# Verb-initial grammars: A multilingual/parallel perspective
# ESRC Project RES-000-23-0505
# Oxford University
# Charles Randriamasimanana

# Malagasy phonetics/phonology & morphology:
# Malagasy ka, tra, na endings of words: ESRC-OX-05-CR104.

# Weak vs strong roots or rather weak vs strong syllables?

# In Malagasy many words end up with a syllable ka, tra or na and certain morphophonological
# processes are likely to occur. This phenomenon was already noticed by grammarians such as
# Aristide Marre in his Grammaire Malgache, Imprimerie Vosgiennes, Epinal, France 1894: 40-46; 
# or R.P. Malzac in his Grammaire Malgache (re-edited and published in 1950:6, 13-14) or in 
# his Dictionnaire Malgache-Français (page VIII), which he completed in 1888. The same point
# is made in Essai de Grammaire Malgache, Société d'Editions Géographiques, Maritimes et Coloniales
# 1931: 58-60. The same view was taken up by most if not all subsequent grammarians writing about 
# Malagasy till Keenan and Polinsky (1998:581) came up with a novel definition.

# Problem # 1: Keenan & Polinsky (1998:581) make the general claim that the shape of the Malagasy
# passive suffix is (C)Vna, where C represents a consonant, V a vowel and na the final syllable of
# the passive form. According to Keenan & Polinsky (1998) the choice of the relevant vowel /a/ or /i/
# is determined by the verb root. Hence their distinction between weak & strong roots. However the 
# data from roots with a passive meaning (see document entitled Distribution of passive affixes in 
# Malagasy) clearly suggests that a PATIENT/THEME-subject requires suffix ina --by and large; whereas
# an ...-Oblique-subject requires suffix ana. So the distinction between weak vs strong roots does not
# appear to be relevant at all.

# Problem # 2: Keenan & Polinsky (1998:577) make the claim that the weak vs strong nature of the
# root plays a role in the drop of final -ka or -tra in such examples as
  tapa-dRabe  (0! 0 0 0)
# tapaka 'broken' Rabe 'name' tapa-dRabe 'broken by Rabe';
  fanta-dRabe  (0! 0 0 0)
# fantatra 'known' Rabe 'name' fanta-dRabe 'known by Rabe'
# This yields sequences which do not exist in Malagasy as neither tapa-dRabe nor fanta-dRabe
# can qualify as Malagasy! See below a description of phenomena involving a drop of
# { ka, tra, na } not just in roots, but in specific environments in all kinds of words.

# Instead of the forms proposed in K & P, in Colloquial Malagasy we will have something like 
  tapak'i Rabe
# tapak+O       i    Rabe
# be-decided-by D.sg Rabe
# "has been/was decided by Rabe'
# where O represents the elided final vowel a of the final syllable ka; in the actual spelling
# of the sequence, this O is replaced by an apostrophe;

  fantatr'i Rabe
# fantatr+O   i Rabe
# be-known-to D.sg Rabe
# 'be known to Rabe'
# where O represents the elided final vowel a of the final syllable tra; in the actual spelling
# of the sequence, this O is replaced by an apostrophe.

# From the above data it is obvious that the relevant unit is not the root, i.e. the word, but 
# rather the relevant syllable --obviously we also need to take the environment into account.

# Instead of the above K & P sequences, in standard Malagasy, we will have something like

  tapak'Ingahy Rabe
# tapak+O       Ingahy    Rabe
# be-decided-by D.sg      Rabe
# "has been/was decided by Mr. Rabe'
# where O represents the elided final vowel a of the final syllable ka; in the actual spelling
# of the sequence, this O is replaced by an apostrophe;

 fantatr'Ingahy Rabe
# fantatr+O   Ingahy Rabe
# be-known-to D.sg   Rabe
# 'be known to Mr. Rabe'
# where O represents the elided final vowel a of the final syllable tra; in the actual spelling
# of the sequence, this O is replaced by an apostrophe;

