Vasta - Some issues I write some comments - LENE Location adverbs are not accepted as locative or illative: Ruoktot, vulos not accepted as illative Olgun, maddin, lulde, guhkkin not accepted as locative: ex. sii leat olgun: - svaret burde inneholde en lokativ LENE: This one is for Saara "Geaid X dát leat?" questions receive the feedback "Svaret burde inneholde en akkusativ". These are genitive questions. Although the accusative and genitive forms are similar, should we not make a distinction here? Also, it seems that there is a requirement to leave out the possession NP. Ex. Geaid niibi dát lea? Dat lea etniid/mánáid niibii - Svaret burde inneholde en akkusativ Dat lea mánáid. - Accepted LENE: This one I have fixed in sme-ped.cg3 by disambiguating "geaid" Numerals ========== Numeral rules requiring the gen form in nouns following a numeral bigger than 1, should not apply to plural forms. ex: Gallit gaskabeaivvit leat dáid beassážiid? Dáid beassážiid leat golmmat gaskabeaivvit (Not accepted, although it is correct) - Sjekk bøyninga av substantivet etter numeral Dáid beassážiid leat golmmat gaskabeivviid = (accepted as answer, although it is wrong) LENE: This one I have fixed in sme-ped.cg3 by improving the rules (were made for Sg) Gradation errors =================== Geaidda don attát gásaid. "Gássa" in the meaning of cardboard box has invisible gradation, also according to the ped lexicon. "Gássa" in the meaning "gas" is not natural in this context. LENE: Have added form-info to "gássa" in nouns.xml Maid lávllu-guovttos bargaba? Lávlu belongs to ACTOR, and should not have gradation. This is also according to our ped-lexicon, but still the gradated form occurs in the games. LENE: This one is for Saara - the system looks for lemma instead of lemma_id Rections ========== Rection verbs are only required to have objects of a certain case, when the verbs occur in the questions. However, these verbs can also be used as answers to the Maid X barget odne/ikte? questions, and the case of the following noun is not checked. All these answers were accepted: Sii njuvvet bohccuiguin. Sii balle gussii. Son humai áhči mielde. Nouns in the nominative case after transitive verbs are commented on: Sii lohket girji -Det passer ikke med nom. etter et transitivt verb. Sii lohket girjin =Accepted as answer. Should the obvious rection verbs also have a "det passer ikke..." feedback, for example ballat/luohttit/jáhkkit. LENE: I have started working with rections A different problem relating to rections, is that rection questions also seem to accept answers with Advs from DOHKO/DOPPE lists: Masa mun illudin? Don illudit dohko (accepted as answer) Masa sii liikojit? Sii liikojit deike (accepted as answer) I write "seem" because the DOPPE/DOHKO lists do not always function. LENE: I have started working with this Word order ============= Answers beginning with the negation are not accepted: Stunže go bagadeaddji odne? Ii son stunže - er du sikker på at du svarer i riktig person Son ii stunže odne: accepted In yes-no questions it is quite natural to begin with the negation. The feedback is also not informative for the user. LENE: This one I have fixed in sme-ped.cg3 by improving the rules (stunže was disambiguated as Imprt Sg2 - that is why it didn´t accept "Ii") Strange feedback ================= Sometimes the computer does not recognize a space between words: Maid don barggat odne? Odne mun oađašan - "Odnemun" finnes ikke i vårt leksikon. Kan det være en skrivefeil Correction: Mun oađašan odne: - "Munoađašan" finnes ikke i vårt leksikon osv. LENE: Saara is working with this one Other ==== Comitative questions which are only supposed to use COMV verbs, use all verbs, and we get strange questions like: geaiguiin dii lágidehpet? LENE: For Saara. I think the problem is that COMV is a "valency"-feature, and the system perhaps only looks for "sematics". "Man duohkái SUBJ MAINV" questions have defined MAINV to "čiehkádit", but the questions use all verbs, giving things like: Man duohkái gávpebargit málestedje LENE: For Saara.