• Keine Ergebnisse gefunden

4.1  Database  of  roots  and  patterns  

 

Having  established  that  the  relations  between  templatic  verbs  in  Maltese  are  partly  sys-­‐

tematic  and  partly  unpredictable,  it  becomes  very  important  to  quantify  the  regular  and   the  irregular  share  in  the  binyan  system.  This   section,  in  fact,  takes  a  quantitative  ap-­‐

proach,   addressing   such   questions   as   the   following.   How   many   verb-­‐creating   tri-­‐   and   quadri-­‐consonantal  roots  are  there  in  Maltese?  How  many  verbs  does  each  root  create?  

Which  verbal  patterns  are  the  most  or  least  productive?  In  order  to  resolve  these  and   other  related  issues,  it  was  crucial  to  compile  an  exhaustive  list  of  consonantal  roots  in   Maltese  and  the  binyanim  they  appear  in.  In  what  follows,  I  describe  the  process  of  data   collection   and   provide   detailed   analyses   of   the   distribution   and   productivity   of   roots   and  patterns.    

4.1.1  Data  collection  

 

In  the  previous  chapters,  Maltese  templatic  verbs  were  described  as  a  combination  of   two  bound  morphemes,  a  consonantal  root  and  a  binyan  interwoven  within  each  other   in  a   non-­‐concatenative  fashion.  In  Maltese   there  are  around  2000  verb-­‐creating  roots   and   eleven   binyanim,   nine   for   tri-­‐consonantal   roots   and   two   for   quadri-­‐consonantal   roots  (cf.  Table  2.5  for  the  allomorphic  variations  of  each  binyanim,  which  are  dictated   by  different  root  types,  such  as  reduplicative,  weak-­‐final,  etc.).  I  have  also  shown  that,   subject  to  certain  combinatorial  constraints  (cf.  Sect.  2.2.2),  a  tri-­‐consonantal  root  can   in  principle  combine  with  any  of  the  nine  binyanim  to  create  different  verbal  lexemes.  

In   practice,   however,   no   tri-­‐consonantal   root   appears   in   all   nine   patterns.   Rather,   around   two   thirds   of   all   tri-­‐consonantal   roots   appear   in   only   one   or   two   binyanim.  

Quadri-­‐consonantal  roots,  by  contrast,  do  combine  with  both  QI  and  QII  binyanim,  but   about  half  of  them  combine  with  one  pattern  only  (see  data  below).  The  discussion  that   follows   revolves   primarily   around   tri-­‐consonantal   roots   and   patterns.   I   occasionally   make  reference  to  quadri-­‐consonantal  verbs,  but  a  fuller  treatment  of  these  verbs  must   await  future  work  (cf.  Ellul  2010  for  a  brief  analysis).    

Although  roots  seem  to  combine  freely  with  patterns,  the  distribution  of  templatic   verbs  within  the  lexicon  is  not  entirely  random.  As  we  shall  see  in  the  rest  of  this  chap-­‐

ter,   there  is  a   degree  of  organization  in  the  binyan  system,   particularly  in  the   greater   than  chance  co-­‐occurrence  of  patterns  for  a  given  root  when  one  verb  is  derived  from   another.   In   order   to   achieve   a   full   understanding   of   the   way   consonantal   roots   are   mapped  into  the  various  binyanim,  it  was  necessary  to  set  up  an  exhaustive  database  of   tri-­‐  and  quadri-­‐consonantal  roots  and  record  the  number  of  verbs  in  each  pattern.    

Using   Serracino-­‐Inglott’s   (1975-­‐1989)   monolingual   dictionary,   Aquilina’s   (1987-­‐

1990)  bilingual  dictionary,  and  Mifsud’s  (1995a:  272-­‐295)  list  of  loan  verbs  that  have   been  fully  integrated  into  the  root-­‐and-­‐pattern  system  of  Maltese,  a  database  of  all  con-­‐

sonantal  roots  which  give  rise  to  one  or   more  patterns  was  compiled  in  the  form  of  a   spreadsheet.  A  small  extract  from  the  database  for  tri-­‐consonantal  roots  is  reproduced   in  Table  4.1.  A  full  listing  of  the  data  tabulated  by  root  and  pattern  is  given  in  Appendix  I   (tri-­‐consonantal  roots)  &  Appendix  II  (quadri-­‐consonantal  roots).  