# where the apostrophe introduces an AGENT in the genitive case. Now elision is a pretty general 
# phenomenon when it comes to encoding a definite AGENT even when the root does not end up with 
# any of { ka, tra, na} syllables, i.e. in K & P's terms is NOT a weak root. This additional 'na', 
# which shows up after the A-passive, i.e. 'passive.prefix.A + root' is labelled 'epenthetic na'
# [ in later documents, we will use the label 'clitic NA'] as opposed to passive voice suffix 
# '(i)na' or '(a)na':

  Nalefan'i Jeanne ilay entana.
# n-a-lefa-na         i    Jeanne ilay  entana
# past-a.pass-lefa-na D.sg Jeanne DX.sg luggage
# 'The (previous mention) luggage was sent by Jeanne'

# where the active voice is mandefa m+an+(l>d)efa 'pres-prf.an-root.lefa' > m+an+defa'to send';
# passive 1 form = alefa or alternatively alefaso from the passive 1 imperative alefao a+lefa+u 
#'a.pass-root.lefa-suffix.u' or alefaso a+lefa+s+u 'a.pass-stem.lefas-suffix.u' written o
# '(let-it)-be-sent-by.you' and where -n represents something like 'by+AGENT' in the genitive.
# Note that the final vowel a of epenthetic na of n-a-lefa-na has been elided in the vicinity
# of D.sg 'i', a vowel. Note the definite AGENT comprising D.sg 'i' in 'i Jeanne' introduced by
# the epenthetic na. In this connection, note that epenthetic 'na' --unlike suffix '(i)na' or
# '(a)na'-- does not trigger stress shift so that stress falls on 'le' in 'nalefa', and NOT
# of 'fa' as would be the case if this was a case of suffixation.

# NOTE: The a.passive is analysed as a+lefa only; to which is added the epenthetic 'na' [ more
# accurately 'clitic NA', as will be proposed in later documents ] to introduce the AGENT in the
# genitive case. In other words, under this analysis A-passive does not comprise a suffix 'ina'. 
# The primary reason for the existence of this epenthetic 'na' is provided immediately below.

  Nalefana mpianatra ilay entana.
# n-a-lefa-na                 mpianatra    ilay  entana
# past-a.pass-root.lefa-ep.na a/student(s) DX.sg luggage
# 'The (previous mention) luggage was sent by a/some student(s)'.
# Note that the passive 1 indicative form is nalefa n-a-lefa 'past-a.pass-lefa' 
# to which has been added the epenthetic 'na''by'. In this specific case, 
# the AGENT is non-definite;

  Navelan'i Jeanne ilay entana.
# n-a-vela-na                 i    Jeanne ilay  entana
# past-a.pass-root.vela-ep.na D.sg Jeanne DX.sg luggage
# 'The (p.m.) luggage was left behind by Jeanne';

# where the active voice is mamela m+a(n>m)+(v)ela 'pres-prf.an-root.vela' 
# > m+am+ela 'to leave behind'; passive 1 form = avela from passive 1 imperative 
# avelao a+vela+u 'a.pass-vela-suffix.u''(let-it)-be-left-behind-by.you'.
# Note that the final vowel a of the epenthetic na of n-a-vela-na 
# has been elided in the vicinity of D.sg 'i', a vowel;
# Because we are dealing here with epenthetic na --NOT suffix '(i)na' or
# '(a)na'-- the process does not trigger stress shift; so stress falls on 've' in 'navela', and NOT
# of 'la' as would be the case if this was a case of suffixation.


  Navelana mpianatra ilay entana.
# n-a-vela-na                 mpianatra    ilay  entana
# past-a.pass-root.vela-ep.na a/student(s) DX.sg luggage
# 'The (p.m.) luggage was left behind by a/some students(s)';
# Again, because we are dealing here with epenthetic na --NOT suffix '(i)na' or
# '(a)na'-- the process does not trigger stress shift; so stress falls on 've' in 'navela', and NOT
# of 'la' as would be the case if this was a case of suffixation.


