• Keine Ergebnisse gefunden

4.3  A(n  ir)regular  binyan  system  

 

The  aim  of  this  section  is   to   summarize   the   most  important  conclusions  drawn  in  the   previous  and   the   current  chapter,  which  tried  to  (i)  identify  the  regular  and  irregular   aspects  of  the  binyan  system;  (ii)  describe  the  extent  to  which  the  system  is  transparent   and  opaque;  and  (iii)  account  for  the  dual  nature  of  templatic  verbs  by  means  of  a  prin-­‐

cipled  distinction  in  the  process  of  word  formation.  Let  us  give  a  brief  outline  of  the  as-­‐

sumptions  and  results  (Sect.  4.3.1),  in  order  to  bring  together  the  key  contributions  of   this  analysis  of  templatic  verbs  in  Maltese  (Sect.  4.3.2).    

 

4.3.1  Main  findings  and  assumptions  

 

Semantic  (un)relatedness    

The  point  of  departure  for  the  preceding  discussion  was  the  puzzle  that  the  semantics   of  words  containing  the  same  root  consonants  may  be  close  and  relatively  predictable   (e.g.,  xorob  ‘I,   drink’,  xorb  ‘drinking’),   on   the   one   hand,   and   distant   and   idiosyncratic   (e.g.,  xarrab  ‘II,  wet’,  inxtorob  ‘VII,  shrink’),  on  the  other.    

 

Underspecified  roots  

In  order  to  resolve  the  above  puzzle,  I  resorted  to  (i)  one  tenet  of  Distributed  Morphol-­‐

ogy,   the   so-­‐called   Root   Hypothesis;   that   is,   the   idea   that   nouns,   verbs   and   adjectives,   regardless   of   their   morphological   complexity,   may   be   decomposed   into   roots,   atomic  

lexical  elements;  and  to  (ii)  the  assumption  that  roots  are  underspecified  on  three  lev-­‐

els.  First,  roots  are  category-­‐neutral.  It  is  only   when  they   merge  with  a  word-­‐creating   head  that  an  actual  verb,  noun  or  adjective  is  formed.  Second,  roots  are  incomplete  from   a  phonological  point  of  view:  √ktb,  for  instance,  is  unpronounceable  on  its  own.  Third,   the  semantic  content  of  roots  is  not  fully  definable.  Just  as  √ktb  becomes  a  continuous,   pronounceable  string  when  inserted  in  a  verbal  or  nominal  pattern,  roots  acquire  spe-­‐

cific  interpretations  in  the  environment  of  different  patterns.    

 

Contribution  of  patterns  

Contrary  to  the  deterministic  approach  taken  by  traditional  grammars  of  Maltese,  there   is  no  exclusive  mapping  between  the  semantic  and  syntactic  property  of  verbs  and  their   morphological  realization.   Rather,  (i)   the   same  binyan   may  host  verbs  of  different   se-­‐

mantic   types,   e.g.,   a   verb   in   binyan   VII   may   be   passive,   reflexive,   inchoative;9   and   (ii)   verbs  of  the  same  semantic  type  (e.g.,  reflexive)  may  appear  in  different  binyanim  (e.g.,   V,  VI,  VII,  VIII,  X).  A  significant  contribution  of  the  binyanim  to  morphological  transpar-­‐

ency  is  instead  in  terms  of  transitivity.  Some  of  them  (VI,  VII,  IX)  are  inherently  intransi-­‐

tive,  which  means  that  verbs  that  are  typically  transitive,  i.e.  causative,  active  and  non-­‐

reflexive   verbs,   will   not   appear   in   these   non-­‐transitive   binyanim.   The   same   holds   for   binyan  V,  which  is  predominantly  intransitive.  Other  verbal  patterns  are  almost  always   transitive  (II,  III)  and  are  therefore  incongruent  with  inchoatives,  passives  and  reflex-­‐

ives.  The  valency  of  the  remaining  three  patterns,  I,  VIII,  X,  is,  however,  inconsistent.    