The   database   contains   all   the   templatic   verbs   recorded   in   the   mentioned   works,   without   discriminating   between   dialectal,   slang,   bookish,   current   or   dated   forms.   For   instance,  under  the  root  √ktb  one  finds  not  only  the  verbs  kiteb  ‘I,  write,  enroll  (v.t)’  and  

inkiteb  ‘VII,  be  written,  enroll  (v.i)’,  which  have  a  high  frequency  of  use,  but  also  the  verb  

tkieteb  ‘VI,  engage  in  mutual  correspondence’,  which  has  fallen  out  of  use.  I  also  added   some  templatic  verbs,  unrecorded  in  dictionaries  of  Maltese,  which  I  came  across  in  lit-­‐

erary   works   (e.g.,  lajjem   ‘II,   slow   down’,   in   Rużar   Briffa’s   poem  Quo   Vadis?)   or   heard   other  native  speakers  use  (e.g.,  irħas  ‘IX,  become  cheap’  for  more  common  raħas  ‘I,  be-­‐

come  cheap’)  or  which  I  know  as  a  native  speaker  (e.g.,  tmandar  ‘QII,  get  wasted’).  

 

Table  4.1  Extract  from  the  database  of  roots  and  patterns      

Root   I   II   III   V   VI   VII   VIII   IX   X  

                   

krh     kerrah     tkerrah         krieh   stkerrah  

krm   korom   *karrem     *tkarrem     nkaram        

krr   *kerr                 stkerr  

krt     karrat                

ksb   kiseb   kisseb     tkisseb     nkiseb        

ksħ   kesaħ   kessaħ     tkessaħ            

ksr   kiser   kisser     tkisser     nkiser        

kss   kess                  

ktb   kiteb   *kitteb       tkieteb   nkiteb        

ktf     kittef     tkittef            

                   

 

In  principle,  it  is  also  possible  to  compile  an  inventory  of  only  the  verbal  lexemes   that  an  educated  speaker  of  modern  Maltese  has.  The  list  would  be  drastically  reduced,   with  some  roots  appearing  in  fewer  binyanim  (e.g.,  under  √sgħl  one  would  list  sagħal  ‘I,   cough’  but  not  the  obsolete  siegħel  ‘III,  cause  one  to  cough’),  and  other  roots  being  omit-­‐

ted  altogether  (e.g.,  √kbx  and  √wrx  deriving,  among  others,  kebbex  ‘be  unfaithful’  and   werrex  ‘slap  someone’s  face’  respectively).  Since  at  the  time  of  compiling  the  database   there   was   no   reliable   dictionary   or   language   resource   that   provides   categorization   of   lexemes  on  the  basis  of  their  usage,  it  was  not  always  easy  to  decide,  for  instance,  which   forms  are  current  and  which  are  dated.1  With  such  resources  in  hand,  it  is  possible  to   replicate   the   exercise,   providing   additional   results   in   terms   of   the   usage   and   the   fre-­‐

quency  of  use  of  the  lexemes.  A  few  verbs,  which  dictionaries  mark  as  unused,  obsolete    

1  Carrying  corpus  studies  would  certainly  help  make  such  usage-­‐based  decisions.  However,  because  of  the   normative  nature  of  writing,  a  number  of  verbs,  being  dialectal,  tend  to  be  used  more  often  in  the  spoken   rather  than  in  the  written  language.  The  present  lack  of   corpora  of  spoken  Maltese  and  dialectal  Maltese   clearly  does  not  facilitate  the  undertaking  of  such  a  task.  

or  hypothetical  are,  however,  entered  in  the  database,   marked  by  an  asterisk,  but  are   excluded  from  the  calculations  given  here.  