# Note also the following forms involving A-passive 1 verbs with a non-definite AGENT:

  Napetrak'olona ilay entana.
# n+a+petrak'                olona  ilay  entana
# past-a.pass-root.petra<ka> people DX.sg luggage
# 'The (previous mention) luggage was deposited by (some) people';

  Napetraka mpianatra ilay entana.
# n+a+petraka                mpianatra    ilay  entana
# past-a.pass-root.petra<ka> a/student(s) DX.sg luggage
# 'The (p. m.) luggage was deposited by a/some student/students';

  Natolotr'olona ilay entana.
# n+a+tolotr'                olona  ilay  entana
# past-a.pass-root.tolo<tra> people DX.sg luggage
# 'The (p.m.) luggage was handed over by (some) people';

  Natolotra mpianatra ilay entana.
# n+a+tolotra                mpianatra  ilay  entana
# past-a.pass-root.tolo<tra> mpianatra  DX.sg luggage
# 'The (p.m.) luggage was handed over by a/some student/students';

# where the apostrophe now represents an non-definite AGENT in the genitive 
# in the vicinity of a word beginning with a vowel.

# In addition, we note special forms for { ka, tra } in A.passive 1 verbs 
# involving a definite AGENT indicated by the definite article ny:

  Napetraky ny ankizy ilay entana.
# n+a+petraka+ny           ny ankizy   ilay  entana
# past-a.pass-petra<ka+ny> Det.unspec  DX.sg luggage
# 'The (previous mention) luggage was deposited by the children'.

  Natolotry ny ankizy ilay entana.
# n+a+tolotra+ny           ny         ankizy   ilay  entana
# past-a.pass-tolo<tra+ny> Det.unspec children DX.sg luggage
# 'The (p.m.) luggave was handed over by the children'.

  Tsy namelan'                 ny         ankizy   ny         entana.
# not n+a(n>m)+(v)ela+<ana+ny> Det.unspec children Det.unspec luggage
#     n+am+ela+n
# 'None of the luggage was left behind by the children';

# where <ka+ny> becomes ky and <tra+ny> becomes try and where the
# mutation is influenced by the presence of the definite article ny
# showing up on a definite AGENT. On the other hand, the suffix ana
# does not involve a root and so the notion of root does not appear 
# to be relevant to this discussion.

# Problem # 3: Keenan & Polinsky (1998:573) make the claim that' a certain number
# of two-syllable words with weak endings behave as weak roots...' However the problem
# is not at the level of a root, but rather at the level of the final syllable ending 
# a root/word in a relevant environment.

# First, consonant epenthesis is not related to the distinction between weak & strong
# roots since there are strong forms, which require Cons-epenthesis. See Keenan & Polinsky
# (1998:581): fono, tohy, tsinjo, for instance. These forms are not weak in the sense of
# 'weak root' proposed by the authors, i.e. ending with { ka, tra, na }. Yet they
# require consonant epenthesis:

  fono
  fonosina
# fono  Noun 'envelope', passive 1 imperative fonosy funu+s+i 'funu-epenthetic.s-suffix.i'
# where a final i is written y '(let-sth)-be-enveloped-by.you'; passive 1 indicative 
# fonosina 'stem.funus+ina 'stem.funus-suffix.ina''(that-which)-is-usually-enveloped' or 
# '(that-which)-is-being-enveloped';

  tohy
  tohizina
# tohy Noun 'continuation', passive 1 imperative tohizo tuhi+z+u 'tuhi-epenthetic.z-
# suffix.i''(let-sth)-be-continued-by.you'; passive 1 indicative tohizina 'stem.tuhiz+ina
# '(that-which)-is-usually-continued' or '(that-which)-is-being-continued';/

  tsinjo
  tsinjovina
# tsinjo /tsindzu/ Noun 'perception', passive 1 imperative tsinjovy tsindzu+v+i 'tsindzu-ep.
# v-suffix.i''(let-sth)-be-perceived/scrutinized-by.you'; passive 1 indicative tsinjovina
# stem.tsindzuv+ina '(that-which)-is-usually-perceived' or '(that-which)-is-being-perceived'.

