Minutes of the Work Group on Literary Prose: <title>held 2-3 Nov 1991, Hotel Ibis, Brussels <author>Lou Burnard <docnum>TEI &docfile. <date>&docdate. </titlep> <!> </frontm> <!> <body> <p>Present: Lou Burnard (LB), Tom Corns (TC; chair), Christian Delcourt (CD). <p>Apologies for absence: Paul Fortier and Don Ross. <h1>Agenda <p> TC welcomed members of the group and proposed the following (previously circulated) agenda: <ol> <li>The relationship of literary prose to other discourse. <li>Issues relating to the tagging of genre, in theory and in practice <li>Narratological aspects (Speakers, narrators, voices; ordering of events) <li>Tagging at the rhetorical level. <li>Tagging `exoticism' and other non-standard linguistic features. <li>Problems particular to miscellaneous kinds of literary prose. <li>Any other business. </ol> This agenda was agreed, though not taken in exactly the same order as above. <p>In answer to a question about the form that the final report from the work group should take, LB said that while the exact form of the report was a matter for the work group head, incorporation of its substantial recommendations into the text of P2 would be greatly simplified if these were expressed as lists of elements, with their respective tags and attributes, brief definitions and illustrative examples. <h1>Literary discourse and Situational Parameters <p>Discussion focussed initially on whether the situational parameters proposed by TR6 could be adopted for all literary discourse. CD noted that the report of that workgroup did not distinguish <term>re''cit</term> from <term>histoire</term>: it treated discourse independently of its factive content. TC noted that it was concerned only with the situation in which discourse was produced. LB asked whether similar parameters were not appropriate to both. TC felt this might be possible, but that extensions might be needed for example the `mode' of a fictive situation should include values such as `thought', `felt', etc. <p> The distinction between the situation within which a text was itself created, and the situations displayed mimetically within it was felt to be a helpful one. The situational parameters proposed by TR6 were felt to be a scientific way of negotiating the problems of defining genre (see further below); their possible application to the content of a text was agreed to be an exciting extension, of considerable use. CD cited as example the construction of corpora in which the treatment of similar situations by texts emanating from different cultural contexts could be contrasted. He referred to work of this kind carried out on a limited scale by Kristeva et al. As a minimum it should be possible to locate the purported subject of a literary work in time, place and social class. <p> It was not clear that the `topic' heading proposed by TR6 was sufficient for this. CD cited as a further example the contrastive study of idiolects for the author of a work and the characters contained it it. It was noted however that this level of description would be of most use at the text level of description, rather than throughout the course of a text. <h1>Literary and non-literary prose <p>It was agreed that there was a continuum between literary and non-literary prose. Categories blurred inevitably: for example between factual historical writing and fiction. CD noted that in history, speculation is presented as speculation, whereas in fiction, fact is not distinguished as such. The difference between prose and prose poem, or verse and prose in such authors as Wulfstan, depended as much on interpretative strategy as on any formal property. CD observed that metrical effects were forbidden in French prose. TC noted that it should be possible to use the tags proposed by TR11 to mark metrical effects independently of the text type being marked up. This implied that encoding for metre or prosody would need to be added to elements other than verse lines, including arbitrary sequences of words. LB suggested that this might be a use for the arbitrary segmentation <tag>s</tag> tag. <h1>Genre <p>The notion of genre was agreed to be an entirely culturally-determined phenomenon, unstable over time. LB asked whether it might be identified with TR6's parameter 13 (perceived value). This was agreed to be inappropriate as the same perceived genre could have different values in different contexts (for example, `comic-book' vs. `fumetti'). A further category `perceived kind' was proposed. Other examples cited were Vergil's <q>Messianic eclogue</q> which had been regarded both as panegyric and as