The   discussion   that   follows   is   therefore   based   on   an   all-­‐embracing   database   of   templatic  verbs,  which  add  up  to  almost  4000  verbs  derived  from  over  1900  different   consonantal  roots.  A  breakdown  of  the  results  is  given  in  Table  4.2,  which  shows  three   things:  the  total  number  of  tri-­‐  and  quadri-­‐consonantal  roots,  the  number  of  verbs  de-­‐

rived  from  both  kinds  of  roots,  and  finally  the  mean  number  of  patterns  represented  by   root.  One  striking  observation  that  comes  out  of  this  table  is  that  there  are  numerous   roots  that  are  not  used  to  form  actual  templatic  verbs.  Only  25%  of  all  possible  combi-­‐

nations   of   tri-­‐consonantal   roots   with   binyanim   are   attested   verbs   in   Maltese,   leaving   9572   binyan   slots   empty.   By   contrast,   quadri-­‐consonantal   roots   exhibit   much   fewer   gaps,  creating  718  verbs,  which  make  up  75%  of  the  total  number  of  possible  combina-­‐

tions  of  quadri-­‐consonantal  roots  with  binyan  QI  and  binyan  QII.  

 

Table  4.2  Total  consonantal  roots  and  verbs  they  create    

  No.  of  Roots   No.  of  verbs   Mean  patterns/root  

       

Tri-­‐consonantal   1424   3244   2.28  

Quadri-­‐consonantal   478   718   1.5  

Total   1902   3962    

       

 

Before  presenting   more  results,  a   word  is  in  order  on  the  classification   of  homo-­‐

phonous  roots  in  the  database.  In  all,  there  are  around  400  homophonous  roots,  such  as  

√bjd1  bajjad  ‘whitewash’,  √bjd2  bied  ‘lay  eggs’,  and  √sfr1  sfar  ‘turn  yellow’,  √sfr2  saffar  

‘whistle’,  √sfr3  siefer  ‘travel’.  They  are  usually  the  result  of  historical  merges  of  different   phonemes,  in  particular  emphatic  consonants  with  their  non-­‐emphatic  counterparts  (cf.  

Aquilina  1970;  Mifsud  2008;  inter  alia).    

In   general,   it   seems   reasonable   to   have   two   (or   more)   separate   entries   for   such   roots  both  on  etymological  and  semantic  grounds.  However,  because  a  number  of  con-­‐

sonantal  roots  are  assigned   multiple  interpretations  that  are   difficult  to  relate  deriva-­‐

tionally,   at   least   synchronically   (recall   verbs   like  ħareġ  ‘take   out’,  ħarreġ  ‘train’   and   stħarreġ  ‘investigate’,  which  are  all  derived  from  the  same  root,  √ħrġ),  it  is  not  always   easy  to  tell  apart  cases  of  homophony  from  cases   of  multiple  interpretations.  In  some   studies,   such   as   Borg   (1981,   1988),   verbs   derived   from   two   etymologically   distinct  

roots,   like  għalaq  ‘close’   and  għallaq  ‘hang,   strangle’,   are   treated   on   a   par   with   verbs   such  as  fetaħ  ‘open’   and  fettaħ  ‘loosen’,   which  etymologically  belong  to   the  same  root   and  which  in  this  study  are  analyzed  as  multiple  interpretations  of  one  root.  In  relation   to  this,  Fabri  (2009:  4)  writes:  

 

It   is   not   clear   to   what   extent   these   forms   (=   binyanim)   are   productive   in   modern   Maltese,   and   very  little  is  known  about  native  speaker  intuitions  about  the  relations  between  the  members  of  a  

‘family’.  For  example,  to  what  extent  does  a  native  speaker  see  the  verb  għallaq  ‘strangle’  as  a  sec-­‐

ond  form  derivation  of  għalaq  ‘close’,  even  though  historically  the  two  are  unrelated  (see  Aquilina   1990:  954,  957)?  In  other  words,  it  is  not  at  all  clear  how  psychologically  real  these  families  are  to   native  speakers.  