# Indeed, with regard to Keenan & Polinsky's general claim (1998:581) that '...when the verb root
# is weak, the consonant of -na, -ka, -tra may mutate as follows...' The more explicit claim made
# in K & P (1998:573) hinges on an exceptional behaviour based on a finite set of cases. In reality,
# the general situation is as illustrated in the following examples involving words (not just roots) 
# of less than 3 syllables and whose { ka, tra, na } ending is not affected at all:

  Loakan'i Paoly ilay zavatra.
# Loaka-                n(a) i    Paoly ilay  zavatra.
# 'be-with-a-hole-in-it-by   D.sg Paul  DX.sg thing
# 'Paul accidentally made a hole in that thing' or 'Paul managed to make a hole in that thing'
# where loaka /loka/ '(be)-with-a-hole-in-it' is a root with a passive meaning and where the ka ending of
# the first word does not drop (nor does it mutate) simply because this word has less than three
# (3) syllables;
# as was the case earlier, because we are dealing here with epenthetic na --NOT suffix '(i)na' or
# '(a)na'-- the process does not trigger stress shift; so stress falls on 'lo' in 'lo(a)ka', and NOT
# of 'ka' as would be the case if this was a case of suffixation;
# in addition of course, the root 'loaka' describes an  an OUTCOME, i.e. a STATE, not an ACTIVITY.


  ritran'ny masoandro ny rano
# ritra+n          ny         masoandro ny         rano
# be-evaporated-by Det.unspec sun       Det.unspec water
# 'The (previous mention) water was evaporated by the sun';
# where ritra '(be)-dried-up' is a root with a passive meaning and where the tra ending of
# the first word does not drop (nor does it mutate) simply because this word has less than three
# (3) syllables;
# as was the case before, because we are dealing here with epenthetic na --NOT suffix '(i)na' or
# '(a)na'-- the process does not trigger stress shift; so stress falls on 'lo' in 'lo(a)ka', and NOT
# of 'ka' as would be the case if this was a case of suffixation;
# in addition of course, the root 'ritra' describes an OUTCOME, i.e. a STATE, not an ACTIVITY.



  sakan-dry Jeanne
# saka+n+d+ry Jeanne
# cat-of-d-D.pl Jeanne
# 'a cat(s) belonging to Jeanne & associates'
# The noun saka 'cat' only has two syllables and so { ka, tra, na } drop does not apply;

  lokan-dry Jeanne
# loka+n+d+ry Jeanne
# bet-of-d-D.pl Jeanne
# 'a bet made by Jeanne'
# The noun loka /luka/ 'prize, bet' only has two syllables and so { ka, tra, na } drop does not apply;

  ratran-dry Jeanne
# ratra+ n+ d+ry   Jeanne
# injure-of-d-D.pl Jeanne
# 'an injury/injuries sustained by Jeanne & associates'
#  The noun ratra 'injury' only has two syllables and so { ka, tra, na } drop does not apply;


  taonan-dry Jeanne
# taona+n+d+ry Jeanne
# ages-of-d-D.pl Jeanne
# 'ages of Jeanne & associates'
# The noun taona 'year' comprises a diphthong /au/, the word taona has less than three syllables
# and so { ka, tra, na } drop does not apply.