christian prophecy; Defoe's <cit>Robinson Crusoe</cit>, variously regarded as an allegory of salvation, the autobiography of a historical figure, an adventure story, a novel etc. It was agreed that there have been genre taxonomies throughout history, and that (perhaps partly because of the tendency of literary writers to redefine their work with relation to existing taxonomies) they have continually mutated. This was not however a reason to discommend their use. On the contrary the group felt that providing genre information along a diachronic scale would be very valuable. The proposed parameter `perceived kind' should contain information along the lines <q>this text was called a <term>blort</term> in description <term>farble</term></q>; the perceived assessment being supported by a specific bibliographic reference wherever possible. <h1>Rhetoric <p> The group briefly reviewed sources of information about rhetorical figures. These included a transcription of a specific 16th century English rhetoric made by Don Ross; a generic 20th century glossary of rhetorics (Vickers: <cit>In defence of rhetoric</cit>, 1989); a modern classification of rhetorical tropes by their internal structure (Rhetorique ge''nerale, Groupe Mu, 1968). Three areas of concern were identified: <ul> <li>definition of rhetorical structure per se (can the tropes be organised into a meaningful taxonomy?) <li>tagging of trope instances - do they have internal structure? <li>the same text sequence can participate in more than one rhetorical classification </ul> LB suggested that the f.struct mechanism might be used to represent rhetorical tropes, and sketched an example feature structure for zeugma. CD noted that in some cases a subcomponent of a rhetorical trope might be omitted e.g. synecdoche (part for whole), or most obviously metaphor, where the vehicle is present in the text but the tenor is not. This was felt to be surmountable. Where less rigorous or detailed tagging was felt desirable, the note tag might be used. TC noted that if rhetoric were to be reclaimed as a valid structuring of text, then we would have to take it seriously and on its own terms. CD noted that a further attraction of the feature structure mechanism was that it allowed access to the internal structure of a trope, hence reducing the importance, for analytic purposes, of its name. The group concluded that the use of feature structures for representation of rhetorical figures should be investigated more closely, and proposed in the guidelines. Where less exact tagging was required, the existing note tag would be adequate. To propose specific tags for any one rhetorical scheme was felt unproductive. <h1>Relevant text types <p> Over lunch, the group went into brainstorm mode and produced the following initial list of text types, each of which it was felt might require the identification of specific structural elements as noted below: <ul> <li>epistle: addressee, addressor, signature <li>sermon: `proof text' exegesis <li>devotional work: narrative, meditation <li>journal or diary: date, place <li>broadside: illustration with comments: hard to identify title as such etc <li>fable: moral, exemplum <li>comic books: frame, directions which overlap, reln between frames <li>prayer book,liturgy <li>almanacs <li>encyclopaedias, emblem books <li>form-filling documents <li>instructional manuals: decision points <li>inscriptions etc on coins, pots, tombs, monuments <li>elegy <li>commentary <li>libretto <li>florilegium/anthology or commonplace book (mss collections organised by topic; logical/temporal may not fit with physical structure) <li>running titles which change <li>symposia <li>disputations <li>representations of structure e.g. contents page of Burton's anatomy <li>collections of notes (xrefs for same) <li>acts and orders of parliament </ul> <h1>Narratological concerns <h2>7.1. Speaker <p> The discussion of narrative in P1 (7.3.2) needed to be expanded to address the narrator's role. Changes of voice within dialogue were straightforward; indirect quotation slightly less so. Problems of boundary definition and overlap between narrated segments were self evident. After discussion the group agreed on three clearly distinguishable cases: <ul> <li>direct speech <li>speech marked by verba dicendi <li>speech not marked as such </ul> A voice could, and for some kinds of analysis, should be associated with each of these. Determination of the speaker for the last kind might be determined by deixis or by characteristic vocabulary. An article by Ross <fn><cit>Who's talking? how characters become narrators