 

From  this  observation  arises  an  interesting  question  concerning  the  semantic  related-­‐

ness  of  words  sharing  the  same  root  consonants  which  may  be  of  identical  or  different   historical  origin.  However,  this  is  an  issue  which  requires  psycholinguistic  and  corpus-­‐

based   research   that   makes   use   of   such   methods   as   the   Latent   Semantic   Analysis   (cf.  

Landauer  &  Dumais  1997;  Landauer  et  al.  1998;  inter  alia).    

For  the  present  purposes,  it  is  important  to  avoid  confusion  between  semantic  un-­‐

relatedness  that  is  due  to  different  etymologies,  on  the  one  hand,  and  tenuous  semantic   relations  among  verbs  derived  from  the  same  root  by  virtue  of  the  characteristic  of  con-­‐

sonantal  roots  to  be  assigned  multiple  interpretations,  on  the  other.  For  this  reason,  in   line  with  Aquilina’s  (1987-­‐1990)  dictionary,  which  points  out  the  etymological  sources   of  roots,  homophonous  roots  are  distinguished  in  the  database  by  means  of  superscript   numbers,  such  as  √bjd1  and  √bjd2.  

 

4.1.2  Productivity  of  roots  and  patterns  

 

There  are  a  number  of  interesting  ways  to  analyze  the  raw  data  in  order  to  gain  insights   into  how  roots  interact  with  patterns.  In  this  section,  I  take  a  closer  look  at  the  database   from  three  different  yet  related  angles.  I  first  examine  the  distribution  of  roots  across   different  binyanim.   Following   that,  I   give   a  quantitative   analysis  of  the   morphological   productivity  of  binyanim  derived  from  tri-­‐consonantal  roots.  Finally,  I  reinterpret  these   results  in  light  of  the  binyanim’s  preference  for  different  root  types,  weak  and  strong,   with  the  aim  of  determining  the  extent  to  which  binyanim  are  conditioned  by  the  mor-­‐

phonological  shape  of  the  root.  

 

Distribution  of  roots  

Let  us  begin  by  looking  at  the  frequencies  for  patterns  per  root.  Table  4.3a  shows  the   number   of   tri-­‐consonantal   roots   that   appear   in   different   patterns.   Very   few   roots   are   cast  in  more  than  two  patterns:  only  a  third  of  tri-­‐consonantal  roots  appear  in  three  or   more  patterns.  It  is  striking  that  most  of  the  roots  appear  in  only  one  (27%)  or  two  pat-­‐

terns  (40%),  leaving  a  significantly  large  number  of  binyan  slots  empty.  This  means  that   a  very  large  number  of  patterns  are  not  used  for  creating  actual  templatic  verbs  .    

 

Table  4.3a  Actual  number  of  patterns  per  tri-­‐consonantal  roots    

               

Patterns/root   1   2   3   4   5   6   7  

No.  of  roots   380   567   249   161   64   2   1  

Percentage   26.69   39.82   17.49   11.31   4.49   0.14   0.07  

               

 

The  exercise  was  repeated  for  quadri-­‐consonantal  roots.  The  results,  as  summarized  in   Table  4.3b,  show  that  a  little  more  than  half  of  the  total  of  quadri-­‐consonantal  roots  cre-­‐

ate  verbs  in  both  patterns.  The  rest  typically  appear  only  in  binyan  QI,  such  as  √gdwd   gedwed  ‘mutter’,  √ħbrk  ħabrek  ‘strive’,  √slpj  salpa  ‘sail’,  etc.  A  much  smaller  number  is   cast  only  in  QII,  such  as  √gxtr  tgexter  ‘become  unsociable’,  √rnġj  tranġa  ‘be  set  right’.  

The  entire  list  of  quadri-­‐consonantal  roots  and  the  verbs  derived  from  them  is  given  in   Appendix  II.    