# In the last 6 cases the syllables { ka, tra, na} are not affected by a general rule regulating
# words comprising 3 or more syllables in a wide range of phenomena. Note the inserted morpheme
# n meaning 'by+AGENT/EXPERIENCER' when a root verb is concerned; it indicates a genitive case and
# when possession is involved, then the first noun describes an item possessed by somebody. This is
# a potential alternative analysis to the one presented earlier, i.e. suffix na; the only apparent 
# difficulty here is that positing 'n' instead of 'na' will require an epenthetic vowel 'a'; but 
# consult the document entitled 'Malagasy Epenthesis'; 
 

# Contrast the above with the following words, i.e. not just roots but other words as well
# which comprise three (3) or more syllables:

  tola-gaga
# /tulagaga/ '(be)-totally-flabbergasted';
# tolaka+gaga '(be)-embezzled, (be)-bent'+'(be)-surprised''(be)-totally-flabbergasted';
# where root tolaka /tulaka/ comprises three distinct syllables and where the final 
# syllable ka does drop indeed;

  fanta-daza
# '(be)-renowned, (be)-with-a-good-reputation';
# fantatra+laza '(be)-known'+'reputation''(be)-renowned, (be)-with-a-good-reputation';
# where root fantatra /fantatra/ has three separate syllables and where the final 
# syllable tra does drop before l > d and incorporation takes place;

  tazan-davitra
# tazana+lavitra 'perception'+'far-away''distant viewing'; where root tazana 'be-seen'
# /tazana/ comprises three distinct syllables and where the final vowel a of na drops;

  mamoa-pady
# mamoaka          fady i.e. Verb + Object
# m+a(n>m)+(v)oaka fady
# pres-prf.an-let-out taboo
# m+am+oaka
# 'which lets taboo out' or 'morally objectionable'; where the verb mamoaka pronounced
# as ma.mo.ka comprises three distinct syllables and where the final ka of the first word
# does drop when incorporation takes place;

  zava-dehibe
# zavatra lehibe, i.e. Noun + Adjective
# thing   big
# 'a big thing' or 'big things'; where the noun zavatra comprises three distinct syllables
# and where the final syllable tra of the first word /zavatra/ drops during incorporation; 
# note the mutation l > d affecting the initial consonant of the second word /lehibe/;

  hala-jaza
# halatra zaza i.e. RootV + Object
# theft   child(ren)
# 'child theft'; where the noun halatra /halatra/ comprises three distinct syllables and 
# where the final syllable tra of the first word drops during incorporation; note also 
# the mutation z > d+z written as j affecting the initial consonant of the second word
# zaza;

  manan-janaka
# manana zanaka i.e. Verb + Object
# m+anana progeny
# pres-have progeny
# "have children'; where the verb manana /manana/ comprises three different syllables 
# and where the final vowel a of the first word drops during incorporation;

  manam-bady i Jeanne
# manana vady i Jeanne i.e. Verb + Object
# m+anana   vady   i    Jeanne
# pres-have spouse D.sg Jeanne
# "Jeanne has a spouse' 'Jeanne is married'; where again manana has three syllables and
# where homorganic nasal assimilation ( n > m ) has affected the first word -- manan(a) >
# manam while the initial consonant of the second word vady has mutated ( v > b ) yielding
# bady;

# where j = [dz] and where { ka, tra, na --to a lesser extent--} seem to simply drop out. 
# So the relevant phenomenon has to do with those specific syllables as well as the number
# of syllables in any given word, i.e. three or more syllables. Note also that nasal n in 
# Malagasy does not constitute a syllable on its own.

# Problem # 4: Inserted 'n(a)''of/by-(s.one.sth)' or 'belonging-to-s.one' vs '(i)n(a)' derived 
# from PATIENT/THEME suffix ina and affixes no...ina, (...) denotes optionality. Words ending in 
# { ka, tra, na } comprising three (3) or more syllables are automatically subject to the syllable
# /vowel dropping rule described above; but other words require suffixation first.