in fiction</cit>: MLN 91 (1976) pp. 1222-42</fn> listed various deictic and other features that signalled shift of narrative voice. `Pure narrative' as such was not a special case: it represented the speech of a narrator. <p> Specific recommendations: the <tag>in.quot</tag> tag should be removed: it's just speech by the narrator. The <tag>q</tag> tag should be allowed to nest within itself. <note>This needs more thought: consider the following <xmp> <![ CDATA [ <q sp=a>She said <q sp=b>Cluck cluck</q> on Thursday</q> <q sp=a>This <q sp=b>she said </q>is not as easy as it looks</q> ]]> </xmp> In both cases speaker A's remarks are interrupted by speaker B. In the first case, however, <q>cluck cluck</q> is speaker B as reported by speaker A, while in the second speaker A is being reported by speaker B.</note> <p> An attribute `indicator' was suggested as a means of indicating the grounds for attributing a <tag>q</tag> to a particular speaker, with possible values `punctuation' (the default), `verbdic', `other'. <p> Indeterminacy of boundary was felt to be an inherently insoluble problem. An attribute `precision' might be used to indicate the approximate number of words either side of the tag where the speech division occurred. LB noted that there were two kinds of uncertainty: is it a <tag>q</tag> at all? and is the speaker correctly identified? The group suggested a single indeterminate speaker might be defined as target for uncertain attributions. TC noted that this should only rarely be used: more usually, the value would be a list of idrefs indicating the likely candidates. <h2>7.2. Time <p>A narrative presents a series of events in time, but the normal temporal sequence may be violated in various ways e.g. repetition, flashback, prediction, synchrony. With some reluctance, the group proposed that events (or the fictive elements corresponding to them) might be tagged, in which case existing mechanisms such as the AI2 proposals for a timeline might be used to mark their temporal position. CD noted the popularity of this kind of analysis, citing Genette. It was agreed that it might be possible to adapt from Barthes' <cit>S/Z</cit> the notion of identifying those lexemes which carry forward the sequence of the action; such units of the text could then be given identifiers which could be related to points on the timeline. might be a helpful way of structuring narrative. <h1>``exoticisms'' It was agreed that there was a need to extend the notion of `foreign' already present in P1 to tag a variety of exotic words or phrases, including such features as dialect, tabu, demotic, neologism, archaism, technical term, poetic usage and regionalism. After discussion it was agreed that these features could all be characterised along three dimensions <ul> <li> diatopie ( place) <li> diachronie (time) <li> diastratie (social class) </ul> The group therefore proposed the single tag <tag>marked</tag>, with three attributes as follows: <gl> <gt>marked <gd>marks a word or phrase analytically distinguished from its surroundings along one or more of the dimensions of topology, chronology or stratum. </gl> Its attributes: <gl> <gt>place <gd>name of a place <gt>time <gd>neologism, archaism, pseudo-archaism <gt>register <gd>poetic,demotic,tabu,technical <gt>lang, rendition, n, id <gd>as usual </gl> <h1>Review of TR6 W1 <p> The second day of the meeting began with a more detailed review of the situational parameters proposed in the report of TR6, with a view to determining which of these would be most appropriate for recording information of interest in analysing literary texts. <ul> <li>Should verse prose and drama be explicitly distinguished in the header? LB noted that low level tags would indicate presence of verse, prose or drama within a text. For example, <xmp> <![ CDATA [ He said <q sp=Fred>Rats!</q> <sp speaker=FRED> Rats!</sp> ]]> </xmp> Either of these might appear in a novel with similar significance, but the former would be tagged as <tag>q</tag>, because it was presented as direct speech; the latter as <tag>sp</tag> because it was presented as drama. CD felt that we should document at the highest level, if only for indexing purposes as we had already agreed for genre and topic. LB suggested that the `text-category' field in the header would be the most appropriate place for this. TC agreed to discuss this issue with the heads of TR10 and TR11. <li>Domain of use: this was felt to be a very useful way of characterising literary texts. <li>10: purpose : this category was felt to be too naive for the categorisation of literary texts. For