languagetool-org / languagetool

Style and Grammar Checker for 25+ Languages
https://languagetool.org
GNU Lesser General Public License v2.1
12.31k stars 1.39k forks source link

Issues with rules on "scambio del verbo avere" in Italian #681

Open sergiocallegari opened 7 years ago

sergiocallegari commented 7 years ago

Rules on "Scambio del verbo avere" in Italian have issues and cause tons of false positives. Specifically:

1) Rule "Suggerisci sempre lo scambio di 'a' con 'ha'" namely "Always suggest substituting 'ha' for 'a'" should in my opinion be removed. It is not based on grammar in any way, but just on a presumed frequency of occurrence of the forms "a" and "ha" in Italian. Would be like "Always suggest substituting 'to' for 'too'" in English, based on the fact that "to" occurs more frequently than "too". Having the rule disabled by default does not change the fact that this is not a grammar rule.

2) Rule "Suggerisci sempre lo scambio di 'ai' con 'hai'", same as above.

3) Rulegroup "scambio di 'a' ed 'ha'" causes ton of false positives. It contains two rules, one meant to catch "a" followed by a past participle, which seems OK. The second one meant to catch cases where one states possession of something that is available in plural form like "Piero ha case e cavalli" (Piero owns houses and horses) which could get miswritten as "Piero a case e cavalli. It is triggered whenever one has "a" followed by a plural name. Unfortunately, this also catches cases where "a" is used to introduce a location, as in forms like "Sono stato a congressi internazionali" (I have been at international conferences) or is used to introduce reference to something "Si può fare riferimento a partner internalzionali" (one can refer to international partners). I suggest removing this second rule from the ruleset or modifying it to exclude some constructs from the matching.

4) Rule "scambio di 'a' ed 'ai'", same as above. Matching "ai" followed by a plural name causes too many false positives like "Sono stato ai congressi più interessanti" or "Si può fare riferimento ai partner".

danielnaber commented 7 years ago

@paolob67, can you have a look?

paolob67 commented 7 years ago

Hi, those rules were created by a team that works with child that have learning problems to help them with the Italian subject at school. As such, I would not change them since they have this specific purpose. The user has the option of disabling them from the preferences panel if they don't want the false positives.

Thanks.

Paolo

Sent from my iPhone

On 10 Mar 2017, at 10:16, Sergio Callegari notifications@github.com wrote:

Rules on "Scambio del verbo avere" in Italian have issues and cause tons of false positives. Specifically:

Rule "Suggerisci sempre lo scambio di 'a' con 'ha'" namely "Always suggest substituting 'ha' for 'a'" should in my opinion be removed. It is not based on grammar in any way, but just on a presumed frequency of occurrence of the forms "a" and "ha" in Italian. Would be like "Always suggest substituting 'to' for 'too'" in English, based on the fact that "to" occurs more frequently than "too". Having the rule disabled by default does not change the fact that this is not a grammar rule.

Rule "Suggerisci sempre lo scambio di 'ai' con 'hai'", same as above.

Rulegroup "scambio di 'a' ed 'ha'" causes ton of false positives. It contains two rules, one meant to catch "a" followed by a past participle, which seems OK. The second one meant to catch cases where one states possession of something that is available in plural form like "Piero ha case e cavalli" (Piero owns houses and horses) which could get miswritten as "Piero a case e cavalli. It is triggered whenever one has "a" followed by a plural name. Unfortunately, this also catches cases where "a" is used to introduce a location, as in forms like "Sono stato a congressi internazionali" (I have been at international conferences) or is used to introduce reference to something "Si può fare riferimento a partner internalzionali" (one can refer to international partners). I suggest removing this second rule from the ruleset or modifying it to exclude some constructs from the matching.

Rule "scambio di 'a' ed 'ai'", same as above. Matching "ai" followed by a plural name causes too many false positives like "Sono stato ai congressi più interessanti" or "Si può fare riferimento ai partner".

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

sergiocallegari commented 7 years ago

Would it be possible to practice one or more of the following then?

1) Break up the rule sets in individual rules. I'm getting the impression that it is not possible to selectively disable specific rules from rule sets. At least in my Libreoffice with the LanguageTool extension, I am unable to keep the "a + past participle" (which is desirable) while disabling the "a + plural noun". The latter is too intrusive ("ha prodotto contributi a volumi", "ha organizzato eventi a congressi", "a risposto a bandi", ecc. all get marked) and at the same time too fragile ("Luca a soldi" is seen as problematic, while "Luca a molti soldi", which is equally wrong is not) to be useful, IMHO.

2) See if it is possible to have the "a + plural noun" rule modified to be more selective. E.g. Avoid triggering it if the "a" has a verb and a name or just a verb before it. At the same time make it less fragile, e.g. trigger on "a" + "plural adjective" + "plural name".

3) Have "a seguito di" as an exception to the "a" + "past participle" rule.