt>[cat = "SjPer"]</tt></li> |
| 178 |
</ul> |
| 179 |
<p>If we wish to include impersonal subjects (i.e. "SjPer" and "SjImp") we can use a |
| 180 |
regular expression:</p> |
| 181 |
<ul> |
| 182 |
<li><tt>[cat = /Sj.*/]</tt></li> |
| 183 |
</ul> |
| 184 |
<p>We can identify all nodes which are <i>not</i> subjects:</p> |
| 185 |
<ul> |
| 186 |
<li><tt>[cat != /Sj.*/]</tt></li> |
| 187 |
</ul> |
| 188 |
<p>We may also the conjunction (&) operator within the square brackets to specify |
| 189 |
several properties. For example, we can search for subordinate clause subjects by |
| 190 |
requiring the subject to be headed by a finite verb (<tt>type</tt> is "VFin"):</p> |
| 191 |
<ul> |
| 192 |
<li><tt>[cat = /Sj.*/ & type = "VFin"]</tt></li> |
| 193 |
</ul> |
| 194 |
<h4>Assigning a variable name to a node</h4> |
| 195 |
<p>A variable name may be assigned to the node definition. These are useful to refer to |
| 196 |
the same node several times in a complex query and are also used to indicate the |
| 197 |
pivot node to concordance scripts.</p> |
| 198 |
<p>Variable definitions adopt the following syntax:</p> |
| 199 |
<ul> |
| 200 |
<li><tt>#name:[<definition>]</tt></li> |
| 201 |
</ul> |
| 202 |
<p>where <i>definition</i> is a feature specification as described above. Note that |
| 203 |
variable names must begin with hash (#) and are separated from their definition by a |
| 204 |
colon (:).</p> |
| 205 |
<p>For example, we may to construct a concordance in which the subject forms the pivot. |
| 206 |
We define the #pivot variable as follows:</p> |
| 207 |
<ul> |
| 208 |
<li><tt>#pivot:[cat = /Sj.*/]</tt></li> |
| 209 |
</ul> |
| 210 |
<h3>Node relations</h3> |
| 211 |
<p>All but the most simple queries will require more than one node to be defined, and |
| 212 |
will usually require the relationship between the nodes to be specified. </p> |
| 213 |
<p>For example, suppose we wish to identify all subjects headed by the word |
| 214 |
<i>Tristran</i>. First, we define the subject:</p> |
| 215 |
<ul> |
| 216 |
<li><tt>#subject:[cat = /Sj.*/]</tt></li> |
| 217 |
</ul> |
| 218 |
<p>Second, we define the word Tristran as a terminal node:</p> |
| 219 |
<ul> |
| 220 |
<li><tt>#tristran:[word = "Tristran"]</tt></li> |
| 221 |
</ul> |
| 222 |
<p>Finally, we must indicate the relationship between the nodes. The relationship |
| 223 |
between a non-terminal node and the terminal node representing its lexical content |
| 224 |
in the TigerSearch graph is one of direct dominance, labelled ‘L’ (lexical).</p> |
| 225 |
<h4>Direct dominance</h4> |
| 226 |
<p>In TigerSearch, direct dominance is expressed by using the operator ‘>’ with the |
| 227 |
following syntax:</p> |
| 228 |
<ul> |
| 229 |
<li><tt>node >[label] node2</tt></li> |
| 230 |
</ul> |
| 231 |
<p>where <i>node</i> and <i>node2</i> are feature specifications or node variables, and |
| 232 |
label (optional) is a string.</p> |
| 233 |
<p>To identify subjects headed by the word <i>Tristran</i>, the relationship between |
| 234 |
nodes #subject and #tristran is expressed as follows:</p> |
| 235 |
<ul> |
| 236 |
<li><tt>#subject >L #tristran</tt></li> |
| 237 |
</ul> |
| 238 |
<h4>Left corner dominance</h4> |
| 239 |
<p>The ‘>@l’ operator specifies the leftmost terminal node dominated at any depth by a |
| 240 |
non-terminal node. It has the following syntax:</p> |
| 241 |
<ul> |
| 242 |
<li><tt>node >@l tnode</tt></li> |
| 243 |
</ul> |
| 244 |
<p>where <i>node</i> and <i>tnode</i> are feature specifications or node variables, and |
| 245 |
<i>tnode</i> is a terminal node.</p> |
| 246 |
<p>For example, instead of searching for all subjects which are headed by the word |
| 247 |
<i>Tristran</i>, we may wish to identify all subjects <strong>beginning</strong> |
| 248 |
with the word <i>Tristran</i>. This relation would be written as follows:</p> |
| 249 |
<ul> |
| 250 |
<li><tt>#subject >@l #tristran</tt></li> |
| 251 |
</ul> |
| 252 |
<p>Note that there is also a right corner dominance operator ‘>@r’.</p> |
| 253 |
<h4>Precedence</h4> |
| 254 |
<p>The precedence operator ‘.*’ permits the user to specify the word order of two |
| 255 |
terminal nodes with the following syntax:</p> |
| 256 |
<ul> |
| 257 |
<li><tt>tnode .* tnode2</tt></li> |
| 258 |
</ul> |
| 259 |
<p>where <i>tnode</i> and <i>tnode2</i> are feature specifications or node variables |
| 260 |
representing terminal nodes.</p> |
| 261 |
<p> For example, suppose we wish to identify all sentences in which the word Tristran |
