An Gramadóir:
Bugs

Reporting bugs

There are inherent limitations to the approach An Gramadóir takes to grammar checking, and because of this there are errors (involving semantics) it is unlikely ever to catch. On the other hand, if you've looked through some of the known issues below and you think you've found something new, feel free to submit a bug report, support request, or a patch via the sourceforge tracker. Or just send me an email: kscanne at gmail dot com.


Lenition

The copula "gur" is either present/future (no lenition) or past/conditional (lenition) and deciding which sometimes requires distant contextual markers:
An amhlaidh gur daoine iad a chaith an lá ar fad ar an bhus?
Dar leis gur dhaoine réasúnta saibhir atá ina gcónaí i mBaile Átha Cliath.
The same goes for "ar" and all ambiguous combined forms: dar, faoinar, inar, lenar, sular, etc.

One needs semantic information sometimes to tell whether an adjective is being used attributively vs. predicatively, and this affects lenition sometimes:
"Seán broke a window"
Bhí fuinneog bhriste ag Seán.
"Seán had a broken window"
Bhí fuinneog briste ag Seán.
The second error is currently flagged, but at the expense of incorrectly flagging it when used predicatively. Another example:
"He had become famous"
Bhí cáil bhainte amach aige.
"He had a reaping tool"
Bhí uirlis bainte aige.
Or:
"The busy woman was in her office"
Bhí an bhean gnóthach ina hoifig.
"The woman was busy in her office"
Bhí an bhean ghnóthach ina hoifig.
Some adjectives that are almost always used predicatively are given as exceptions, so not flagged:
Bhí an fhéile curtha ar ceal.

"mar a" sometimes precedes an indefinite clause, sometimes definite, about 50/50 in the corpus (compare example below and CB p.146)
Beidh do shíol ina ndeoraithe i dtír nach leo, mar a bheidh siad ina ndaoir.
Chuala tú conas mar a mbeidh siad ina gceap uafáis agus mallacht.

The noun "méid" is either masculine or feminine, depending on semantics.
Intended to mean "amount, quantity" here, so should be masculine:
Ní smaoiníonn siad ar an mhéid eolais atá le fáil.
Intended to mean "size" here, so should be feminine:
Braitheann an méid a bhíonn i scriúnna ar an bhfeidhm a bhíonn le baint astu.

The preposition "ar" lenites in certain situations, sometimes depending on subtle semantic information:
Mar sin de, is é scéal na himirce scéal an pharóiste, ar dóigh.
Mar sin de, is é scéal na himirce scéal an pharóiste ar dhóigh.
Bhí clóscríobhán ar bord sa chúinne.
Bhí clóscríobhán ar bhord na loinge.
"A project that is established now"
Tionscadal atá ar bhun anois.
"A project that is on provisional basis"
Tionscadal atá ar bun sealadach.
"He beat them completely"
Bhuail sé iad ar fhad.
"He struck them across their cheek"
Bhuail sé iad ar fad a leicinn.
"It is in progress" (Ó Donaill, siúl.5)
Tá sé ar shiúl.
"He is gone" (Ó Donaill, siúl.6)
Tá sé ar siúl.
There are lots of similar examples where the unlenited form exists in a set phrase but the lenited form is perfectly legitimate as well. Here are some of the other top offenders: "ar d(h)eireadh", "ar b(h)all", "ar c(h)uairt", "ar t(h)alamh", etc. etc.

"idir" lenites about half the time, depending on semantics:
"There are both men and women in the park"
Tá idir fir agus mná sa pháirc.
"There is a game between men and women in the park"
Tá cluiche idir fhir agus mhná sa pháirc.

"chun" is ambiguous -- requires genitive in the first case:
Tháinig sé chun an cluiche.
Tháinig sé chun an chluiche a fheiceáil.

"a" is hard/impossible to disambiguate in many instances:
"He didn't like contradicting them."
Ní maith leis na daoine a bréagnú.
"The people didn't like her contradictions."
Ní maith leis na daoine a bhréagnú.
Currently, the first of this pair is reported as an error, in which case if the second meaning was intended, this becomes a false positive!
"I don't like person number two."
Ní maith liom duine a dhó.
"I don't like burning someone."
Ní maith liom duine a dó.

Don't yet have accurate markup of definite/indefinite nouns; CB p.16 #6
Rinne sé iarracht ar threoir a thabhairt ina lá féin conas cúraimí Dé a dheighilt ó ghnóthaí Sátain.
Tá Eifisigh, dá réir sin, ar an litir is fairsinge agus is "caitlicí" de litreacha Póil.
Is cosúil gur finné a scríobh é luath go leor tar éis báis Dháiví.