 

Distribution  of  roots  

The  mean  number  of  patterns  per  root  is  around  2,  which  means  that  the  binyan  system   is  full  of  gaps.  In  fact,  on  examining  the  distribution  of  roots  across  patterns,  it  turns  out   that   over   two   thirds   of   tri-­‐consonantal   roots   combine   with   only   one   or   two   patterns,   leaving   several   binyan   slots   empty.   The   rest   typically   occur   in   three   (17%)   or   four   (11%)  different  patterns.  While  no  root  appears  in  all  nine  binyanim,  there  are  very  few   roots  (less  than  5%)  that  are  inserted  in  five,  six,  or  seven  patterns.      

 

Combinatorial  constraints  

The  combination  of  roots  and  patterns  is  subject  to  a  few  phonological  constraints.  Bin-­‐

yan  VIII,  which  involves  a  -­t-­  infixed  between  the  first  C-­‐  and  V-­‐slot,  does  not  combine    

9  The  sole  exception  is  binyan  IX,  which  is  associated  almost  exclusively  with  inchoatives.  

with  roots  whose  first  consonant  is  a  stop  /b,  p,  t,  d,  k,  g,  ʔ/,  an  affricate  /tʃ,  dʒ,  ts/,  a   glide  /j,  w/,  or  silent  għ  and  h.  By  contrast,  it  hosts  roots  with  liquids,  nasals,  fricatives,   and  sibilants  as  first  radical.  For  phonotactic  reasons,  certain  roots  are  not  inserted  in   binyan   X.   Being   characterized   by   a   prefix  st-­,   it   does   not   combine   with   roots   starting   with  /t,  tʃ,  d,  dʒ,  ts/  etc.,  as  they  would  create  onset  clusters  that  are  not  permissible  in   Maltese.  Silent-­‐medial  roots  tend  not  appear  in  binyan  II  and  V,  as  the  virtual  phonemes   għ  and  h  cannot   be   geminated.   In   addition,   there   are   some   dispreferences   for   root   types.  For  instance,  reduplicative  roots  tend  not  to  be  embedded  in  binyan  III  and   VI;  

weak-­‐initial  roots  do  not  appear  in  VIII  (see  restriction  above  on  glides  as  first  radical),   and  they  rarely  appear  in  I  and  VII;  weak-­‐medial  roots  hardly  ever  occur  in  III,  VI  and   VIII;  and  weak-­‐final  roots  do  not  usually  combine  with  II,  V  and  IX.  

 

Correspondences  among  patterns    

Subject   to   such  combinatorial   constraints,   tri-­‐consonantal   roots   can   in   principle   com-­‐

bine  with  any  of  the  nine  binyanim  to  create  different  verbal  lexemes.  However,  an  ex-­‐

amination  of  the  distribution  of  roots  across  patterns  revealed  that  only  four  binyanim   are  really  productive,  I,  II,  V  and  VII.  The  other  four  are  underrepresented,  especially  X.  

Several  grammars  of  Maltese  observe  four  similarities  between  the  productive  and   the   non-­‐productive   patterns.   First,   on   the   equivalence   of   binyan   II   and   III,   Sutcliffe   (1936:  84)  remarks,  “[t]he  third  form  […  ]  is  practically  an  extension  of  the  second,  and   has   the   same   meanings”.   Similar   observations   have   been   made   by   Borg   (1981,   1988)   and  Cachia  (1994:  207),  among  others.    

Second,  a  parallel  equivalence  obtains  between  binyan  V  and  VI,  which,  as  we  have   seen  above,  regularly  mark  the  passives,  reflexives  and  inchoatives  of  the  correspond-­‐

ing  transitive  verbs  in  II  and  III  respectively.10    

Third,   the   qualitative   analysis   in   Sect.   4.2   has   also   brought   to   light   the   fact   that   17%  of  the  absolute  and  contextual  synonyms  are  marked  by  VII  and  VIII.  That  several   verbs  in  VII  and  VIII  have  identical  meanings  was  already  observed  by  Sutcliffe  (1936:  

97),  “[s]ome  verbs  are  found  both  in  the  seventh  and  in  the  eighth  form  without  change   of  meaning.”    