 

Table  4.3b  Actual  number  of  patterns  per  quadri-­‐consonantal  roots    

     

Patterns/root   1   2  

No.  of  roots   238   240  

Percentage   49.79   50.21  

     

 

 

The  rest  of  this  chapter  deals  with  the  interrelations  of  tri-­‐consonantal  roots  with   patterns,   leaving  an  in-­‐depth   analysis  of  quadri-­‐consonantal  roots   and   patterns  to  fu-­‐

ture   research.   Let   us   now   turn   to   the   morphological   productivity   of   the   nine   tri-­‐

consonantal  binyanim.    

   

Morphological  productivity  of  binyanim  

A  more  interesting  question  concerns  the  morphological  productivity  of  patterns.  Say-­‐

ing  that  a  given  binyan  is  productive  does  not  mean  that  it  is  more  often  used  for  the   production  of  neologisms  or  for  the  integration  of  loan  verbs.  Rather,  in  this  subsection,   I  measure  the  profitability  of  the  various  binyanim  in  the  lexicon;  that  is,  the  extent  to   which   they   are   actively   used.   An   analysis   of   patterns   in   terms   of   their   availability   re-­‐

quires   future   psycholinguistic   and   corpus-­‐based   research.2   Research   on   Semitic   lan-­‐

guages  along  these  lines  has  been  carried  out,  among  others,  by  Bolozky  (1999)  for  Is-­‐

raeli  Hebrew  and  Verheij  (2000)  for  Biblical  Hebrew.  

In  Table  4.4,  I  give  the  productivity  rates  of  the  nine  binyanim,  indicated  by  Roman   numerals.  Over  three  fourths  of  the  total  number  of  tri-­‐consonantal  roots  create  verbs   in   binyan   I,   II   and   V.   Another   relatively   productive   pattern   is   VII,   accommodating   around   10%   of   tri-­‐consonantal   roots.   The   other   patterns   are   virtually   unproductive,   with  only  13%  of  the  roots  appearing  in  the  five  of  them  put  together.    

 

Table  4.4  Productivity  of  binyanim  derived  from  tri-­‐consonantal  roots    

                   

Pattern   I   II   III   V   VI   VII   VIII   IX   X  

Frequency   696   994   102   790   137   337   91   73   24  

%  of  roots   21.45   30.64   3.14   24.35   4.22   10.39   2.80   2.25   0.74  

                   

 

Here,  binyanim  were  treated  individually.  In  the  next  section,  I  tackle  the  interrelations   among  patterns  on  the  basis  of  their  co-­‐occurrence  on  the  same  roots,  thereby  assess-­‐

ing  the  degree   to  which  the  binyanim  can  be   said  to  form  a   system.  Before   going  into   this  analysis,  however,  let  us  study  the  relationship,  if  any,  between  the  formal  aspects   of  the  root  and  the  productivity  of  the  patterns.          

 

Binyanim  and  root  types  

Having  looked  at  how  roots  are  distributed  across  patterns  and  at  the  tendency  of  roots   to  combine  with  particular  patterns,  I  now  address  the  issue  of  possible  preferences  a   given  binyan  might  have  for  different  root  types.    

In  Sect.  2.2.1,  roots  were  categorized  into  different  types,  drawing  a  main  distinc-­‐

 

2  In  Sect.  2.4,  I  briefly  discuss  the  distinction  between  the  availability  and  profitability  of  templatic  verbs  in   Maltese.  

tion  between  strong  and  weak  roots.  While  the  consonants  of  strong  roots  are  always   audible  and  normally  not  affected  by  morphonological  processes,  weak  roots  have  the   glides  /j/  or  /w/  as  one  of  their  radicals,  which  are  silent  in  some  words,  giving  rise  to  

‘irregular’  forms.  Strong  roots  were  further  subdivided  into  regular  C1C2C3  and  redupli-­‐

cative  C1C2C2.  There  are  three  different  subtypes  of  weak  roots,  depending  on  the  posi-­‐

tion  of  the  glide,  in  initial,  medial  or  final  position.  As  discussed  in  detail  in  Ch.  2,  root   types  determine  the  syllabic  structure  of  verbs  and  have  an  effect  on  their  inflectional   morphology.  These  five  root  types  are  summarized  in  Table  2.4,  reproduced  here  with   some  minor  changes  as  Table  4.5.  

