• Keine Ergebnisse gefunden

6.2  Comparing  the  two  morphologies    

 

In  order  to  be  able  to  compare  the  two  systems  and  arrive  at  a  unified  account  of  the   two   verb   formation   strategies   in   Maltese,   it   was   important   to   study   root-­‐and-­‐pattern   morphology  first.  Since  templatic  verbs  constitute  a  closed  class  of  verbs,  it  was  possi-­‐

ble   to   compile   an   exhaustive   database   of   all   consonantal   roots   and   the   patterns   they   occur  in,  and  come  up  with  a  classification  based  on  the  lexical  semantics  and  morpho-­‐

logical  marking  of  the  verbs.  In  what  follows  I  illustrate  the  main  findings  of  this  exami-­‐

nation,  as  discussed  in  Ch.  3  and  Ch.  4,  and  propose  a  way  of  extending  this  analysis  to   concatenative  verbs,  as  shown  in  Ch.  5.      

 

6.2.1  Analyzing  templatic  verbs  in  Maltese  

 

In  this  study,  templatic  verbs  are  understood  as  the  end  result  of  the  interdigitation  of  a   category-­‐neutral,  consonantal  root  with  a  binyan,  a  verbal  pattern  which  has  a  specific   syllabic   structure,   and   slots   for   vowels   and   root   consonants   to   fit   in.   The   binyan   not   only  makes  the  root  consonants  pronounceable,  but  it  also  determines  the  lexical  cate-­‐

gory:  it  is  only  when  roots  are  inserted  in  a  particular  pattern  that  an  actual  verb,  noun   or  adjective  is  formed.  This  implies  that  all  templatic  verbs  must  be  in  the  form  of  a  bin-­‐

yan.  

I  argued  that  the  binyanim  belong  to  the  domain  of  derivation,  not  to  the  marking  of   inflectional  categories  such  as  person,  number,  tense-­‐aspect,  and  gender.  They  display   two  main  characteristics  of  word  formation.  First,  they  are  not  completely  productive:  

the  binyan  system  is  full  of  gaps  (cf.  Sect.  4.1).  Second,  word  formation  creates  words   with  lexicalized  meanings,  and  when  cast  in  different  verbal  patterns,  roots  may  take  on   specialized,  non-­‐transparent  interpretations.  See  the  discussion  in  Sect.  2.2.2  for  further   arguments   that   point   to   the   conclusion   that   binyan   morphology   is   essentially   deriva-­‐

tional.    

A   root   can   in   principle   combine   with   several   binyanim   to   form   different   verbs.  

However,  no  tri-­‐consonantal  root  combines  with  all  nine  patterns.  Over  two  thirds  of  all   tri-­‐consonantal  roots  appear  in  only  one  or  two  patterns.  The  mean  number  of  patterns   per  root  is  around  2,  which  means  that  many  binyan  slots  are  left  empty.    

The  combination  of  roots  and  binyanim  is  subject  to  a  number  of  phonological  con-­‐

straints.  For  instance,  roots   starting   with   /k,  t,  tʃ,   dʒ/   etc.  cannot  appear  in  binyan  X,  

which  is  characterized  by  the  prefix  st-­,  as  they  would  create  onset  consonant  clusters   that  are  not  permissible  in  Maltese,  *kst,  *tst  etc.  In  addition,  an  examination  of  the  pos-­‐

sible  relationships  between  patterns  and  root  types,  strong  (regular,  reduplicative)  and   weak  (initial,  medial,  final),  revealed  that  binyan  III  and  VI  disfavor  reduplicative  roots,   and  that  weak-­‐final  roots  are  unlikely  to  be  inserted  in  binyan  II,  V  and  IX.  See  Sect.  4.1   for  further  (dis)preferences.    

An  analysis  of  the  distribution  of  roots  across  patterns  showed  that  only  four  pat-­‐

terns  are  really  productive  in  Maltese,  namely  binyan  I,  II,  V  and  VII.  Moreover,  I  argued   in   Sect.   4.3   that,   in   terms   of   morphological   productivity,   morphosyntactic   similarity,   and   complementarity   in   distribution,   templatic   verbs   in   Maltese   make   up   a   four-­‐way   system:   III,   VI,   VIII,   IX   are,   as   it   were,   the   shadow   binyanim   of   II,   V,   VII   and   I   respec-­‐

tively.  In  general,  this  quasi-­‐allomorphy  of  patterns  is  phonologically  conditioned.  The   remaining   pattern,   binyan   X,   is   highly   underrepresented,   as   it   combines   with   only   24   roots,  making  up  less  than  one  percent  of  the  total  number  of  tri-­‐consonantal  roots  in   Maltese.    