| 262 |
heads the subject and precedes the main clause verb.</p> |
| 263 |
<p>We need to add two additional conditions to the query in the previous section. First, |
| 264 |
we need to identify the terminal node containing the main verb of the sentence: i.e. |
| 265 |
the lexical realization of the non-terminal node ‘Snt’:</p> |
| 266 |
<ul> |
| 267 |
<li><tt>#snt:[cat = "Snt"] >L #verb</tt></li> |
| 268 |
</ul> |
| 269 |
<p>You may have noticed that #verb has no feature specification. This is perfectly valid |
| 270 |
in TigerSearch query syntax. In practice, we know that only one node can be linked |
| 271 |
to #snt by an ‘L’ relation in the corpus. #Verb is thus defined by its relation to |
| 272 |
#snt rather than by its features.</p> |
| 273 |
<p>We then need to specify that the word Tristran precedes the verb:</p> |
| 274 |
<ul> |
| 275 |
<li><tt>#tristran .* #verb</tt></li> |
| 276 |
</ul> |
| 277 |
<p>Finally, we need to clarify that #subject is the the subject of #snt. Otherwise, we |
| 278 |
risk finding subjects of a subordinate clause which happen to precede the main |
| 279 |
clause verb:</p> |
| 280 |
<ul> |
| 281 |
<li><tt>#snt >D #subject</tt></li> |
| 282 |
</ul> |
| 283 |
<p>Putting it all together, the query is as follows:</p> |
| 284 |
<ul> |
| 285 |
<li><tt>#subject:[cat = /Sj.*/] >L #tristran:[word = "Tristran"] <br /> & |
| 286 |
#snt:[cat = "Snt"] >L #verb <br /> & #tristran .* #verb <br /> & |
| 287 |
#snt >D #subject</tt></li> |
| 288 |
</ul> |
| 289 |
<p>There is also a direct precedence operator, ‘.’, which specifies that the two |
| 290 |
terminal nodes must be directly adjacent.</p> |
| 291 |
<h4>Negation</h4> |
| 292 |
<p>It is important to learn one (extremely frustrating) golden rule of Tiger query |
| 293 |
syntax:</p> |
| 294 |
<ul> |
| 295 |
<li>you can negate a feature specification (e.g. <tt>[cat != "SjPer"]</tt>);</li> |
| 296 |
<li>you can negate a relation between nodes (e.g. <tt>#subject !>L |
| 297 |
#tristran</tt>)</li> |
| 298 |
<li><strong>but you can’t negate the existence of a node!</strong></li> |
| 299 |
</ul> |
| 300 |
<p>In practice, this means that when we write:</p> |
| 301 |
<ul> |
| 302 |
<li><tt>#snt:[cat = "Snt"] !>D #subject:[cat = /Sj.*/]</tt></li> |
| 303 |
</ul> |
| 304 |
<p>we have <strong>not</strong> found all null subject main clauses. Instead, we have |
| 305 |
asked for sentences (#snt) which contain a subject node (#subject) which is |
| 306 |
<strong>not</strong> the subject of a sentence. TigerSearch will return all |
| 307 |
sentences with subjects in a subordinate clause.</p> |
| 308 |
<p>The SRCMF corpus provides a partial work-around for this problem by using the |
| 309 |
<i>dom</i> feature. The <i>dom</i> feature of a non-terminal node lists the cat |
| 310 |
features of all nodes linked to it by a ‘D’ edge in alphabetical order separated by |
| 311 |
an underscore. For example, the ‘Snt’ node in the example tree has two dependants: |
| 312 |
SjPer and AuxA. It therefore has a <i>dom</i> property ‘AuxA_SjPer’.</p> |
| 313 |
<p>As a result, we can identify all main clauses without subjects by negating the |
| 314 |
<i>dom</i> feature:</p> |
| 315 |
<ul> |
| 316 |
<li><tt>#snt:[cat = "Snt" & dom != /.*Sj.*/]</tt></li> |
| 317 |
</ul> |
| 318 |
<p>This will return all ‘Snt’ nodes whose <i>dom</i> property does not contain the |
| 319 |
characters ‘Sj’: in other words, a main clause without an expressed subject.</p> |
| 320 |
<h4>Syntactic variation</h4> |
| 321 |
<p>TigerSearch syntax is quite flexible, and we may express queries in a number of ways. |
| 322 |
For example, the query identifying all subjects headed by the word <i>Tristran</i> |
| 323 |
may be expressed using three statements...</p> |
| 324 |
<ul> |
| 325 |
<li><tt>#subject:[cat = /Sj.*/] <br /> & #tristran:[word = "Tristran"] <br /> |
| 326 |
& #subject >L #tristran</tt></li> |
| 327 |
</ul> |
| 328 |
<p>... or two statements, e.g.:</p> |
| 329 |
<ul> |
| 330 |
<li><tt>#subject:[cat = /Sj.*/] <br /> & #subject >L #tristran:[word = |
| 331 |
"Tristran"]</tt></li> |
| 332 |
</ul> |
| 333 |
<p>... or one statement:</p> |
| 334 |
<ul> |
| 335 |
<li><tt>#subject:[cat = /Sj.*/] >L #tristran:[word = "Tristran"]</tt></li> |
| 336 |
</ul> |
| 337 |
<p>... or without variable names:</p> |
| 338 |
<ul> |
| 339 |
<li><tt>[cat = /Sj.*/] >L [word = "Tristran"]</tt></li> |
| 340 |
</ul> |
| 341 |
<p>Where multiple statements are used, the order of statements is irrelevant. |
| 342 |