 

Table  4.5  The  main  types  of  tri-­‐consonantal  roots  in  Maltese    

Root  type   Example(s)   Verb(s)   Meaning(s)  

       

Strong   regular   √ktb   kiteb   write  

  reduplicative   √xmm   xamm   smell  

       

Weak   initial   √wħl,  √jsr   weħel,  jassar   stick,  enslave  

  medial   √dwr,  √tjr   dar,  tar   turn,  fly  

  final   √dgħw,  √qrj   dagħa,  qara   swear,  read  

       

 

Let  us  now  look  into  the  issue  of  whether  the  morphonological  shape  of  the  root,   i.e.  if  it  is  strong  or  weak,  puts  a  constraint  on  the  selection  of  a  binyan.  If  this  were  the   case,  then  this  would  suggest  that  the  distribution  of  binyanim  is  to  some  degree  condi-­‐

tioned  by  formal  properties  of  the  root.    

To  begin  with,  the  total  number  of  roots  and  the  verbs  they  form  are  laid  out  in  Ta-­‐

ble  4.6  according  to  the  root  type.3  Two  main  observations  can  be  drawn  from  this  ta-­‐

ble.  First,  over  two  thirds  of  the  tri-­‐consonantal  roots  in  Maltese  are  strong,  the  major-­‐

ity   of   which   are   regular   rather   than   reduplicative.   Second,   around   27%   of   templatic   verbs  (excluding  those  formed  from  quadri-­‐consonantal  roots)  are  derived  from  weak   tri-­‐consonantal  roots.  

   

3  Note  that  one  particular  root,  √wċċ,  is  grouped  with  reduplicative  roots,  but  can  instead  be  entered  with   weak-­‐initial   roots.  Also  note  that  the   12   irregular   roots,  i.e.  the  roots  that  for   historical  reasons   have  the   first  radical  missing  (e.g.,  √’ħd  creating  ħa  ‘I,  take’  and  ittieħed  ‘VI,  be  taken’),  are  classified  with  the  strong   regular  roots,  unless  their  last  radical  is  a  glide  (e.g.,  √’dj  forming  idda  ‘VIII,  shine’),  in  which  case  they  are   listed  with  weak-­‐final  roots.  

Table  4.6  Tri-­‐consonantal  roots  and  verbs  they  create  grouped  by  root  type  

       

   

   

   

   

The  chart  reveals  that:  

• All  binyanim  have  a  high  tendency  to  combine  with  strong  regular  roots,  especially   VI   and   VIII.   The   only   exception   is   X,   which   shows   an   almost   equal   preference   for   regular,  reduplicative  and  weak-­‐final  roots.  

• Relatively  speaking,  reduplicative  roots  appear   most  frequently  in  X.  They  are   em-­‐

bedded  frequently  in  I,  II,  V,  VII  and  VIII,  but  almost  never  in  III  and  VI.    

Weak-­initial  roots  never  appear  in  VIII  and  hardly  ever  in  I  and  VII.  They  are  rela-­‐

tively  more  commonly  cast  in  II,  III,  V,  VI  and  X.        

• Binyan  III,  VI  and  VIII  do  not  generally  accommodate  weak-­medial  roots,  which  tend   to  interleave  with  II,  V,  IX  and  X.    

• In  proportion  to  all  types,  weak-­final  roots  combine  most  frequently  with  III  and  X.  