 In  Ch.  3,  I  showed  that  the  binyan  system  is  inherently  regular  and  irregular  from   three   different   angles.   First,   when   a   root   is   embedded   in   more   than   one   pattern,   this   very   often   correlates   with   a   regular   argument   structure   alternation   (e.g.,   √ktb,  kiteb  

‘write’   –  inkiteb  ‘be   written’).   However,   two   or   more   verbs   containing   the   same   root   consonants  may  have  two  semantically  distant  interpretations  (e.g.,  √xrb,  xorob  ‘drink’  

–  xarrab  ‘wet’)  or  may  be  synonymous  (e.g.,  √slf,  silef  ‘lend’  –  sellef  ‘lend’).  The  qualita-­‐

tive  analysis  carried  out  in  Ch.  4  brought  to  light  the  fact  that  61%  of  all  combinations  of   tri-­‐consonantal   roots   with   patterns   mark   an   argument   alternation,   18%   are   synony-­‐

mous  verbs,  9%  correspond  to  multiple,  semantically  distant  interpretations,  and  12%  

are  cases  of  roots  that  derive  singleton  verbs.  Second,  from  a  formal  point  of  view,  tem-­‐

platic   verbs   sharing   a   single   consonantal   root   may   either   have   the   same   vowel   se-­‐

quence,  as  in  kiteb  and  inkiteb,  or  display  vocalic  variation,  as  in  xorob  and  xarrab.  Fi-­‐

nally,  an  examination  of  the  distribution  of  verbs  in  the  lexicon  reveals  that  verbs  de-­‐

rived  from  the  same  root  may  be  fully  productive  (e.g.,  passives  in  binyan  V,  tqassam  ‘be   distributed’,  always  have  an   active   counterpart,   typically  in  binyan  II,  qassam  ‘distrib-­‐

ute’)  or  exhibit  gaps  in  the  system  (e.g.,  inchoatives  in  IX,  smar  ‘get  tanned’,  do  not  al-­‐

ways  have  a  corresponding  causative).    

These  conflicting  aspects  of  the  binyan  system,  I  argue,  can  be  explained  if  we:    

 

(a) assume   that   roots   are   category-­‐neutral   and   semantically   underspecified   (it   is   only  when  they  are  inserted  in  a  word  pattern  that  their  lexical  category  is  de-­‐

termined  and  that  they  acquire  a  specific  interpretation);  

(b) recognize  that  there  is  no  one-­‐to-­‐one  relation  between  the  semantic  and  syntac-­‐

tic  properties  of  verbs  and  their  morphological  form  (the  same  binyan  may  host   verbs  of  different  semantic  types  (e.g.,  a  verb  in  binyan  I  may  be  active,  causa-­‐

tive,  inchoative),  and  verbs  of  the  same  semantic  type  (e.g.,  active)  may  appear   in  different  patterns  (e.g.,  I,  II,  III,  VIII));  

(c) assume,  in  line   with  Marantz  (1997,  2001,  2007),   Arad  (2003a,  2003b,  2005),   and  others,  that:    

(i) words   can   be   formed   either   by   merging   a   root   with   a   category-­‐bearing   head  or  by  combining  an  existing  word  with  a  new  syntactic  head;  

(ii) word   formation   from   roots   gives   rise   to   irregular,   non-­‐productive   proc-­‐

esses,  whereas   word  formation  from  previously  formed  words  results  in   regular,  productive  processes.  

 

From  the  qualitative  analysis  in  Ch.  4,  we  concluded  that  one  main  source  of  irregu-­‐

larity  in  the  binyan  system  is  syncretism.  There  is  no  one-­‐to-­‐one  mapping  between  the   morphological   form   of   verbs   and   their   semantic-­‐syntactic   structure.   This   can   be   ob-­‐

served  in  two  ways.  First,  the  same  combination  of  binyanim  may  mark  more  than  one   semantic-­‐syntactic  category.  A  root  in,  say,  binyan   I  and  II  may  represent  a  causative-­‐

inchoative   alternation   (e.g.,   √għlj,  għola  ‘rise’,  għolla   ‘raise’),   a   pair   of   synonyms   (e.g.,  

√bdl,  bidel,  biddel   ‘change’),   multiple   interpretations   (e.g.,   √tbgħ  tebagħ  ‘print’,  tebba’  

‘stain’),   and   others.   Second,   the   same   semantic-­‐syntactic   feature   may   be   marked   by   more   than   one   combination   of   patterns.   For   instance,   the   transitive-­‐reflexive   alterna-­‐

tion  may  be  realized  in  II  –  V,  I  –  VII,  III  –  VI,  etc.  (cf.  Sect.  4.2).    