Confusingly for programmers, you may reference variables before assigning a value, |
| 343 |
e.g.:</p> |
| 344 |
<ul> |
| 345 |
<li><tt>#subject >L #tristran & #tristran:[word = "Tristran"] & |
| 346 |
#subject:[cat = /Sj.*/]</tt></li> |
| 347 |
</ul> |
| 348 |
<h2><a name="concordances"></a>Using concordances</h2> |
| 349 |
<p>The SRCMF project has developed a number of concordances to present the results of |
| 350 |
TigerSearch queries in tabular format. Three concordances are currently |
| 351 |
implemented:</p> |
| 352 |
<ul> |
| 353 |
<li>basic concordance</li> |
| 354 |
<li>single word pivot concordance</li> |
| 355 |
<li>pivot and block concordance</li> |
| 356 |
</ul> |
| 357 |
<p>These concordances produce a text CSV file.</p> |
| 358 |
<h3>Principles</h3> |
| 359 |
<p>The concordances use the names of variables from the TigerSearch query to identify |
| 360 |
the syntactic constituents which should form the focus of the table. All |
| 361 |
concordances require a #pivot variable to be present in the query.</p> |
| 362 |
<p>For example, the following query is correct in TigerSearch, but <strong>will |
| 363 |
not</strong> produce a concordance:</p> |
| 364 |
<ul> |
| 365 |
<li><tt>[word = /Tristr?a[nm][sz]?/]</tt></li> |
| 366 |
</ul> |
| 367 |
<p>To produce a concordance, the query must identify a node as the #pivot, for |
| 368 |
example:</p> |
| 369 |
<ul> |
| 370 |
<li><tt><strong>#pivot:</strong>[word = /Tristr?a[nm][sz]?/]</tt></li> |
| 371 |
</ul> |
| 372 |
<h3>Basic concordance</h3> |
| 373 |
<p>The basic concordance has four columns:</p> |
| 374 |
<ul> |
| 375 |
<li>sentence ID</li> |
| 376 |
<li>left context</li> |
| 377 |
<li>pivot</li> |
| 378 |
<li>right context</li> |
| 379 |
</ul> |
| 380 |
<p>The #pivot can be any node in the syntactic tree, either a single word or a larger |
| 381 |
structure. Currently, only lexical information (not annotation) can be shown in the |
| 382 |
basic concordance.</p> |
| 383 |
<p>For example, we may wish to create a concordance of all the main clause subjects |
| 384 |
containing the word ‘Tristran’:</p> |
| 385 |
<ul> |
| 386 |
<li><tt>#snt:[cat = "Snt"] >D #pivot:[cat = "SjPer"] & #pivot >* [word = |
| 387 |
/Tristr?a[nm][sz]?/]</tt></li> |
| 388 |
</ul> |
| 389 |
<p>Note that the #pivot variable is attached to the subject node (cat = "SjPer").</p> |
| 390 |
<p>Below is a selection of the results from the concordance:</p> |
| 391 |
<table border="1"> |
| 392 |
<tr> |
| 393 |
<th>ID</th> |
| 394 |
<th>contexte gauche</th> |
| 395 |
<th>pivot</th> |
| 396 |
<th>contexte droite</th> |
| 397 |
</tr> |
| 398 |
<tr> |
| 399 |
<td>beroul_pb:8_lb:234_1263227636.06</td> |
| 400 |
<td>di por averté Ce saciés vos de verité Atant s' en est Iseut tornee</td> |
| 401 |
<td>Tristran</td> |
| 402 |
<td>l' a plorant salüee Sor le perron de marbre bis Tristran s' apuie ce</td> |
| 403 |
</tr> |
| 404 |
<tr> |
| 405 |
<td>beroul_pb:13_lb:415_1264876249.02</td> |
| 406 |
<td># croiz Einz croiz parole fole et vaine Ma bone foi me fera saine Tristran |
| 407 |
[remest] a qui * mot poise </td> |
| 408 |
<td>Tristran tes niés </td> |
| 409 |
<td>vint soz cel pin Qui * est laienz en cel jardin Si me manda</td> |
| 410 |
</tr> |
| 411 |
<tr> |
| 412 |
<td>beroul_pb:134_lb:4365_1268928771.68</td> |
| 413 |
<td>moi le reçoive En sus l' atent s' espee tient Goudoïne autre voie tient</td> |
| 414 |
<td>Tristran [remest] a qui * mot poise</td> |
| 415 |
<td>Ist du * buison cela part toise Mais por noient quar cil s' esloigne</td> |
| 416 |
</tr> |
| 417 |
</table> |
| 418 |
<p>Note that the pivot may be one or more words.</p> |
| 419 |
<h3>What do the square brackets ([]), slashes (/), asterisks (*) and hashes (#) |
| 420 |
mean?</h3> |
| 421 |
<p>The third example in the above table contains [square brackets] in the pivot. These |
| 422 |
are used in all concordances to indicate <strong>words which occur between parts of |
| 423 |
a discontinuous syntactic constituent</strong>.</p> |
| 424 |
<p>The annotated subject in this sentence is <i>Tristran ... a qui mot poise</i>. The |
| 425 |
main verb of the sentence, <i>remest</i>, is not part of the subject, but occurs |
| 426 |
between its two parts. The verb <i>remest</i> is included in the pivot column, but |
| 427 |
surrounded by square brackets.</p> |
| 428 |
<p>This means that:</p> |
| 429 |
<ul> |
| 430 |
<li>the pivot column contains <strong>all parts</strong> of discontinuous |
| 431 |
pivots;</li> |
| 432 |
<li>reading the concordance from left to right will always give the original |