It  is  relatively  common  for  this  root  type  to  be  inserted  in  I,  VI  and  VII,  but  not  in  II,   V  and  IX.4      

• Binyan   III   and   VI   tend   to   combine   with   regular   and   weak-­‐final   roots,   and   are   underrepresented  on  the  other  three  root  types.  

• It   is   more   often   that   binyan   II,   V   and   IX   combine   with   regular   and   weak-­‐medial   roots  than  with  other  root  types.  However,  II  and  V  also  show  a  relative  preference   for  reduplicative  roots.    

 

 

4  There  is  only  one  weak-­‐final  root  which  combines  with  IX,  √ħlw,  creating  ħliel  ‘become  sweet’,  where  the   root  is  reanalyzed  as  a  reduplicative  one,  with  the  final  glide  turning  into  a  liquid.      

These  findings  suggest  that  there  is  some  morphological  conditioning  of  binyanim   by   root   type,   which   can   be   couched   in   terms   of   (dis)preferences.   While   regular   roots   generally   combine   with   all   binyanim,   reduplicative   roots   have   a   strong   dispreference   for  III  and  VI.  The  combination  of  weak-­‐initial  roots  with  I,  VII  and  VIII  is  dispreferred   or  even  not  attested.  There  is  a  relatively  low  number  of  verbs  formed  by  the  combina-­‐

tion  of  weak-­‐medial  roots  and  binyan  III,  VI  and  VIII,  on  the  one  hand,  and  of  weak-­‐final   roots  and  binyan  II,  V  and  IX,  on  the  other.  These  observations  are  summarized  in  Table   4.8.  

 

Table  4.8  (Dis)preferences  of  combinations  of  root  types  and  binyanim      

Patterns   Preferred  type   Example   Dispreferred  type  

       

I,  VII,  VIII   R.   √xħt   W-­‐I  

  Red.   √sdd    

II,  V,  IX   R.   √ċkn   W-­‐I  

  W-­‐M   √twl   W-­‐F  

III,  VI   R.   √ħrs   Red.  

  W-­‐F   √vrj   W-­‐M  

X   R.   √nbħ   W-­‐I  

  Red.   √knn    

  W-­‐F   √ħbj    

       

 

 

4.1.3  Co-­‐occurrence  of  patterns  

 

We  have,  so  far,  examined  the  binyanim  separately,  observing  that,  among  other  things,   I,   II   and   V   are   the   predominant   patterns   in   the   system,   and   that   VII,   although   much   smaller,   is   still   considerably   larger   than   the   other   five   binyanim,   which   together   ac-­‐

commodate  only  13%  of  all  tri-­‐consonantal  roots.  In  this  section,  I  analyze  the  patterns   as  one  system  by  looking  at  the  relations  between  binyanim,  especially  in  terms  of  their   co-­‐occurrence  on  the  same  roots.    

Table  4.9   gives   the  frequencies  of  roots  appearing   in  only   two  binyanim.   When  a   root   combines   with   two   patterns,   there   is   a   very   high   chance   it   selects   II   and   V,   e.g.,  

√wtq,  wettaq  ‘implement’,  twettaq  ‘be   implemented’.   It   is   also   relatively   frequent   for  

roots   creating   two   binyanim   to   appear   in   I   and   VII,   e.g.,   √bdj,  beda  ‘begin’,  inbeda  ‘be   begun’.   Other   less   frequent   combinations   include   I   and   II,   e.g.,   √għlj,  għola   ‘go   up   (prices)’,  għolla  ‘put  up  (prices)’,  and  III  and  VI,  as  √ħrs,  ħares  ‘protect’,  tħares  ‘be  pro-­‐

roots   creating   two   binyanim   to   appear   in   I   and   VII,   e.g.,   √bdj,  beda  ‘begin’,  inbeda  ‘be   begun’.   Other   less   frequent   combinations   include   I   and   II,   e.g.,   √għlj,  għola   ‘go   up   (prices)’,  għolla  ‘put  up  (prices)’,  and  III  and  VI,  as  √ħrs,  ħares  ‘protect’,  tħares  ‘be  pro-­‐