# Cases of words not ending in { ka, tra, na } but where clitic 'na' 'of/by' is inserted to 
# indicate either AGENT/EXPERIENCER or POSSESSION:

  ranon'ny mpianatra
# rano+ na ny         mpianatra
# water-of Det.unspec students(s)
# 'The students' water'
# Initially the root word /ranu/ does not end up with { ka, tra, na }; however when the clitic na is
# added, the syllable/vowel drop rule will apply eliding the vowel a of clitic na in the environ-
# ment of the definite article 'ny'; the possessor phrase is definite;

  ranona mpianatra
# rano+ na mpianatra
# water-of students(s)
# 'A/some students' water'
# Note the clitic na, which shows up on the surface as the possessor phrase is not definite,
# thus lacking the relevant definite article 'ny';

  bokin'ny mpianatra
# buki+na ny         mpianatra
# book-of Det.unspec student(s)
# 'A/some book/books of the students'
# The root word /buki/ does not end up with { ka, tra, na }; however when the clitic na is added,
# the syllable/vowel drop rule will apply eliding the vowel a of clitic na in the environment of the
# definite article 'ny'; the possessor phrase is definite;

  bokina mpianatra
# boki+na mpianatra
# book-of student(s)
# 'A/some book/books of a/some students'
# Note the clitic na, which shows up on the surface as the possessor phrase is not definite,
# thus lacking the definite article 'ny';

  asan'ny mpianatra
# asa+n'  ny         mpianatra
# work-of Det.unspec student(s)
# 'work(s) of the students'
# The root word asa does not end up with { ka, tra, na }; however when the clitic na is added,
# the syllable/vowel drop rule will apply eliding the vowel a of clitic na in the environment of the
# definite article 'ny'; the possessor phrase is definite;

  asana mpianatra
# asa+na  mpianatra
# work-of student(s)
# A/some 'work(s) of a/some student(s)'
# Note the clitic na, which shows up on the surface as the possessor phrase is not definite,
# thus lacking the definite article 'ny';

  rononon'ny mpianatra
# rununu+n'ny         mpianatra
# milk-of  Det.unspec student(s)
# 'The students' milk'
# The root word /rununu/ does not end up with { ka, tra, na }; however when the clitic na is added,
# the syllable/vowel drop rule will apply eliding the vowel a of clitic na in the environment of the
# definite article 'ny'. The possessor phrase is definite;

  rononona mpianatra
# rununu+na mpianatra
# milk-of   student(s)
# 'A/some student(s)' milk'
# Note the clitic na, which shows up on the surface as the possessor phrase is not definite,
# thus lacking the definite article 'ny';

  vatomamin'ny mpianatra
# vatumami+n'ny        mpianatra
# sweets-of Det.unspec students
# 'The students' sweets'
# The word /vatumami/ does not end up with { ka, tra, na }; however when the clitic na is added,
# the syllable/vowel drop rule will apply eliding the vowel a of ctici na in the environment of the
# definite article 'ny'; the possessor phrase is definite;

  vatomamina mpianatra
# vatumami+na mpianatra
# sweets-of   students
# 'A/some student(s)' sweets'
# Note the clitic na, which shows up on the surface as the possessor phrase is not definite,
# thus lacking the definite article 'ny';

  fahasalaman'ny mpianatra
# fahasalamana ny         mpianatra
# health-of    Det.unspec students
# 'The students' health'
# The word fahasalamana ends up with { ka, tra, na }; therefore the syllable/vowel drop rule will apply
# eliding the vowel a of final syllable na in the environment of the definite article 'ny'. The possessor 
# phrase is definite;

  fahasalamana mpianatra
# fahasalamana mpianatra
# health-of    students
# 'A/some student/students' health'
# Note the final syllable na, which is not an added suffix as the word already ends up with 'na' (strictly
# speaking; however this is a derived noun, see the relevant document dealing with Nominals in Malagasy);

  hitan'ny mpianatra
# hita+na    ny         mpianatra
# be-seen-by Det.unspec students
# 'seen-by the students'
# Initially the root verb hita 'be-seen' does not end up with { ka, tra, na }; however the
# AGENT/EXPERIENCER will eventually be appended to the root 'hita' by epenthetic 'na'; therefore the  
# syllable/vowel drop rule will apply and the final vowel 'a' of epenthetic 'na' will be elided in the environment
# of the definite article 'ny' (i.e. since the AGENT/EXPERIENCER/.. is definite);
# note the root 'hita' describing an OUTCOME, i.e. an 'achieved' STATE.