| 433 |
sentence.</li> |
| 434 |
</ul> |
| 435 |
<p>Slashes (/) indicate division between sentences in the syntactic annotation. These |
| 436 |
will not correspond to the editor’s division into sentences as shown in the |
| 437 |
punctuation.</p> |
| 438 |
<p>Asterisks (*) indicate that the preceding word has two syntactic functions (e.g. |
| 439 |
<i>qui</i> in <i>a qui mot poise</i> is both a relator and a subject). They may |
| 440 |
usually be ignored.</p> |
| 441 |
<p>Hashes (#) are related to the representation of coordination, and may always be |
| 442 |
ignored.</p> |
| 443 |
<h3>Single word pivot concordance</h3> |
| 444 |
<p>The single word pivot concordance has a variable number of columns, based on the |
| 445 |
following structure:</p> |
| 446 |
<ul> |
| 447 |
<li>ID</li> |
| 448 |
<li>Left context outside the SRCMF sentence containing the pivot</li> |
| 449 |
<li>Left context within the SRCMF sentence containing the pivot</li> |
| 450 |
<li>Pivot</li> |
| 451 |
<li>Structure headed by the pivot</li> |
| 452 |
<li>Function of the structure headed by the pivot</li> |
| 453 |
<li>Right context within the SRCMF sentence containing the pivot</li> |
| 454 |
<li>Right context outside the SRCMF sentence containing the pivot</li> |
| 455 |
</ul> |
| 456 |
<p>The single word pivot concordance is designed to give as much information as possible |
| 457 |
about a single word. For example, a concordance could be created around the word |
| 458 |
"Tristran":</p> |
| 459 |
<ul> |
| 460 |
<li><tt>#pivot:[word = /Tristr?a[nm][sz]?/]</tt></li> |
| 461 |
</ul> |
| 462 |
<p>Below is a selection of the results from the concordance (some columns are |
| 463 |
omitted):</p> |
| 464 |
<table border="1"> |
| 465 |
<tr> |
| 466 |
<th>Left context in sentence</th> |
| 467 |
<th>Pivot</th> |
| 468 |
<th>Pivot-headed structure</th> |
| 469 |
<th>Right context in sentence</th> |
| 470 |
</tr> |
| 471 |
<tr> |
| 472 |
<td>Sire</td> |
| 473 |
<td>Tristran</td> |
| 474 |
<td>Tristran</td> |
| 475 |
<td>por Deu le roi Si grant pechié avez de moi Qui * me mandez a itel ore</td> |
| 476 |
</tr> |
| 477 |
<tr> |
| 478 |
<td></td> |
| 479 |
<td>Tristran</td> |
| 480 |
<td>Tristran tes niés</td> |
| 481 |
<td>tes niés vint soz cel pin Qui * est laienz en cel jardin</td> |
| 482 |
</tr> |
| 483 |
<tr> |
| 484 |
<td># Que por Yseut que por</td> |
| 485 |
<td>Tristranz</td> |
| 486 |
<td>que por Tristranz</td> |
| 487 |
<td>Mervellose joie menoient</td> |
| 488 |
</tr> |
| 489 |
</table> |
| 490 |
<p>The ‘pivot-headed structure’ gives the noun phrase of which the word <i>Tristan</i> |
| 491 |
is head. In the second example, for instance, the word <i>Tristran</i> heads the |
| 492 |
structure <i>Tristan tes niés</i>.</p> |
| 493 |
<p>Note that words appearing in the ‘pivot-headed structure’ column are also found in |
| 494 |
the two context columns. The original sentence may be read across the columns left |
| 495 |
context — pivot — right context.</p> |
| 496 |
<h3>Pivot and block concordance</h3> |
| 497 |
<h4>Introduction</h4> |
| 498 |
<p>The pivot and block concordance is designed to highlight the position of certain |
| 499 |
constituents, called ‘blocks’ (e.g. the subject) with respect to a pivot (e.g. the |
| 500 |
verb). The resulting CSV files are complex, with a large number of columns, and are |
| 501 |
intended as the basis for more detailed analysis in spreadsheet software.</p> |
| 502 |
<p>The pivot and block concordances has the following basic structure:</p> |
| 503 |
<ul> |
| 504 |
<li>ID</li> |
| 505 |
<li>Left context outside the SRCMF sentence containing the pivot</li> |
| 506 |
<li>Left context within the SRCMF sentence containing the pivot</li> |
| 507 |
<li>Pre-pivot blocks</li> |
| 508 |
<li>Pivot</li> |
| 509 |
<li>Post-pivot blocks</li> |
| 510 |
<li>Right context within the SRCMF sentence containing the pivot</li> |
| 511 |
<li>Right context outside the SRCMF sentence containing the pivot</li> |
| 512 |
</ul> |
| 513 |
<p>As with the other concordances, TigerSearch queries must define a #pivot variable. |
| 514 |
However, any number of variables whose name begins ‘#block’ may be defined. At least |
| 515 |
one ‘#blockXX’ variable is required.</p> |
| 516 |
<p>For example, the following query will generate a pivot and block concordance to show |
| 517 |
the position of the subject (#block1) with respect to the finite verb (#pivot):</p> |
| 518 |
<ul> |
| 519 |
<li><tt>#snt:[cat = "Snt"] >D #block1:[cat = "SjPer"] & #snt >L |