example, we felt it would be hard to distinguish edification from persuasion. <li>11: topic - as previously noted, library classification schemes were regarded as particularly inadequate for literary texts because they lacked topic information. <li>Dramatis personae : it was not clear where these should most properly be included. If they were regarded as participants, an attribute under 4 (d) `fictiveness of participating voice' would be needed. Alternatively, the list of primary producer/addressors should be followed by a list of represented producers. <li>5,6,7: very useful <li>13 perceived value : this should be expanded to perceived status; it might then include our proposals above about genre and reception. We noted that several of the value terms given in the TR6 report are associated with hierarchies of genre that genre taxonomies have traditionally produced.<fn>This conclusion substantially altered the provisional position arrived at the previous day.</fn> <li>14: Plagiarism should be distinguished from issues of transcription since it involves a value judgment. It is a recension of a text which denies or disguises its real origins. LB noted that this category was not intended to capture the relation between this text and others, just its `authenticity'. </ul> <h1>Miscellanea <p> We reviewed the list of text types prepared on the previous day and agreed that it would be helpful to organise them into some kind of hierarchy. A start should be made on identifying higher level abstractions required by several texttypes e.g. the privileging of part of the text as exemplum, moral, prooftext. <p> In discussion, the need for a general purpose tag to identify such features as sententiae, proverbs, oaths, maxims, catchphrases was identified. The tag <tag>formula</tag> was proposed with attributes <att>type</att> and <att>reference</att>. The latter optionally specified a canonical source, e.g. Tilley number for proverbs. <h1>Concluding actions <ul> <li>LB to prepare and circulate draft minutes by 5 Nov <li>TC/CD to approve draft for distribution within 24 hours <li>TC to draft initial report and circulate by 9 Nov </ul> In conclusion, the group thanked CD for having organised the meeting. </body> <!> </gdoc>

.sr docfile = &sysfnam. ;.sr docversion = 'Draft';.im teigmlp1 .* Document proper begins. .sr docdate '10 November 1991' Minutes of the Work Group on Literary Prose: <title>held 2-3 Nov 1991, Hotel Ibis, Brussels <author>Lou Burnard <docnum>TEI &docfile. <date>&docdate. </titlep> <!> </frontm> <!> <body> <p>Present: Lou Burnard (LB), Tom Corns (TC; chair), Christian Delcourt (CD). <p>Apologies for absence: Paul Fortier and Don Ross. <h1>Agenda <p> TC welcomed members of the group and proposed the following (previously circulated) agenda: <ol> <li>The relationship of literary prose to other discourse. <li>Issues relating to the tagging of genre, in theory and in practice <li>Narratological aspects (Speakers, narrators, voices; ordering of events) <li>Tagging at the rhetorical level. <li>Tagging `exoticism' and other non-standard linguistic features. <li>Problems particular to miscellaneous kinds of literary prose. <li>Any other business. </ol> This agenda was agreed, though not taken in exactly the same order as above. <p>In answer to a question about the form that the final report from the work group should take, LB said that while the exact form of the report was a matter for the work group head, incorporation of its substantial recommendations into the text of P2 would be greatly simplified if these were expressed as lists of elements, with their respective tags and attributes, brief definitions and illustrative examples. <h1>Literary discourse and Situational Parameters <p>Discussion focussed initially on whether the situational parameters proposed by TR6 could be adopted for all literary discourse. CD noted that the report of that workgroup did not distinguish <term>re''cit</term> from <term>histoire</term>: it treated discourse independently of its factive content. TC noted that it was concerned only with the situation in which discourse was produced. LB asked whether similar parameters were not appropriate to both. TC felt this might be possible, but that extensions might be needed for example the `mode' of a fictive situation should include values such as `thought', `felt', etc. <p> The distinction between the situation within which a text was itself created, and the situations displayed mimetically within it was felt to be a helpful one. The situational parameters proposed by TR6 were felt to be a scientific way of