  hitana mpianatra
# hita+na mpianatra
# be-seen-by students
# 'seen-by a/some student(s)'
# Note that the epenthetic 'na' shows up on the surface since the AGENT/EXPERIENCER is not definite;
# again, note the root 'hita' describing an OUTCOME, i.e. an 'achieved' STATE.

  voan'ny mpianatra
# vua-na ny mpianatra
# 'be-hit by the students'
# Initially the root verb voa 'be-hit' does not end up with { ka, tra, na }; however the
# AGENT/CAUSER will eventually be introduced by epenthetic 'na'; therefore the syllable/vowel drop 
# rule will apply and the final vowel 'a' of epenthetic 'na' will be elided in the environment of
# the definite article 'ny' (i.e. since the AGENT/EXPERIENCER/.. is definite);
# note the root 'voa' describing an OUTCOME, i.e. an 'achieved' STATE.

  voana mpianatra
# vua-    na mpianatra
# 'be-hit by a/some student(s)'
# Note the epenthetic 'na' since the AGENT/EXPERIENCER is not definite;
# note the root 'voa' describing an OUTCOME, i.e. an 'achieved' STATE.


# Contrast epenthetic 'na' 'of/by' vs 'ina' 'PATIENT/THEME-subject (alternative affixes
# 'no...ina') forces us to slightly modify as follows our analysis of a sentence found
# in Keenan (2000: 27-47) :

  Hitan'i Paoly ilay trano.
# Hita-na         i    Paoly ilay  trano.
# root.hita-ep.na D.sg Paul  DX.sg house(s)
# Lit. "The (previous mention) house was seen by Paul."
# "Paul saw the (previous mention) house"
# where the clitic 'na' appended to the root 'hita' drops its final vowel a in view of the fact
#  that the AGENT/EXPERIENCER is definite and comprises the D.sg 'i';
# as was the case before, because we are dealing here with epenthetic na --NOT suffix '(i)na' 
# of passive.1 circumfix 'no...ina' nor suffix '(a)na' of passive.1 circumfix 'no...ana'
# -- the process does not trigger stress shift; so stress falls on 'hi' in 'hita', and NOT
# of 'ta' as would be the case if this was a case of suffixation;
# in addition of course, the root 'hita' describes an OUTCOME, i.e. a punctual STATE and
# does NOT describe a DELIBERATE kind of ACTIVITY.

  Hitan'ny mpianatra ilay trano.
# Hita-na         ny         mpianatra   ilay  trano.
# root.hita-ep.na Det.unspec student(s)  DX.sg house(s)
# Lit. "The (previous mention) house was seen by the students."
# "The students see/saw the (previous mention) house"
# where the clitic 'na' drops its final vowel a in view of the fact that the AGENT/
# EXPERIENCER is definite and comprises the definite article 'ny';
# as was the case before, because we are dealing here with epenthetic na --NOT suffix '(i)na' 
# of passive.1 circumfix 'no...ina' nor suffix '(a)na' of passive.1 circumfix 'no...ana'
# -- the process does not trigger stress shift; so stress falls on 'hi' in 'hita', and NOT
# of 'ta' as would be the case if this was a case of suffixation;
# in addition of course, the root 'hita' describes an OUTCOME, i.e. a punctual
# STATE and does NOT describe a DELIBERATE kind of ACTIVITY.