| 520 |
#pivot</tt></li> |
| 521 |
</ul> |
| 522 |
<p>In essence, the central section of the resulting concordance will take the following |
| 523 |
form:</p> |
| 524 |
<table border="1"> |
| 525 |
<tr> |
| 526 |
<th>Left context</th> |
| 527 |
<th>Block</th> |
| 528 |
<th>Pivot</th> |
| 529 |
<th>Block</th> |
| 530 |
<th>Right context</th> |
| 531 |
</tr> |
| 532 |
<tr> |
| 533 |
<td></td> |
| 534 |
<td>Li rois</td> |
| 535 |
<td>pense</td> |
| 536 |
<td></td> |
| 537 |
<td>que par folie Sire Tristran vos aie amé</td> |
| 538 |
</tr> |
| 539 |
<tr> |
| 540 |
<td>Si</td> |
| 541 |
<td></td> |
| 542 |
<td>voient</td> |
| 543 |
<td>il</td> |
| 544 |
<td># Deu et son reigne</td> |
| 545 |
</tr> |
| 546 |
</table> |
| 547 |
<p>Where the subject is pre-verbal, it appears in the block column to the left of the |
| 548 |
pivot. Where it is post-verbal, it appears in the block column to the right of the |
| 549 |
pivot.</p> |
| 550 |
<h4>Why are there square brackets ([]) and curly brackets ({}) in the concordance?</h4>
|
| 551 |
<p>As with other concordances, square brackets denote <strong>words occurring between |
| 552 |
two parts of a discontinuous unit</strong>. The difference in this concordance |
| 553 |
is that blocks may be discontinuous, as well as the pivot.</p> |
| 554 |
<p>Curly brackets denote <strong>words which occur between the block and the |
| 555 |
pivot</strong> (or, in more complex examples, between two blocks).</p> |
| 556 |
<table border="1"> |
| 557 |
<tr> |
| 558 |
<th>Left context</th> |
| 559 |
<th>Block</th> |
| 560 |
<th>Pivot</th> |
| 561 |
<th>Block</th> |
| 562 |
<th>Right context</th> |
| 563 |
</tr> |
| 564 |
<tr> |
| 565 |
<td></td> |
| 566 |
<td>Vos {n'}</td>
|
| 567 |
<td>entendez</td> |
| 568 |
<td></td> |
| 569 |
<td>pas la raison</td> |
| 570 |
</tr> |
| 571 |
<tr> |
| 572 |
<td>Dex qel pitié</td> |
| 573 |
<td></td> |
| 574 |
<td>Faisoit</td> |
| 575 |
<td>{a} {mainte} {gent} li chiens</td>
|
| 576 |
<td></td> |
| 577 |
</tr> |
| 578 |
<tr> |
| 579 |
<td></td> |
| 580 |
<td>Ta parole [est] [tost] [entendue] Que li rois la roïne prent</td> |
| 581 |
<td>est</td> |
| 582 |
<td></td> |
| 583 |
<td>tost entendue Que li rois la roïne prent</td> |
| 584 |
</tr> |
| 585 |
<tr> |
| 586 |
<td></td> |
| 587 |
<td>Tuit [s'] [escrïent] la gent du * reigne {s'}</td>
|
| 588 |
<td>escrïent</td> |
| 589 |
<td></td> |
| 590 |
<td>la gent du * reigne</td> |
| 591 |
</tr> |
| 592 |
</table> |
| 593 |
<p>In the table above, note the use of curly brackets in the first example to mark the |
| 594 |
negative adverb <i>n’</i>, which occurs between the subject-block <i>vos</i> and the |
| 595 |
verb-pivot <i>entendez</i>. In the second example, the prepositional phrase <i>a |
| 596 |
maintes gens</i> is marked with curly brackets, as it separates the verb-pivot |
| 597 |
<i>Faisoit</i> from the post-verbal subject-block <i>li chiens</i>.</p> |
| 598 |
<p>In the third example, a discontinuous subject <i>Ta parole ... que li rois la roïne |
| 599 |
prent</i> appears in a pre-verbal block. <strong>The pre- or post-verbal |
| 600 |
position of a block is determined by the position of its first word relative to |
| 601 |
the pivot</strong>. The words <i>est tost entendue</i>, which separate the two |
| 602 |
parts of the block, are marked with square brackets. </p> |
| 603 |
<p>In the fourth example, the word <i>s’</i> appears (i) in square brackets, between the |
| 604 |
two halves of a discontinuous subject-block and (ii) in curly brackets, between the |
| 605 |
first part of the discontinuous subject <i>tost</i> and the verb-pivot |
| 606 |
<i>escrïent</i>.</p> |
| 607 |
<h4>Why are there so many columns? I only asked for one block!</h4> |
| 608 |
<p>The pivot and block concordance shows <strong>only one result per pivot</strong>. |
| 609 |
Continuing to work with the same example, if a single verb-pivot has multiple |
| 610 |
subject-blocks (which is quite possible in cases of coordination), each subject |
| 611 |
occupies a separate column:</p> |
| 612 |
<table border="1"> |
| 613 |
<tr> |
| 614 |
<th>Block3</th> |
| 615 |
<th>Block2</th> |
| 616 |
<th>Block1</th> |
| 617 |
<th>Pivot</th> |
| 618 |
<th>Block</th> |
| 619 |
</tr> |
| 620 |
<tr> |
| 621 |
<td>Ne tor</td> |
| 622 |
<td>ne mur</td> |
| 623 |
<td>ne fort chastel {Ne} {me}</td>
|
| 624 |
<td>tendra</td> |
| 625 |
<td></td> |
| 626 |
</tr> |
| 627 |
</table> |
| 628 |
<p>However, due to the way the number of columns is calculated, it is possible that some |