Fourth,  Borg  (1981:  90)  identifies  a  parallelism  between  intransitive  verbs  in  bin-­‐

yan   I   and   verbs   in   IX:   “Semantically,   first   form   verbs   such   as   ‘kiber’   “he   grew”   and    

10  Borg  (1981:  Ch.3)  notes  a  semantic  equivalence  of  V,  VI  and  VII,  which  is  confirmed  by  the  results  in  Sect.  

4.2.    

‘għolob’  “he  became  lean”  are  indistinguishable  from   ninth  form  verbs   such   as  ‘ċkien’  

“he  grew  small”  and  ‘ħxien’  “he  grew  fat”.”    

The  semantic  and  syntactic  equivalence  between  the  productive  I,  II,  V,  VII  and  the   respective  non-­‐productive  IX,  III,  VI,  VIII  is  mirrored  by  morphological  similarity:    

 

• Both   binyan   V   (tC1vC2C2vC3)   and   VI   (tC1vvC2vC3)   involve   a   prefix  t-­,   which   under-­‐

goes  complete  assimilation  when  the  first  radical  of  the  root  is  a  coronal.    

• Binyan  I  (C1vC2vC3)  and  IX  (C1C2vvC3),  by  contrast,  are  prefixless.  In  addition,  some   inflectional  forms  of  I  and  IX  are  identical,  namely   the  first  and   second  persons  of   the  perfective.  Compare,  for  instance,  kiteb  ‘I,  write’,  1SG  ktib-­t  and  1PL  ktib-­na,  with   swied  ‘IX,  get  dark/tanned’,  1SG  swid-­t  and  1PL  swid-­na.    

• Some  grammars,  such  as  Cachia  (1994:  207),  claim  that  the  only  difference  between   II   (C1vC2C2vC3)   and   III   (C1vvC2vC3)   is   that   the   latter   usually   combines   with   roots   whose  second  radical  is  an  ungeminatable  silent  għ  or  h,  though  they  also  note  that   some  binyan  III  verbs  are  derived  from  roots  with  a  /r/  or  /f/  as  their  middle  radi-­‐

cal,  which  can  be  geminated.    

• Finally,   consider   the   formal   equivalence   of   binyan   VII   (ntC1vC2vC3)   and   VIII   (C1tvC2vC3),  especially  when  they  combine  with  nasal-­‐  and  liquid-­‐initial  roots,  such   as  √nfħ,  creating  intnefaħ  and  intefaħ,  which,  in  addition,  happen  to  be  synonymous,  

‘inflate,  v.i’.  Besides,  recall  that  metathesis  takes  place  with  the  prefixal  t-­  of  binyan   VII   and   roots   that   have   a   sibilant   for   first   radical,   yielding   forms   such   as  inxtorob   (for   *int-­xorob)   ‘shrink’.   At   first   glance,   these   verb   forms   appear   to   be   hybrids   of   binyan  VII  (prefix  n-­)  and  VIII  (infix  -­t-­),  and,  in  fact,  in  some  dictionaries  they  are   listed  as  VII+VIII  verbs  (cf.  Aquilina  1987-­‐1990).    

 

These   correspondences   suggests   that   the   four   pairs   of   binyanim   have   a   quasi-­‐

allomorphic  status.  They  are  virtually  in  complementary  distribution.  This  claim  is  sup-­‐

ported  by  the  analysis  of  the  co-­‐occurrence  of  patterns  on  the  same  root  in  Sect.  4.1.3,   which  revealed  that  less  than  one  percent  of  all  tri-­‐consonantal  roots  appear  in  both  I  –   IX  (0.98%)  and  II  –  III  (0.47%).11  An  examination  of  Table  4.13  shows  that  1.4%  of  tri-­‐

consonantal  roots  appear  in  V  and  VI,  and  that  2.09%  co-­‐occur  in  VII  and  VIII.  Note  that   when  two  verbs  in  V  and  VI  share  the  same  root,  the  latter  verb  form  is  very  often  dated    

11  A  root  appearing  in  I  and  IX  may  correspond  to  an  instance  of  multiple  interpretation  (e.g.,  √sbħ,  sebaħ  

‘dawn’,  sbieħ  ‘become   beautiful’)   or,   more   likely,   to   two   synonymous   verbs,   e.g.,   √twl,  tal,  twal   ‘become   long’,  √smn,  simen,  smien  ‘get  fat’.  One  tends  to  be  more  frequent  or  current  (twal,  simen)  than  the  other.  

