Fundamentals in the Structure and History of Russian: A Usage-Based Approach

Chapter 3. Usage and Cognition

56 3. USAGE AND COGNITION   (2)   underlying  form     #  р  э  к  a´  #NOM.SG   #  р  э´  к  и  #NOM.PL       consonant  softening     #  р’э  к  а´   #  р’э´  к’и       ikanje   #  р’и  к  а´   does  not  apply       phonetic   #[р’и  к  а´]   #[р’э´  к’и]     This   is   how   generative   phonology   accounts   for   the   relationship   be-­‐‑ tween  [и]  in  река´  and  [э]  in  ре´ки.  An  underlying  (abstract)  form  of  a   word  is  set  up  and  all  the  actually  occurring  forms  of  that  word  are   generated  from  the  underlying  form.   3.2. Underlying Forms Versus Usage The  generative  approach  has  much  to  recommend  it,  not  the  least  be-­‐‑ ing   its   systematic   methodology.   We   do   not   wish   to   discard   what   is   useful   about   this   method,   but   generative   phonology   has   some   im-­‐‑ portant  drawbacks.  First,  the  underlying  form—such  as  #рэка´#—does   not  occur  in  Russian  in  any  word.  Paired  consonants  are  always  soft   before  front  vowels,  and  we  know  that  the  sound  [э]  can  occur  only   under  stress.  Therefore  the  proposed  underlying  form  does  not  exist  in   reality.  Generativists  answer  that  the  underlying  form  is  a  handy  ab-­‐‑ straction  that  generates,  with  proper  rules,  all  forms  that  do  exist.  Our   point  is  that  as  an  abstraction  it  represents  no  actually  occurring  form.     In  a  usage-­‐‑based  approach,  such  as  the  one  presented  in  this  text,   variants   such   as   [р’ика´]   and   [р’э´к’и]   are   learned   by   experience   and   exist  side  by  side  in  the  speaker’s  mind.  There  is  no  need  to  generate   either  one.  The  speaker  can  choose  whichever  one  is  needed  in  a  given   context.  Generalizations  (i.e.,  rules)  which  relate  alternations  emerge  in   a  speaker’s  mind  based  on  actual  linguistic  input,  i.e.,  based  on  what   the  speaker  actually  says  and  hears  every  day.  In  a  usage-­‐‑based  pro-­‐‑ gram,  the  structure  of  the  language  is  overt,  it  is  not  derived  from  a   deeper  (unverifiable)  abstraction.  It  therefore  seems  reasonable  to  as-­‐‑ sert  that  a  usage  based  description,  such  as  the  one  presented  in  this   text,   renders   a   better   picture   of   reality   than   the   generative-­‐‑based   description.     3.3. SCHEMAS 57 3.3. Schemas Throughout  this  text  we  use  a  simple  tool,  the  schema,  to  describe  re-­‐‑ lationships  between  specific  linguistic  tokens  and  the  meaningful  cate-­‐‑ gories  to  which  they  belong.  For  the  various  forms  of  the  word  река´,   for  example,  we  write  the  schema:       (3)     ‘RIVER’           [р’ика´]   [р’э´к’и]   [р’ику´]   …         (The  three  dots  above  indicate  there  are  other  forms  that  could  fit  in   here,  too.  We’re  just  not  bothering  to  list  all  the  forms  of  this  word.)   Note  that  in  this  schema  only  actually-­‐‑occurring  forms  are  part  of  the   description.  When  children  learning  Russian  first  hear  these  words  (in   any  case  form)  they  store  them  away  in  memory  as  pronounced,  hav-­‐‑ ing   learned   what   their   referent   is.   Later   they   misspell   the   word   in   school,   assuming   that   и   stands   for   all   instances   of   [и],   because   they   have  heard  [р’ика´]  (or  another  singular  form)  much  more  often  than   the   plural   [р’э´к’и].   (Shteinfeldt   reports   this   word   is   used   84%   in   the   singular,  16%  in  the  plural.)   The  second  reason  why  we  find  the  generative  approach  untenable   is  that  it  implies  that  whenever  speakers  want  to  use,  say,  the  genitive   form  of  the  word,  they  apply  one  or  more  phonological  rules  to  an   abstraction  to  come  up  with  the  proper  pronunciation.  We  think  this  is   unlikely.  Even  more  unlikely  is  the  flip  side  of  that  coin.  If  speakers   hear  [р’ик’а´],  they  must  apply  some  rule  to  degenerate  this  word  to  its   basic  abstract  form  [рэка´].  The  problem  with  this  is  that,  as  discussed   in  chapter  two,  phonetic  [и]  may  be  associated  with  /э/,  /и/,  /a/,  or  /o/!   No  generative  rule  could  pick  the  proper  association  without  looking   at  other  forms  of  the  word,  which  would  essentially  make  it  a  usage-­‐‑ based  approach.     In   cognitive   grammar   we   deal   with   “abstractions”   (such   as   pho-­‐‑ nemes   /a/,   /o/,   /т/,   etc.),   but   only   as   names   of   categories   of   sounds.   Thus  the  phoneme  /a/  is  an  abstraction  in  that  it  represents  all  the  pos-­‐‑ sible  pronunciations  of  a  low  back  vowel  in  Russian,  but  it  also  repre-­‐‑ sents  the  pronunciation  of  [a],  a  low  back  vowel.  The  same  is  true  with   meaning.   The   “abstraction”   /RIVER/   refers   to   whatever   rivers   the   58 3. USAGE AND COGNITION speaker  knows  about  and  may  even  represent  a  prototypical  river,  but   it   also   relates   to   the   Volga   or   some   other   specific   actually   occurring   river.  The  abstraction  is  nothing  more  than  a  useful  category  name  that   comes  from  the  actually  encountered  instances.         (4)     /category/         [instance1]   [instance2]     [instance3]         Where  the  category  name  is  one  of  the  instances  and  stands  for  the  rest.     Why  is  any  of  this  important?  Aren’t...