We   also   noted   that   regularity   in   the   system   stems   from   two   main   sources.   First,   there   is   a   strong   tendency   for   complementary   distribution   among   the   four   categories   roots  fall  into  (argument  alternations,  multiple  interpretations,  synonyms,  singletons).  

For  instance,  when  a  root  is  inserted  in  VII  and  VIII,  this  almost  always  correlates  with   two  synonyms  and  hardly  ever  with  an  argument  alternation.  Second,  argument  alter-­‐

nations  are  strikingly  regular  with  respect  to  directionality.  While  rejecting  the  deter-­‐

ministic   approach   of   traditional   grammars   of   Maltese   that   find   neat   associations   be-­‐

tween   patterns  and  semantic  roles  such  as  causative  and  passive,   we  can  observe  the   significant  contribution  of  the  binyanim  in  terms  of  transitivity.  The  system  is,  to  a  cer-­‐

tain   extent,   morphologically   transparent   as   it   imposes   transitivity   constraints   on   the   binyanim.  From  the  fact  that  some  of  them  are  inherently  or  predominantly  intransitive   (V,  VII,  IX)  and  others  are  predominantly  transitive  (II,  III),  it  follows  that  argument  al-­‐

ternations   are   very   often   unidirectional,   with   passives,   reflexives   and   non-­‐causatives   appearing   in   the   patterns   that   typically   host   intransitive   verbs   and   their   active,   non-­‐

reflexive   and   causative   counterparts   appearing   in   the   predominantly   transitive   pat-­‐

terns.      

Although  argument  alternations  display   a   great   degree  of  regularity,   we  observed   that  two  alternations,  the  causative-­‐inchoative  and  causative-­‐noncausative,  are  irregu-­‐

lar  in   several  respects.   On   the  basis  of  the   structural  distinction  between  root   deriva-­‐

tion  and   word  derivation,  I  argue   that  passives  and  reflexives   are  verb   derived,  while   causatives  and  non-­‐causatives  are  root  derived.  In  Sect.  4.3,  I  presented  morphonologi-­‐

cal   and   semantic   evidence   for   this   claim.   Let   us   go   over   the   key   differences   between   root-­‐derived  and  verb-­‐derived  verbs.  

To   begin   with,   passives   and   reflexives   have   the   same   formal   properties   (e.g.,   se-­‐

quence   of   vowels,   medial   gemination)   as   the   transitive   verbs   they   are   derived   from.  

Causatives   and   inchoatives,   by   contrast,   may   display   vocalic   variation.   In   an   overall   analysis,  it  turns  out  that  vocalic  variation  among  verbs  containing  the  same  root  is  al-­‐

ways  associated  with  root  derivations,  i.e.  multiple  interpretations  (e.g.,  fired  ‘separate’  

–  farrad  ‘make  unpaired’),  synonyms  (e.g.,  silef  –  sellef  ‘lend’),  causative-­‐inchoative  (e.g.,   kiber  ‘grow  v.i’  –  kabbar  ‘grow  v.t’),  causative-­‐noncausative  (e.g.,  żifen  ‘dance’  –  żeffen  

‘make  one  dance’).      

Related  to  this  is  the  observation  that  passives  and  reflexives  are   morphologically   more  complex  than  the  transitive  verbs  they  alternate  with.  Causatives  and  inchoatives,   on  the  other  hand,  may  be  both  morphologically  marked.  Besides,  there  are  cases  where   one  labile  verb  marks  both  the  causative  and  inchoative  verb.      

The  third  piece  of  evidence  is  in  terms  of  productivity.  While  passives  and  reflexives   always   have   a   transitive   counterpart,   it   is   not   difficult   to   find   causatives   that   stand   alone,   without   a   corresponding   inchoative   (e.g.,  sann  ‘sharpen’),   and   inchoatives   that   lack  a  causative  counterpart  (e.g.,  batta  ‘abate’).    