| 629 |
will be empty. These may be deleted in the spreadsheet software, if you wish.</p> |
| 630 |
<p>Note that the concordance will <strong>never</strong> represent the two halves of a |
| 631 |
<strong>single discontinuous</strong> block in separate columns. The following |
| 632 |
representation therefore indicates a coordination:</p> |
| 633 |
<table border="1"> |
| 634 |
<tr> |
| 635 |
<th>Left context</th> |
| 636 |
<th>Block</th> |
| 637 |
<th>Pivot</th> |
| 638 |
<th>Block</th> |
| 639 |
<th>Right context</th> |
| 640 |
</tr> |
| 641 |
<tr> |
| 642 |
<td></td> |
| 643 |
<td>Tristran {en}</td>
|
| 644 |
<td>bese</td> |
| 645 |
<td>{la} {roïne} {Et} ele</td>
|
| 646 |
<td>lui par la saisine</td> |
| 647 |
</tr> |
| 648 |
</table> |
| 649 |
<p>The SRCMF of the sentence in this table identifies <strong>two coordinated |
| 650 |
subjects</strong> of the verb <i>bese</i>. One is pre-verbal (<i>Tristran</i>), |
| 651 |
one is post-verbal (<i>ele</i>); both occupy separate blocks.</p> |
| 652 |
<h3>Adding annotation information</h3> |
| 653 |
<p>When a concordance is launched from the TXM-web interface, you may specify which |
| 654 |
properties of terminal and non-terminal nodes you wish to see in the |
| 655 |
concordance.</p> |
| 656 |
<ul> |
| 657 |
<li>On the ‘Export Concordance’ form, use the drop-down lists of ‘Non-terminal |
| 658 |
features’ and ‘Terminal Features’.</li> |
| 659 |
<li>Select the features of terminal and non-terminal nodes that you wish to show in |
| 660 |
the concordance from the two drop-down lists.</li> |
| 661 |
<li>Click ‘OK’.</li> |
| 662 |
</ul> |
| 663 |
<p>Each added property will be placed in a separate column next to the block or pivot. |
| 664 |
For example, if the ‘cat’ property is selected for non-terminal nodes, and the ‘pos’ |
| 665 |
property is selected for terminal nodes, the query above will produce the following |
| 666 |
concordance:</p> |
| 667 |
<table border="1"> |
| 668 |
<tr> |
| 669 |
<th>Left context</th> |
| 670 |
<th>Block</th> |
| 671 |
<th>Block Cat</th> |
| 672 |
<th>Pivot</th> |
| 673 |
<th>Pivot Pos</th> |
| 674 |
<th>Block</th> |
| 675 |
<th>Block Cat</th> |
| 676 |
<th>Right context</th> |
| 677 |
</tr> |
| 678 |
<tr> |
| 679 |
<td></td> |
| 680 |
<td>Li rois</td> |
| 681 |
<td>SjPer</td> |
| 682 |
<td>pense</td> |
| 683 |
<td>VERcjg</td> |
| 684 |
<td></td> |
| 685 |
<td></td> |
| 686 |
<td>que par folie Sire Tristran vos aie amé</td> |
| 687 |
</tr> |
| 688 |
<tr> |
| 689 |
<td>Si</td> |
| 690 |
<td></td> |
| 691 |
<td></td> |
| 692 |
<td>voient</td> |
| 693 |
<td>VERcjg</td> |
| 694 |
<td>il</td> |
| 695 |
<td>SjPer</td> |
| 696 |
<td># Deu et son reigne</td> |
| 697 |
</tr> |
| 698 |
</table> |
| 699 |
<h2><a name="tags"></a>Tagset</h2> |
| 700 |
<h3>Non-terminal nodes</h3> |
| 701 |
<p>Non-terminal nodes have the following properties and values:</p> |
| 702 |
<h4>cat</h4> |
| 703 |
<p>Gives the syntactic function of the element. For more details, please refer to the <a |
| 704 |
target="_blank" href="http://srcmf.org">SRCMF |
| 705 |
website</a>.</p> |
| 706 |
<ul> |
| 707 |
<li><a name="Apst"></a><strong>Apst</strong>: Vocative (fr. apostrophe)</li> |
| 708 |
<li><a name="AtObj"></a><strong>AtObj</strong>: Object attribute</li> |
| 709 |
<li><a name="AtRfc"></a><strong>AtRfc</strong>: Attribute of reflexive pronoun</li> |
| 710 |
<li><a name="AtSj"></a><strong>AtSj</strong>: Subject attribute</li> |
| 711 |
<li><a name="Aux"></a><strong>Aux</strong>: Auxiliated non-finite verb (neither |
| 712 |
passive nor active)</li> |
| 713 |
<li><a name="AuxA"></a><strong>AuxA</strong>: Auxiliated non-finite verb |
| 714 |
(active)</li> |
| 715 |
<li><a name="AuxP"></a><strong>AuxA</strong>: Auxiliated non-finite verb |
| 716 |
(passive)</li> |
| 717 |
<li><a name="Circ"></a><strong>Circ</strong>: Adjunct (fr. circonstant)</li> |
| 718 |
<li><a name="Cmpl"></a><strong>Cmpl</strong>: Complement</li> |
| 719 |
<li><a name="Coo"></a><strong>Coo</strong>: Coordination</li> |
| 720 |
<li><a name="GpCoo"></a><strong>GpCoo</strong>: Coordinated group (conjunct)</li> |
| 721 |
<li><a name="Insrt"></a><strong>Insrt</strong>: Inserted clause</li> |
| 722 |
<li><a name="Intj"></a><strong>Intj</strong>: Interjection</li> |
| 723 |
<li><a name="ModA"></a><strong>ModA</strong>: Modifier (attached)</li> |
| 724 |
<li><a name="ModD"></a><strong>ModD</strong>: Dislocated (detached) modifier</li> |
| 725 |
<li><a name="Ng"></a><strong>Ng</strong>: Negation</li> |
| 726 |