When  a  root  appears  in  II  and  III,  the  verbs  created  are  synonymous,  e.g.,  √wld,  welled,  wieled  ‘give  birth’,  

√ġld,  ġelled,  ġieled  ‘provoke  a  fight’.  The  verb  in  III  is  usually  dated.      

and   synonymous   with   the   binyan   V   verb,   e.g.,  iġġieneb  ‘move   aside’,  tbiedel  ‘be   ex-­‐

changed’,  tħaseb  ‘worry’,  tħalat  ‘be  mixed’.  The  number  of  roots  co-­‐occurring  in  VII  and   VIII  is  relatively  ‘higher’  because  of  the  propensity  roots  have  for  creating  synonyms  in   these  two  patterns.    

In   terms   of   productivity,   morphosyntactic   similarity,   and   distribution   in   the   lexi-­‐

con,  we  can  therefore  conclude  that  Maltese  templatic  verbs  make  up  a   four-­‐way  sys-­‐

tem   consisting   of   binyan   I,   II,   V   and   VII,   which   in   a   few   cases   alternate   with   their  

‘shadow’  binyanim  IX,  III,  VI  and  VIII  respectively.  Binyan  selection  is  often  phonologi-­‐

cally  conditioned.   For  instance,   silent-­‐medial  roots  almost   always   appear  in  III  rather   than  II,  because  they  cannot  be  geminated,  and  weak-­‐medial  roots  are  more  likely  to  be   inserted  in  V  than  VI  (cf.  the  (dis)preferences  of  root  types  by  binyanim  Sect.  4.1.2).  Fi-­‐

nally,   binyan   X   is   highly   underrepresented,   hosting   only   24   roots   (0.74%),   seven   of   which  have  practically  fallen  out  of  use  (cf.  Sect  2.2.2).  

 

Vowel  sequences  of  binyanim  

In  passing,  it  was  noted  that  the  vowel  sequence  of  some  verbs  derived  from  one  root  is   the   same   throughout,   e.g.,  ħareġ  ‘I,   take   out’,  ħarreġ  ‘II,   train’,  tħarreġ  ‘V,   train,   v.i’,  

inħareġ   ‘VII,   be   taken   out’,  stħarreġ  ‘X,   investigate’,   while   the   vowels   of   other   verbs   change   from   one   pattern   to   the   other,  kiber  ‘grow,   v.i’,  kabbar  ‘grow,   v.t’,  tkabbar  ‘be   grown’.  

 

Four  categories  

A  qualitative  analysis  of  the  interactions  between  roots  and  patterns  has  shown  that  tri-­‐

consonantal  roots  fall  into  four  major  categories.  Argument  alternations,  such  as  active-­‐

passive  and  causative-­‐inchoative,  make  up  the  largest  group.  The  other  three  categories   consist  of  roots  that  derive  singleton  verbs,  roots  that  acquire  multiple  interpretations,   and  roots  that  form  synonymous  verbs.    

 

4.3.2  Verb  formation  and  the  four  categories  

 

To  conclude  this  section  on  the  binyan  system  in  Maltese,  I  integrate  the  discussion  on   the   morphological   realization   of   the   four   categories   roots   fall   into   with   the   structural   distinction  between  root  derivation  and  word  derivation,  which  was  introduced  in  Ch.  3  

to  explain  why  the  system  has  both  transparent  and  opaque  elements.  Let  us   go  over   the  most  important  observations  concerning  the  distinction  in  word  formation.        