Finally,  from  a  semantic  point  of  view,  it  is  very  rare  for  passives  and  reflexives  to   take  on  an  interpretation  that  is  not  available  in  their  active  and  non-­‐reflexive  counter-­‐

parts,   as  discussed  in   Ch.  3.  Causatives,  by  contrast,  acquire   specialized  and  idiomatic   meanings   that   do   not   exist   with   their   corresponding   inchoatives.   The   same   holds   for  

inchoatives,  which  may  be  assigned  idiosyncratic  interpretations  that  are  not  present  in   the  causative  verbs  they  alternate  with.        

 

6.2.2  Extending  the  analysis  to  concatenative  verbs  

 

The  causative-­‐inchoative  alternation  was  chosen  as   the  point  of   departure  for  a  com-­‐

parative  treatment  of  templatic  and  concatenative  verbs  in  Maltese  because  it  is  a  basic   argument  alternation  which  makes  up  around  one  third  of  all  verb  alternations  in  tem-­‐

platic  verbs,  and  because  it  brings  into  play  various  morphological,  syntactic  and  lexical   semantic  aspects  of  the  verbs.    

An  examination  of  the  formal  encoding  of  the  alternation  in  Maltese  revealed  that   there  is  variation  in  the  marking  of  the  alternation,  which  goes  hand  in  hand  with  the   distinction  between  the  two  verb  formation  strategies.  Templatic  verbs  typically  mark   the   alternation   by   means   of   two   different   morphologically   related   verbs,   with   the   in-­‐

choative  verb  being  generally  morphologically  more  complex  than  the  causative  one.  By   contrast,   concatenative   verbs   do   not   express   the   alternation   overtly:   labile   verbs,   i.e.  

ambitransitive   verbal   forms,   are   used   to   mark   both   the   causative   and   the   inchoative   alternant,   as   in  it-­tifel   iċċarġja   l-­mobile  ‘the   boy   recharged   the   mobile’   and  il-­mobile   iċċarġja  ‘the   mobile  recharged’.  In  a  few  cases,  the  reflexive   marker  ruħ-­‐  plus   a  direct   object  clitic  is  used  to  mark  the  inchoative  alternant,  as  in  il-­problema  intensifikat  ruħha  

‘the  problem  has  intensified’.  

In  Ch.   4  and   Ch.  5,  I  argued  that  the   morphological  irregularities  of   the  causative-­‐

inchoative  alternation  are  best  captured  by  a  root-­‐based  analysis.  In  spite  of  the  differ-­‐

ences  in  the  formal  encoding  of  the  alternation  by  templatic  and  concatenative  verbs,  a   unified  analysis  of  the  causative-­‐inchoative  alternation  in  Maltese  can  be  provided  if  we   assume  that  all  verbs  taking  part  in  the  alternation  are  cases  of  root  derivation.    

From  a  sentence  creation  task  and  a  corpus  study  on  a  set  of  around  40  labile  verbs   in   Maltese,   we   concluded   that   labile   verbs   do   not   all   behave   in   the   same   way.   Even   though  they  fail  to  show  any  morphological  mark  that  would  distinguish  the  causative   or  inchoative  alternant  as  formally  more  marked  than  the  other,  some  of  them  are  more   likely  to  be  used  in  transitive  constructions,  while  others  tend  to  occur  in  intransitive   frames.    

This   difference   among   alternating   labile   verbs   was   explained   in   lexical   semantic   terms.  Assuming  that  the  morphosyntactic  structure  of  verbs  is  directly  related  to  their   lexical  semantic  structure,  I  argue  that  the  variation  found  in  labile  verbs  in  Maltese  is  

sensitive  to  the  kind  of  event  named  by  the  verbs  in  question,  whether  it  is  internally  or   externally  caused.  Externally  caused  events,  like  iċċarġja  ‘recharge’,  are  conceptualized   as  coming  about  due  to  a  force  external  to  the  entity  undergoing  the  change  of  state.  By   contrast,  in  internally  caused  events,  like  sparixxa  ‘vanish’,  the  means  for  bringing  about   the   change   of   state   is   conceptualized   as   residing   in   the   entity   undergoing   the   change   itself.  It  follows  from  this  that  externally  caused  labile  verbs  are  used  more  frequently   in  transitive  constructions,  whereas  internally  caused  labile  verbs  occur  more  often  in   intransitive  frames.