negotiating the problems of defining genre (see further below); their possible application to the content of a text was agreed to be an exciting extension, of considerable use. CD cited as example the construction of corpora in which the treatment of similar situations by texts emanating from different cultural contexts could be contrasted. He referred to work of this kind carried out on a limited scale by Kristeva et al. As a minimum it should be possible to locate the purported subject of a literary work in time, place and social class. <p> It was not clear that the `topic' heading proposed by TR6 was sufficient for this. CD cited as a further example the contrastive study of idiolects for the author of a work and the characters contained it it. It was noted however that this level of description would be of most use at the text level of description, rather than throughout the course of a text. <h1>Literary and non-literary prose <p>It was agreed that there was a continuum between literary and non-literary prose. Categories blurred inevitably: for example between factual historical writing and fiction. CD noted that in history, speculation is presented as speculation, whereas in fiction, fact is not distinguished as such. The difference between prose and prose poem, or verse and prose in such authors as Wulfstan, depended as much on interpretative strategy as on any formal property. CD observed that metrical effects were forbidden in French prose. TC noted that it should be possible to use the tags proposed by TR11 to mark metrical effects independently of the text type being marked up. This implied that encoding for metre or prosody would need to be added to elements other than verse lines, including arbitrary sequences of words. LB suggested that this might be a use for the arbitrary segmentation <tag>s</tag> tag. <h1>Genre <p>The notion of genre was agreed to be an entirely culturally-determined phenomenon, unstable over time. LB asked whether it might be identified with TR6's parameter 13 (perceived value). This was agreed to be inappropriate as the same perceived genre could have different values in different contexts (for example, `comic-book' vs. `fumetti'). A further category `perceived kind' was proposed. Other examples cited were Vergil's <q>Messianic eclogue</q> which had been regarded both as panegyric and as christian prophecy; Defoe's <cit>Robinson Crusoe</cit>, variously regarded as an allegory of salvation, the autobiography of a historical figure, an adventure story, a novel etc. It was agreed that there have been genre taxonomies throughout history, and that (perhaps partly because of the tendency of literary writers to redefine their work with relation to existing taxonomies) they have continually mutated. This was not however a reason to discommend their use. On the contrary the group felt that providing genre information along a diachronic scale would be very valuable. The proposed parameter `perceived kind' should contain information along the lines <q>this text was called a <term>blort</term> in description <term>farble</term></q>; the perceived assessment being supported by a specific bibliographic reference wherever possible. <h1>Rhetoric <p> The group briefly reviewed sources of information about rhetorical figures. These included a transcription of a specific 16th century English rhetoric made by Don Ross; a generic 20th century glossary of rhetorics (Vickers: <cit>In defence of rhetoric</cit>, 1989); a modern classification of rhetorical tropes by their internal structure (Rhetorique ge''nerale, Groupe Mu, 1968). Three areas of concern were identified: <ul> <li>definition of rhetorical structure per se (can the tropes be organised into a meaningful taxonomy?) <li>tagging of trope instances - do they have internal structure? <li>the same text sequence can participate in more than one rhetorical classification </ul> LB suggested that the f.struct mechanism might be used to represent rhetorical tropes, and sketched an example feature structure for zeugma. CD noted that in some cases a subcomponent of a rhetorical trope might be omitted e.g. synecdoche (part for whole), or most obviously metaphor, where the vehicle is present in the text but the tenor is not. This was felt to be surmountable. Where less rigorous or detailed tagging was felt desirable, the note tag might be used. TC noted that if rhetoric were to be reclaimed as a valid structuring of text, then we would have to take it seriously and on its own terms. CD noted that a further attraction of the feature structure mechanism was that it