  Hitana mpianatra ilay trano.
# Hita-na         mpianatra     ilay  trano.
# root.hita-ep.na a/student(s)  DX.sg house(s)
# Lit. "The (previous mention) house was seen by a/some student(s)."
# "A student/some students saw the (previous mention) house"
# where the epenthetic or clitic 'na' shows up as is on the surface;
# as was the case before, because we are dealing here with epenthetic na --NOT suffix '(i)na' or
# '(a)na'-- the process does not trigger stress shift; so stress falls on 'hi' in 'hita', and NOT
# of 'ta' as would be the case if this was a case of suffixation;
# in addition of course, the root 'hita' describes an unplanned OUTCOME, i.e. a 
# STATE and does NOT describe a DELIBERATE kind of ACTIVITY.


# Ungrammatical
  Nohitan'ny lehilahy ny trano.(0! 0 0 0)
# no-       hita-   ina        ny         lehilahy ny         trano
# past.pass-root.hita-suff.ina Det.unspec man/men  Det.unspec house(s)
# "The houses were seen by the men."
# This sequence is from Keenan (2000: 27-47) in Formal Issues in 
# Austronesian Linguistics, eds. Ileana Paul, Vivianne Philips and
# Lisa Travis. Dordrecht, the Netherlands: Kluwer Academic Press.
# No-          hita-ina     ny         lehilahy ny         trano.
# no.....in(a) hita         ny         lehilahy ny         trano
# [ + CONTROL] [ - CONTROL] Det.unspec man/men  Det.unspec house(s).
# Note the emergence of the passive voice suffix ina, which is part of the
# affixes 'no...ina''past.tense...suffix.ina'. However, this sentence simply
# does NOT exist in Malagasy! Indeed, it involves an internal contradiction:
# 'hita' would be a verbal stem, i.e. describing a DELIBERATE kind of ACTIVITY 
# and NOT a root, i.e. describing an ACCIDENTAL, unplanned kind of EVENT; 
# the affixes no...ina indicate explicitly that the phrase 'ny lehilahy' 
# 'the men' encodes an AGENT, who is in full control of the event, whereas the 
# PERCEPTION root verb hita requires an EXPERIENCER, i.e. in this case 'ny lehilahy',
# who is definitely not in control of the event at all.
# Indeed the combination 'root.hita + passive suffix ina' is illegal.


## Appendix: Apparent exceptions

   tahotra alika
   tahotr'alika

   ny tahotra alika
## ny         tahotra alika
## Det.unspec fear-of dogs
##             V      Object
## the        fear-of dog
## 'the fear s.one has of dogs';
## where the final vowel a of /tahutra/ is not elided as it corresponds
## to a root verb taking a direct object complement 'alika';

   ny tahotr'alika
## ny         tahotra-O alika
## Det.unspec fear-of   dogs
##            N         Possessive
## the        fear-of   dogs
## 'the fear dogs have'
## where O represents the elided final vowel a of syllable tra of /tahutra/
## encoding POSSESSOR in the genitive case phrase 'alika';

   hevitra adala
   hevitr'adala
## hevitra adala
## hevitra    =        adala
## an idea(s) is/are   foolish
## '(a) foolish idea(s)';
## where 'hevitra' is a headnoun in a restrictive relative clause comprising
## the adjectival phrase 'adala'; therefore 'hevitra' does not lose its final
## vowel 'a';

## hevitr'adala
## hevitr-O adala
## idea(s)-of a fool/fools
## 'idea(s) of a fool/fools'
## where O represents the elided final vowel a of syllable tra of hevitra
## encoding POSSESSOR in the genitive case phrase 'adala';

   olona alika
   olon'alika
## olona alika
## olona        =     alika
## a person(s) is/are dog(s)
## '(a) terrible person(s)/people';
## where 'olona' is a headnoun in a restrictive relative clause comprising
## the noun phrase 'alika'; therefore 'hevitra' does not lose its final
## vowel 'a';


## olon'alika
## olon-O         alika
## a person(s)-of dogs
## 'a person(s) devoted to dogs'
## where O represents the elided final vowel a of syllable na of 'olona'
## encoding POSSESSOR in the genitive case phrase 'alika';