Anaphora resolution
Tá Kim Jong-Il ag smaoineamh ar a pobal.
Tá Kim Basinger ag smaoineamh ar a phobal.
Leabhar álainn clóbhuailte, arna foilsiú agam.
Tuairisc álainn clóbhuailte, arna fhoilsiú agam.
"Tomás is being praised."
Tá Tomás á moladh.
"Tomás is praising her."
Tá Tomás á mholadh.

No attempt at handling the vocative; CB p.12 #2
Bua ann féin é, a Treasa, go ndéanann duine a dhícheall!

Disambiguating past from imperative verbs
"Keep me on my toes"
Choinnigh mé ar mo bharraicíní.
"I kept on my toes"
Coinnigh mé ar mo bharraicíní.

Similarly, disambiguating imperfect from incorrectly lenited past autonomous
"The ling fishermen were killed."
Mharaíodh iascairí an langa.
"Fishermen used to kill the ling."
Maraíodh iascairí an langa.

Should have rule disallowing lenition on past autonomous, but disambiguating the 1st decl. past autonomous from the verbal noun is hard
"The man left his invitation."
Thréig an fear a cuireadh.
"Leave the man that was buried."
Tréig an fear a chuireadh.

Needs chunking: several adjectives goverened by same noun
Ní raibh baint aici le gníomhaíocht ar bith polaitiúil le cúig bliana anuas.


Eclipsis

Anaphora resolution
Tá brí ina colainn siúd!
Tá brí ina gcolainn siúd!

It can be hard to distinguish the genitive plural from the common plural:
Tuigimid freisin go bhfuil cathú ar dhaoine na dtithe seo a dhíol.
Tuigimid freisin go bhfuil cathú air fearas na tithe seo a dhíol.
Tá siad tar éis na Gardaí.
Tá siad tar éis na nGardaí a cháineadh.
Sometimes it is genuinely ambiguous:
Ní teanga na comharsan amháin a bhíonn i gceist.
Ní teanga na gcomharsan amháin a bhíonn i gceist.


Prefix t

Disambiguation of "leis":
Glacfar leis an am seo.
Glacfar leis an t-am seo.
Tagging genitive correctly without benefit of the prefix "t" cue:
"It was the business of the Old Testament prophets".
Ba é gnó fháithe an Sean-Tiomna é.
"The Old Testament was the business of translators".
Ba é gnó aistritheoirí an tSean-Tiomna é.
Tagging "an" correctly without benefit of the prefix "t" cue:
An suim is airde.
An tsuim agat?
"an" copula
Ní fios an ainm dílis air.
"an" article
Ní fios an t-ainm dílis air.


Prefix h

Anaphora resolution
Tá Kim Jong-Il ag smaoineamh ar a háit dúchais.
Tá Kim Basinger ag smaoineamh ar a áit dúchais.

Reputable sources have prefix "h" after "ní", others omit it, corpus is split.
Ní amháin dúinne in Éirinn ach do na milliúin eile ar fud na hEorpa.
Ní hamháin dúinne in Éirinn ach do na milliúin eile ar fud na hEorpa.
Ní ionann mise agus mo bhean.
Ní hionann mise agus mo bhean.


Agreement

Should have a rule flagging plural noun+singular adjective, but it is not always clear when the noun is attributive vs. predicative w/o parsing. It is indeed an error here:
"The blind people are there"
Tá na daoine dall ann.
But not here ("The people are blind"):
Tá na daoine dall.


Direct/indirect relative

"mar" introduces direct or indirect depending on semantics (CB p.146):
"Leave that as it is"
Fág siúd mar a bhfuil sé.
"Leave that where it is"
Fág siúd mar atá sé.


Other Rules

Should have a rule flagging noun+indefinite adjective (úd,sin,etc.)+descriptive adjective (CB p.60 #8) but this can be problematic without a shallow parse.
This is OK ("That man is tall"):
Tá an fear sin ard.
But here "ard" should precede "sin" ("That tall man is there"):
Tá an fear sin ard ann.

Would like to require comparative after "is" but it is hard to disambiguate the superlative marker from the copula:
Tá sé an lá is fada.
But this is fine:
Is fada an lá.

Should be a rule preventing "beirt", "triúr", etc. from being used with an impersonal noun, see CB p.78 Note. This requires some semantic markup.
beirt chat ann.

NGBC p.184 (b), should use singular noun after "ár", "bhur", plural "a" when each individual has just one of the noun in question:
Beidh siad ar ár n-aigní.
But this is OK if each of us has two cars:
Beidh siad inár gcarranna.


Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.