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