allowed access to the internal structure of a trope, hence reducing the importance, for analytic purposes, of its name. The group concluded that the use of feature structures for representation of rhetorical figures should be investigated more closely, and proposed in the guidelines. Where less exact tagging was required, the existing note tag would be adequate. To propose specific tags for any one rhetorical scheme was felt unproductive. <h1>Relevant text types <p> Over lunch, the group went into brainstorm mode and produced the following initial list of text types, each of which it was felt might require the identification of specific structural elements as noted below: <ul> <li>epistle: addressee, addressor, signature <li>sermon: `proof text' exegesis <li>devotional work: narrative, meditation <li>journal or diary: date, place <li>broadside: illustration with comments: hard to identify title as such etc <li>fable: moral, exemplum <li>comic books: frame, directions which overlap, reln between frames <li>prayer book,liturgy <li>almanacs <li>encyclopaedias, emblem books <li>form-filling documents <li>instructional manuals: decision points <li>inscriptions etc on coins, pots, tombs, monuments <li>elegy <li>commentary <li>libretto <li>florilegium/anthology or commonplace book (mss collections organised by topic; logical/temporal may not fit with physical structure) <li>running titles which change <li>symposia <li>disputations <li>representations of structure e.g. contents page of Burton's anatomy <li>collections of notes (xrefs for same) <li>acts and orders of parliament </ul> <h1>Narratological concerns <h2>7.1. Speaker <p> The discussion of narrative in P1 (7.3.2) needed to be expanded to address the narrator's role. Changes of voice within dialogue were straightforward; indirect quotation slightly less so. Problems of boundary definition and overlap between narrated segments were self evident. After discussion the group agreed on three clearly distinguishable cases: <ul> <li>direct speech <li>speech marked by verba dicendi <li>speech not marked as such </ul> A voice could, and for some kinds of analysis, should be associated with each of these. Determination of the speaker for the last kind might be determined by deixis or by characteristic vocabulary. An article by Ross <fn><cit>Who's talking? how characters become narrators in fiction</cit>: MLN 91 (1976) pp. 1222-42</fn> listed various deictic and other features that signalled shift of narrative voice. `Pure narrative' as such was not a special case: it represented the speech of a narrator. <p> Specific recommendations: the <tag>in.quot</tag> tag should be removed: it's just speech by the narrator. The <tag>q</tag> tag should be allowed to nest within itself. <note>This needs more thought: consider the following <xmp> <![ CDATA [ <q sp=a>She said <q sp=b>Cluck cluck</q> on Thursday</q> <q sp=a>This <q sp=b>she said </q>is not as easy as it looks</q> ]]> </xmp> In both cases speaker A's remarks are interrupted by speaker B. In the first case, however, <q>cluck cluck</q> is speaker B as reported by speaker A, while in the second speaker A is being reported by speaker B.</note> <p> An attribute `indicator' was suggested as a means of indicating the grounds for attributing a <tag>q</tag> to a particular speaker, with possible values `punctuation' (the default), `verbdic', `other'. <p> Indeterminacy of boundary was felt to be an inherently insoluble problem. An attribute `precision' might be used to indicate the approximate number of words either side of the tag where the speech division occurred. LB noted that there were two kinds of uncertainty: is it a <tag>q</tag> at all? and is the speaker correctly identified? The group suggested a single indeterminate speaker might be defined as target for uncertain attributions. TC noted that this should only rarely be used: more usually, the value would be a list of idrefs indicating the likely candidates. <h2>7.2. Time <p>A narrative presents a series of events in time, but the normal temporal sequence may be violated in various ways e.g. repetition, flashback, prediction, synchrony. With some reluctance, the group proposed that events (or the fictive elements corresponding to them) might be tagged, in which case existing mechanisms such as the AI2 proposals for a timeline might be used to mark their temporal position. CD noted the popularity of this kind of analysis, citing Genette. It was agreed that it might be possible to adapt from Barthes' <cit>S/Z</cit> the notion of identifying those lexemes which carry forward the sequence