<li><a name="NgPrt"></a><strong>NgPrt</strong>: Negative particle (e.g. <i>pas</i>, |
| 727 |
<i>mie</i></li> |
| 728 |
<li><a name="nSnt"></a><strong>nSnt</strong>: Non-sentence</li> |
| 729 |
<li><a name="Obj"></a><strong>Obj</strong>: Object</li> |
| 730 |
<li><a name="RelC"></a><strong>RelC</strong>: Coordinated relator</li> |
| 731 |
<li><a name="RelNC"></a><strong>RelNC</strong>: Non-coordinating relator</li> |
| 732 |
<li><a name="Regim"></a><strong>Regim</strong>: Regime</li> |
| 733 |
<li><a name="Rfc"></a><strong>Rfc</strong>: Reflexive pronoun</li> |
| 734 |
<li><a name="Rfx"></a><strong>Rfx</strong>: Doubled reflexive pronoun (e.g. <i>nous |
| 735 |
... <strong>nous-mêmes</strong></i>)</li> |
| 736 |
<li><a name="SjImp"></a><strong>SjImp</strong>: Impersonal subject</li> |
| 737 |
<li><a name="SjPer"></a><strong>SjPer</strong>: Personal subject</li> |
| 738 |
<li><a name="Snt"></a><strong>Snt</strong>: Sentence</li> |
| 739 |
</ul> |
| 740 |
<h4>type</h4> |
| 741 |
<p>Gives the syntactic category of the head of the structure.</p> |
| 742 |
<ul> |
| 743 |
<li><a name="VFin"></a><strong>VFin</strong>: Finite verb form</li> |
| 744 |
<li><a name="VInf"></a><strong>VInf</strong>: Infinitive</li> |
| 745 |
<li><a name="VPar"></a><strong>VPar</strong>: Participle</li> |
| 746 |
<li><a name="nV"></a><strong>nV</strong>: Non-verbal</li> |
| 747 |
</ul> |
| 748 |
<h4>dom</h4> |
| 749 |
<p>A ‘dom’ property is added to each non-terminal node in the tree listing the functions |
| 750 |
of all its dependants and relators in alphabetical order, separated by underscores. |
| 751 |
For example, if a finite verb has a subject, object and two adjuncts, the property |
| 752 |
[dom = "Circ_Circ_Obj_SjPer"] will be added.</p> |
| 753 |
<p>This resolves to an extent the problem of ‘negative’ queries. Recall that it is |
| 754 |
impossible to query the non-existence of a node:</p> |
| 755 |
<ul> |
| 756 |
<li><tt>#clause:[type = "VFin"] !>D #suj:[cat = "SjPer"]</tt></li> |
| 757 |
</ul> |
| 758 |
<p>Contrary to appearances, this query DOES NOT mean ‘node #suj does not exist’: it |
| 759 |
means that the node #suj exists, but is not dependant on #clause.</p> |
| 760 |
<p>However, it is possible to find all finite verbs without a subject by using the dom |
| 761 |
property of the finite verb:</p> |
| 762 |
<ul> |
| 763 |
<li><tt>#clause:[type = "VFin" & dom != /.*SjPer.*/]</tt></li> |
| 764 |
</ul> |
| 765 |
<p>The query specifies that we wish to find a node #clause which is a finite verb and |
| 766 |
does not have the string ‘SjPer’ in the list of dependant nodes given by the dom |
| 767 |
property.</p> |
| 768 |
<h4>coord</h4> |
| 769 |
<p>A ‘coord’ property is added to each non-terminal node in the tree. If the node |
| 770 |
represents a coordinated structure, [coord = "y"].</p> |
| 771 |
<p>For example, in the sentence <i>Sade et douz est quanqu’est de li</i> (gcoin1: p. 3, |
| 772 |
l. 31), <i>sade</i> and <i>douz</i> are coordinated AtSj. The non-terminal nodes |
| 773 |
dominating the words <i>sade</i> and <i>douz</i> have the properties [cat = "AtSj" |
| 774 |
& coord="y"].</p> |
| 775 |
<p>The ‘coord’ property exists primarily to allow non-coordinated structures to be |
| 776 |
identified. In the original format, this is not possible, as it would require a |
| 777 |
query specifying the non-existence of a node [cat = "Coo"]. However, with the coord |
| 778 |
property, it is possible to restrict a query to non-coordinated structures only:</p> |
| 779 |
<ul> |
| 780 |
<li><tt>#suj:[cat = "SjPer" & coord != "y"]</tt></li> |
| 781 |
</ul> |
| 782 |
<h4>headpos</h4> |
| 783 |
<p>A ‘headpos’ property is added to each non-terminal node in the tree. If the text is |
| 784 |
correctly annotated at the deep level, each non-terminal node representing a |
| 785 |
structure should directly dominate at most one terminal node in the tree, the word |
| 786 |
representing the lexical content of the head of the structure. If this is the case, |
| 787 |
the ‘headpos’ property is equal to the ‘pos’ property of the dominated terminal |
| 788 |
node. Thus:</p> |
| 789 |
<ul> |
| 790 |
<li><tt>#node:[headpos = "NOMcom"]</tt></li> |
| 791 |
</ul> |
| 792 |
<p>is equivalent to:</p> |
| 793 |
<ul> |
| 794 |
<li><tt>#node >L #lexnode:[pos = "NOMcom"]</tt></li> |
| 795 |
</ul> |
| 796 |
<p>The headpos property does not improve the usability of the corpus in TigerSearch, but |
| 797 |
is useful in producing concordances, providing a more detailed morpho-syntactic tag |