In   line   with   Marantz   and   Arad’s   model   of   word   formation,   couched   within   the   framework  of  Distributed  Morphology,  I  argued  in  Ch.  3  that   words  can  be  formed  in   syntax  in  two  ways:  (i)  by  merging  a  category-­‐free  root  with  a  syntactic  head,  turning   the   root   in   a   noun,   verb   or   adjective,   or   (ii)   by   combining   a   previously   formed   word   with  a  new  syntactic  head.  From  this  distinction  follows  a  locality  constraint  on  the  in-­‐

terpretations   and   forms   of   roots:   root-­‐formed   words   are   more   likely   to   display   mor-­‐

phonological  irregularities  and  semantic  idiosyncrasies  than  word-­‐formed  words.  Refer   to  Sect.  3.3  for  empirical  evidence  from  different  languages,  including  Maltese,  in  favor   of  the  distinction  between  root  and  word  derivation,  and  for  potential  counterexamples   to  the  predictions  made  by  the  model.      

Based  on  these  assumptions  and  generalizations,  we  can  reinterpret  the  morpho-­‐

syntactic  realization  and  semantic  behavior  of  the  four  categories  (singletons,  argument   alternations,  multiple  interpretations,  and  synonyms)  in  light  of  the  distinction  in  word   formation.  To  preempt,  the  result  is  that  singleton  verbs,  synonyms,  multiple  interpre-­‐

tations,   causative,   inchoative   and   noncausative   verbs   are   root-­‐derived   whereas   pas-­‐

sives  and  reflexives  are  word-­‐derived.12  Let  us  now  survey  the  two  groups  in  some  de-­‐

tail,  beginning  with  cases  of  root  derivation.        

 

Root-­derived  verbs  

Unsurprisingly,   root   derivation   gives   rise   to   verbs   belonging   to   very   different   catego-­‐

ries.  Verbs  directly  created  from  a  root  range  from  verbs  that  stand  on  their  own  (sin-­‐

gletons)  to  verbs  that  are  morphologically  related  to  other  verbs  (causatives,  synonyms,   etc.);  from  verbs  whose  meanings  are  systematically  related  (synonyms,  verbs  entering   the   causative-­‐inchoative   alternation)   to   verbs   that   are   semantically   quite   far   apart   (multiple  interpretations).      

Singletons  are  root-­‐derived  for  two  main  reasons.  First,  they  display  many  gaps  in   the   system.   The   actual   presence   of   singletons   means   there   are   eight   other   binyanim,   which,   at   the   intersection   with   that   same   root,   yield   a   phonologically   ill-­‐formed   verb   (e.g.,  a  stop-­‐initial  root  in  binyan  VIII)  or  a  verb  form  that  cannot  be  interpreted.  Sec-­‐

ond,   the   morphological   marking   of   singletons   is   relatively   variable,   in   the   sense   that    

12  Strikingly,  these  findings  echo  to  a  great  extent  the  conclusions  Arad  (2005:  Ch.  6)  draws  for  templatic   verbs  in  Hebrew,  suggesting  strong  similarities  in  the  verbal  systems  of  Semitic  languages.  

they   may   appear   in   any   of   the   nine   patterns,   as   the   following   examples   show,   even   though  most  of  them  occur  in  I  and  II  (cf.  Sect  4.2.1.1).  

 

Root   Binyan   Verb   Meaning  

       

√xrq   I   xeraq   suit  

√slj   II   salla   swear  

√lgħb   III   liegħeb   drool  

√ħsr   V   tħassar   take  pity  on  

√bgħl   VI   tbagħal   work  hard  

√fxx   VII   infexx   erupt  in  

√sjd   VIII   stad   fish  

√żrq   IX   żrieq   become  azure  

√’dn   X   stieden   invite  

 

It  goes  without  saying  that  verbs  with  multiple  interpretations  are  also  formed  at   the  first  merging  of  a  root  with  a  functional  head.  As  Arad  (2005:  203)  rightly  observes,   the   only   difference   between   these   verbs   and   singletons   lies   in   “the  number  of   verbal   environments   in   which   they   are   assigned   an   interpretation.”   In   fact,   unlike   roots   that   take   on   an   interpretation   in   one   pattern   only,   the   roots   belonging   to   the   category   of   multiple   interpretations   acquire   a   different   interpretation   in   more   than   one   pattern.  