of the action; such units of the text could then be given identifiers which could be related to points on the timeline. might be a helpful way of structuring narrative. <h1>``exoticisms'' It was agreed that there was a need to extend the notion of `foreign' already present in P1 to tag a variety of exotic words or phrases, including such features as dialect, tabu, demotic, neologism, archaism, technical term, poetic usage and regionalism. After discussion it was agreed that these features could all be characterised along three dimensions <ul> <li> diatopie ( place) <li> diachronie (time) <li> diastratie (social class) </ul> The group therefore proposed the single tag <tag>marked</tag>, with three attributes as follows: <gl> <gt>marked <gd>marks a word or phrase analytically distinguished from its surroundings along one or more of the dimensions of topology, chronology or stratum. </gl> Its attributes: <gl> <gt>place <gd>name of a place <gt>time <gd>neologism, archaism, pseudo-archaism <gt>register <gd>poetic,demotic,tabu,technical <gt>lang, rendition, n, id <gd>as usual </gl> <h1>Review of TR6 W1 <p> The second day of the meeting began with a more detailed review of the situational parameters proposed in the report of TR6, with a view to determining which of these would be most appropriate for recording information of interest in analysing literary texts. <ul> <li>Should verse prose and drama be explicitly distinguished in the header? LB noted that low level tags would indicate presence of verse, prose or drama within a text. For example, <xmp> <![ CDATA [ He said <q sp=Fred>Rats!</q> <sp speaker=FRED> Rats!</sp> ]]> </xmp> Either of these might appear in a novel with similar significance, but the former would be tagged as <tag>q</tag>, because it was presented as direct speech; the latter as <tag>sp</tag> because it was presented as drama. CD felt that we should document at the highest level, if only for indexing purposes as we had already agreed for genre and topic. LB suggested that the `text-category' field in the header would be the most appropriate place for this. TC agreed to discuss this issue with the heads of TR10 and TR11. <li>Domain of use: this was felt to be a very useful way of characterising literary texts. <li>10: purpose : this category was felt to be too naive for the categorisation of literary texts. For example, we felt it would be hard to distinguish edification from persuasion. <li>11: topic - as previously noted, library classification schemes were regarded as particularly inadequate for literary texts because they lacked topic information. <li>Dramatis personae : it was not clear where these should most properly be included. If they were regarded as participants, an attribute under 4 (d) `fictiveness of participating voice' would be needed. Alternatively, the list of primary producer/addressors should be followed by a list of represented producers. <li>5,6,7: very useful <li>13 perceived value : this should be expanded to perceived status; it might then include our proposals above about genre and reception. We noted that several of the value terms given in the TR6 report are associated with hierarchies of genre that genre taxonomies have traditionally produced.<fn>This conclusion substantially altered the provisional position arrived at the previous day.</fn> <li>14: Plagiarism should be distinguished from issues of transcription since it involves a value judgment. It is a recension of a text which denies or disguises its real origins. LB noted that this category was not intended to capture the relation between this text and others, just its `authenticity'. </ul> <h1>Miscellanea <p> We reviewed the list of text types prepared on the previous day and agreed that it would be helpful to organise them into some kind of hierarchy. A start should be made on identifying higher level abstractions required by several texttypes e.g. the privileging of part of the text as exemplum, moral, prooftext. <p> In discussion, the need for a general purpose tag to identify such features as sententiae, proverbs, oaths, maxims, catchphrases was identified. The tag <tag>formula</tag> was proposed with attributes <att>type</att> and <att>reference</att>. The latter optionally specified a canonical source, e.g. Tilley number for proverbs. <h1>Concluding actions <ul> <li>LB to prepare and circulate draft minutes by 5 Nov <li>TC/CD to approve draft for distribution within 24 hours <li>TC to draft initial report and circulate by 9 Nov </ul> In conclusion, the group thanked CD for having organised the meeting. </body> <!> </gdoc>