| 798 |
for the head of a structure than the SRCMF ‘NV’ (non-verbal) type tag.</p> |
| 799 |
<p>If the non-terminal node directly dominates more than one terminal node, the |
| 800 |
algorithm generating the headpos property makes an calculated guess as to which word |
| 801 |
is the head, and inserts the tag of this word as the ‘headpos’. For example, if a |
| 802 |
non-terminal node dominates a word with pos ‘NOMcom’ and a word with pos ‘DETdef’, |
| 803 |
the algorithm will guess that the noun is the head, and insert the headpos |
| 804 |
‘NOMcom?’.</p> |
| 805 |
<p>Note that headpos values which have been ‘guessed’ are always suffixed by a question |
| 806 |
mark (e.g. NOMcom?). There will be no guessed headpos values in texts with full NP |
| 807 |
annotation.</p> |
| 808 |
<h3>Terminal nodes</h3> |
| 809 |
<p>Terminal nodes have the following properties:</p> |
| 810 |
<h4>pos</h4> |
| 811 |
<p>Part-of-speech tag (Cattex). For more information, please refer to the <a |
| 812 |
target="_blank" href="http://bfm.ens-lyon.fr/article.php3?id_article=323">Cattex |
| 813 |
documentation</a> on the <a target="_blank" href="http://bfm.ens-lyon.fr/">BFM website</a>.</p> |
| 814 |
<h4>form</h4> |
| 815 |
<p>Each word has a property “form”. For texts in prose, the value of the “form” tags is |
| 816 |
always “prose”. For texts in verse, the form tag is:</p> |
| 817 |
<ul> |
| 818 |
<li>“vers_first” for the first word in a line;</li> |
| 819 |
<li>“vers_end” for the last word in a line;</li> |
| 820 |
<li>“vers” for other words.</li> |
| 821 |
</ul> |
| 822 |
<p>It is thus possible to formulate a TS query focusing on words at the beginning or end |
| 823 |
of a line of verse:</p> |
| 824 |
<ul> |
| 825 |
<li><tt>[word = "Tristran" & form = "vers_end"]</tt></li> |
| 826 |
</ul> |
| 827 |
<p>In <i>Aucassin and Nicolete</i>, the form tag correctly distinguishes the verse and |
| 828 |
prose sections of the text.</p> |
| 829 |
<h4>q</h4> |
| 830 |
<p>Each word has a property “q”. This is equal to ‘y’ when the word occurs as part of |
| 831 |
direct discourse, and ‘n’ when it does not. This annotation is automatically |
| 832 |
generated by the BFM team from the position of quote marks in the text.</p> |
| 833 |
<h2><a name="sample"></a>Sample queries</h2> |
| 834 |
<p> The following sample queries may be tested by copying and pasting into the query |
| 835 |
panel. </p> |
| 836 |
<p>Find all main clause verbs:<br /> |
| 837 |
<tt>[cat = "Snt"]</tt></p> |
| 838 |
<p>Find all structures introduced by a preposition:<br /> |
| 839 |
<tt>#n >R #relnc:[cat = "RelNC"]<br /> & #relnc >L [pos = /PRE.*/]</tt><br /> |
| 840 |
</p> |
| 841 |
<p>Find all post-verbal NP subjects:<br /> |
| 842 |
<tt>#verb:[type = "VFin"] >D #suj:[cat = "SjPer" & type="nV"]<br /> & #suj |
| 843 |
>L [pos = /NOM.*/] <br /> & #suj >@l #sword<br /> & #verb >L |
| 844 |
#vword<br /> & #vword .* #sword</tt></p> |
| 845 |
<p>Find indefinite subjects introduced by <q>qui</q>:<br /> |
| 846 |
<tt>[type = "VFin"] >D #suj:[cat = "SjPer"]<br /> & #suj >R #relnc:[cat = |
| 847 |
"RelNC"]<br /> & ( #relnc >L [word = /[QqKk]u?i/]<br /> | #relnc >~dupl |
| 848 |
[word = /[QqKk]u?i/] )</tt><br /></p> |
| 849 |
<p>Find sentences with coordinated subjects:<br /> |
| 850 |
<tt>#coo:[cat = "Coo"] >~coord #sj1:[cat = "SjPer"]<br /> & #coo >~coord |
| 851 |
#sj2:[cat = "SjPer"]<br /> & #sj1 $ #sj2</tt></p> |
| 852 |
<p>Find sentences with possible <q>gapping</q> of the finite verb (i.e. coordination of |
| 853 |
subject–predicate pairs):<br /> |
| 854 |
<tt>#gpcoo1:[cat = "GpCoo"] >~ #suj1:[cat = "SjPer"]<br /> & #gpcoo1 $.* |
| 855 |
#gpcoo2:[cat = "GpCoo"]<br /> & #gpcoo2 >~ #suj2:[cat = "SjPer"]<br /> & |
| 856 |
#gpcoo1 >~ #pred1:[cat = /Cmpl|Obj|AtSj/]<br /> & #gpcoo2 >~ #pred2:[cat = |
| 857 |
/Cmpl|Obj|AtSj/]<br /></tt> |
| 858 |
</p> |
| 859 |
<h1> Useful links</h1> |
| 860 |
<ul> |
| 861 |
<li><a target="_blank" href="https://listes.cru.fr/wiki/srcmf/index">SRCMF wiki</a></li> |
| 862 |
<li><a target="_blank" href="http://srcmf.org">SRCMF website</a></li> |
| 863 |
<li><a |
| 864 |
target="_blank" href="http://www.ims.uni-stuttgart.de/projekte/TIGER/TIGERSearch/oldindex.shtml" |
| 865 |
>TIGERSearch website</a></li> |
| 866 |
<li><a target="_blank" href="http://bfm.ens-lyon.fr/">BFM website</a></li> |
| 867 |
<li><a target="_blank" href="http://textometrie.ens-lyon.fr/?lang=en">TXM website</a></li> |
| 868 |
</ul> |
| 869 |
</div> |
| 870 |
</body> |
| 871 |
</html> |