And  these  interpretations  are  irreconcilable.  Consider  a  couple  of  examples  mentioned   in   Sect.   4.2.1.3:   the   interpretations   that   √lqt   or   √tbgħ   are   assigned   in   binyan   I   (laqat  

‘hit’,  tebagħ  ‘print’)  and  in  binyan  II  (laqqat  ‘collect’,  tebba’  ‘stain’)  are  not  mutually  de-­‐

rived.  In  addition,  the  analysis  of  the  morphological  realization  of  the  roots  in  this  cate-­‐

gory  demonstrated  that  there  is  no  clear  pattern  as  regards  the  forms  that  mark  verbs   with  multiple  interpretations.    

Although  synonyms  tend  to  be  realized  in  particular  groups  of  patterns  such  as  I  –   II  (e.g.,  bidel,  biddel  ‘change’)  and   VII  –  VIII   (e.g.,  instadd,  stadd  ‘clog   up’),   as  shown  in   Sect  4.2.1.4,  they  nonetheless  constitute  a  case  of  root  derivation  because  they  exhibit  a   lot  of  variation  in  the  morphological  marking.  In  fact,  synonymous  verbs  may  appear  in   a  variety  of  26  different  pairs,  15  different  triplets,  and  7  different  quadruplets.    

In  addition,  both  multiple  interpretations  and  synonyms  may  include  vocalic  varia-­‐

tion,  which  is  typical  of  root  derivation.  Some  examples  of  synonyms  derived  from  the   same  root  that  have  different  vowel  sequences  include  the  following:    

 

Root   Binyan   Verb   Meaning  

       

√ġbd   V   ġebbed   be  pulled  repeatedly  

  VII   inġibed   be  pulled  

√qlj   I   qela   fry  

  II   qalla   fry  

√slf   I   silef   lend  

  V   sellef   lend  

√sqj   I   seqa   give  water  to  

  II   saqqa   water,  irrigate  

 

Roots   that   are   assigned   multiple   interpretations   in   the   context   of   different   binyanim   may  also  have  different  vowel  patterns.  Consider  the  following  verb  forms:    

 

Root   Binyan   Verb   Meaning  

       

√tkk   II   tikkek   put  dots  on  letters  

  V   ittekkek   have  blemishes  

√xrb   I   xorob   drink  

  II   xarrab   wet  

 

Of   particular   interest   are   the   already   mentioned   roots   √frd   and   √tbgħ,   each   creating   four  different  verbs  in  I,  II,  V  and  VII.  The  verbs  in  I  –  VII  and  II  –  V  mark  different  argu-­‐

ment  alternations,  such  as  active-­‐passive  tebba’  ‘stain’,  ittebba’  ‘be  stained’,  and  transi-­‐

tive-­‐reflexive  fired  ‘separate,  v.t’,  infired  ‘separate,  v.i’.    

 

Root   Binyan   Verb   Meaning   Binyan   Verb   Meaning  

√frd   I   fired   separate  (v.t)   II   farrad   make  unpaired  

  VII   infired   separate  (v.i)   V   tfarrad   become  unpaired  

√tbgħ   I   tebagħ   print   II   tebba’   stain  (v.t)  

  VII   intebagħ   be  printed   V   ittebba’   stain  (v.i)  

 

The   two   alternating   pairs   are,   however,   difficult   to   relate   derivationally.   Because   all   verb  forms   are   derived  from  the   same   semantically  underspecified  root,  it  is  perhaps   conceivable  to  metaphorically  associate  the  state  of  being  unpaired  with  that  of  being   separated.   Similarly,   it   is   not   impossible   to   find   figurative   links   between   the   event   of   staining  and  the  event  of  printing.  However,  they  do  not  semantically  entail  each  other   in  the  same  way  that,  for  instance,  an  event  of  separating  reciprocally  entails  the  event  

of  separating.    

Evidence  in  support  of  this  claim  that  these  two  roots  are  assigned  multiple  inter-­‐

pretations  in  binyan  I  and  II  comes  from  the  morphonology  of  the  verb  pairs.  The  verb   forms   derived   from   √frd   show   vocalic   variation,  i-­e   in   binyan   I,  a-­a   in   binyna   II.   The   verbs   created   from   √tbgħ   retain   the   same   vowel   sequence,   however   the   virtual   pho-­‐

neme  għ  gets  realized   as   /ħ/  in  binyan  I,   /ˈtɛbɐħ/,  and   has  no   phonetic  realization  in   binyan  II,  /ˈtɛbbɐ/.  I  argue  that  the  verbs  in  binyan  VII  and  V  are  verb-­‐derived  from  the   verbs  in  I  and  II  respectively.  For  this  reason,  they  inherit  not  only  the  form  of  the  base   verbs  (same  vowel  sequence,  same  phonetic  realization  of  għ),  but  also  the  semantics  of   the  corresponding  verbs  (they,  in  fact,  constitute  cases  of  argument  alternations).  These   observations  are  in  line  with  the  predictions  about  the  morphonological  and  semantic   behavior   of   root-­‐derived   and   word-­‐derived   words.   Let   us   now   turn   to   more   cases   of   verb  derivation,  leaving  causative  and  non-­‐causative  verbs  for  last.  

 

Verb-­derived  verbs  

The  above  cases  of  root  derivation,  where  verbs  are  built  at  the  first  merging  of  a  root   with  a  functional  head,  involve  gaps  in  the  system,  special  meanings  assigned  in  differ-­‐

ent  verbal  environments,  vowel  change,  and  a  great  deal  of  variation  in  the  morphologi-­‐

cal   realization   of   the   verbs   falling   in   a   given   category.   Passives   and   reflexives,   on   the   contrary,  are  almost  always  morphologically  and  semantically  regular.  An  examination   of   their   behavior   in   the   binyan   system   suggests   that   they   must   be   instances   of   verb   derivation,  as  they  are  dependent  on  the  form  and  meaning  of  their  transitive  counter-­‐

parts.    

Evidence  for  this  claim  comes  first  of  all  from  the  asymmetry  between  actives  and   passives,  on  the  one  hand,  and  non-­‐reflexives  and  reflexives,  on  the  other.  While  all  pas-­‐

sives  have  a  corresponding  active  verb,  there  are  some  active  verbs,  such  as  ċeda  ‘con-­‐

cede’,  għama  ‘blind’,  stieden  ‘invite’,  which  stand  on  their  own,  without  a  passive  coun-­‐

terpart.   In   such   cases   it   is   possible   to   resort   to   the   periphrastic   passive   (7a)   or   the   pseudo-­‐passive  (7b).  

 

 (7a)   Il=kelliem     ġie       mistieden      

DEF=speaker     come.PFV.3SG.M     inivite.PST.PTCP   mil-­l=kumitat     organizzattiv.  

from-­‐DEF=committee   organizing    

‘The  speaker  was  invited  by  the  organizing  committee.’  

(7b)   Lil       ħu=h         sawt-­u=h    

OBJ       brother=his     beat-­‐PFV.3PL=him     u       għam-­ew=h       minn       waħd-­a.  

and       blind-­‐PFV.3PL=him     from     one-­‐F  

‘His  brother  was  beaten  and  blinded  in  one  eye.’  

 

Similarly,  reflexives  are   derivationally  related  to  their  corresponding   transitive  verbs.  

There  are  many  transitive  verbs  which  lack  a  reflexive  counterpart,  but  there  is  not  one   reflexive  templatic  verb  that  exists  independent  of  a  transitive  verb  form.  Such  verbs  as   xorob  ‘drink’,  wera  ‘show’,  wieġeb  ‘answer’,  can  only  be  reflexivized  periphrastically,  as   in  (8),  if  they  can  be  conceived  in  a  reflexive  sense  at  all.  

 

(8a)   It=tifel       wera       ruħ=u       sodisfatt.    

DEF=boy     show.PFV.3SG.M     soul=his   pleased  

DEF=boy     show.PFV.3SG.M     soul=his   pleased