These  words  bear  examination .sx   
  It  seems  to  have  been  the  case  that  in  the  parent  dialect  , CF  , 
these  words  formed  part  of  a  lexically-marked  stratum  of  loans  from  
Classical  Latin .sx   Yet  in  medieval  Latin  , " all  unstressed  
vowels  had  become  lax  "  ( Halle  and  Keyser  ) .sx  
Consequently  , in  the  parent  dialect  , these  words  are  arguably  
stressed  not  according  to  the  native  French  paradigm  ( which  would  
have  tended  to  make  all  such  words  oxytone  ) , but  according  to  
Latinate  principles :sx   the  final  syllable  is  skipped  ; then  the  
penultimate  syllable  is  analysed  , and  if  it  is  light  , that  too  is  
skipped  , and  stress  is  assigned  to  the  antepenult .sx   This  principle  
CF  seems  to  gift  to  English .sx   Notice  , however  , that  in  two  of  the  
examples  above  (  Tydeus  and  Zepherus  ) , antepenultimate  
stress  can  also  be  accounted  for  by  the  generalising  of  the  OE  Word  
Rule  , which  will  assign  strongest  stress  to  the  initial  syllables  
( where  antepenultimate  =  initial) .sx   In  other  words  , we  are  now  
dealing  with  two  stress  principles  , but  with  one  result :sx   in  many  
cases  , the  survival  of  native  English  principles  of  stress  
( specifically  , the  Word  Rule  ) yields  identical  outputs  to  
stress-patterning  based  on  a  new  principle .sx   
  This  is  by  no  means  an  unusual  situation  in  language .sx   Giegerich  
( 1985 :sx 23  ) , for  example  , notes  that  the  same  thing  happens  in  
German :sx   " a  Latinate  stress  rule  has  entered  into  the  
borrowing  Germanic  language   .sx ...  One  of  the  reasons  for  the  
survival  of  this  Latinate  stress  rule  , and  indeed  for  its  
productivity  , may  be  its  compatibility  with  native  German  
vocabulary .sx   Note  that  the  [Latinate :sx   McC]  rule  copes  with  native  
and  nonnative  vocabulary  alike  and  no  such  distinction  has  to  be  
maintained  in  the  phonology  of  English  "   .sx   
  In  terms  , then  , of  the  English  of  the  14th  century  and  beyond  , 
what  we  are  looking  at  is  a  very  familiar  pattern :sx   the  central  
stress-assignment  rule  is  in  certain  words  skipping  up  to  two  light  
syllables  before  assigning  stress .sx   Halle  and  Keyser  , for  example  , 
write  of  this  that  " The  nonnative  vocabulary  of  Chaucer  
consisted  of  two  types  of  words  , namely  , learned  words  largely  of  
Latin  origin  and  everyday  words  borrowed  from  Old  French  or  
Anglo-Norman .sx   These  two  classes  had  different  stress  patterns .sx   The  
words  of  Latin  origin  were  stressed  on  the  antepenultimate  vowel  if  
the  penultimate  syllable  ended  with  a  weak  cluster  [=  was  a  light  
syllable  ; McC]  ; otherwise  , they  were  stressed  on  the  penultimate  
vowel  "  ( ) .sx  Thus  what  we  are  dealing  with  in  many  of  
the  loans  from  Central  French  are  words  derived  directly  from  Latin  
where  , as  we  have  already  seen  , final  syllables  may  be  analysed  as  
extrametrical .sx   It  is  from  this  point  on  , I  think  , that  we  can  trace  
the  history  of  Noun  Extrametricality  in  English .sx   
  Once  this  happens  , something  very  interesting  and  , as  far  as  I  
know  , unique  in  English  phonology  happens .sx   We  have  a  subsequent  
period  in  English  where  metrical  overlap  and  ambiguity  occurs .sx   
Arguably  , this  period  is  at  its  height  in  the  16th  century  , the  
period  when  it  seems  that  Latin-based  rules  begin  to  be  fully  
operative  in  English  -  and  of  course  , the  period  when  the  
quantitative  metrical  experiments  take  place .sx   
  But  let  me  return  to  the  concept  of  metrical  overlap .sx   The  point  
is  made  by  Dobson .sx   Speaking  specifically  about  late  ME  and  early  
ModE .sx   , he  notes  " the  coexistence  in  English  of  two  modes  of  
stressing  "  ( 1957  , Vol.  ) .sx  These  modes  were  the  earlier  
French  ( A-N  ) mode  , the  mode  of  right-strength  and  end-stress  , and  
the  mode  which  almost  wholly  replaced  it  , the  generalising  of  the  
native  Word  Rule  allied  with  the  new  Latinate  principles  , which  
tended  to  shift  the  primary  stress  to  the  initial  syllable .sx   Once  
this  happens  , we  have  , as  I  have  noted  , a  large  group  of  words  
which  had  secondary  stress  at  their  right  edges .sx   But  what  happened  
to  these  secondary  -  stressed  syllables ?sx   The  answer  is  that  
in  both  disyllables  and  trisyllables  , they  tend  to  be  reduced :sx   
thus  , in  disyllables  , against  empire  , and  increase  ( where  
secondary  stress  is  kept  ) , we  have  e.g.  captain  and  
  pleasure  , where  secondary  stress  is  lost .sx   As  Dobson  notes  , in  
Romance  disyllables  , " especially  if  they  were  or  became  
'popular'  , shift  of  stress  often  led  to  the  second  syllable  
becoming  totally  unstressed  "  ( 1957  , Vol.  ) .sx  This  also  
occurs  in  trisyllables  , albeit  less  frequently  ( see  e.g.  Dobson  
1957  , Vol.  ) .sx  One  bench  mark  is  Levins'  Manipulus  
vocabulorum  ,  printed  in  1570  , which  suggestively  marked  the  
following  ( primary  ) stressings :sx   mem  o  rial  , 
or  i  ginal  , geom  e  trical  ,  as  well  as  
penultimate  stress  on  ori  e  ntal  , 
sacram  e  ntal  , accid  e  ntal  and  final  
stress  on  e.g.  deb  a  te  , div  i  ne  
etc.  The  pattern  is  familiar :sx   although  there  is  a  little  way  to  go  
yet  , it  is  essentially  that  of  today  ( see  further  Halle  and  Keyser  
) .sx  
  Given  Dobson's  dating  , and  the  present  reconstruction  , it  would  
seem  that  secondary  stresses  at  the  right  edges  of  words  are  in  
many  cases  being  reduced  at  exactly  the  same  time  as  Latinate  
stress  patterning  , and  with  it  , the  possibility  of  final  syllables  
being  extra  -  metrical  , is  entering  the  language .sx   In  short  , 
we  have  a  unique  position  in  which  Latin  stress  rules  are  , at  a  
certain  stage  in  linguistic  history  , compatible  with  native  
vocabulary  and  stress-principles  ( see  again  the  quote  from  
Giegerich  1985 :sx 23  , above  p.  18) .sx   Whether  on  lexical  monosyllables  , 
disyllables  , or  trisyllables  , the  old  rules  , and  the  new  , could  
produce  similar  , and  in  many  cases  identical  outputs .sx   By  
  c.  1570  it  is  possible  to  see  the  new  lineaments  of  
English  stress  patterning .sx   In  ( 12  ) below  I  give  a  graphic  outline  
summary  of  the  chief  developments  in  English  stress  patterning  from  
  c.  1100  to  c.  1600 .sx   
  figure&caption   .sx   
  3 .sx   But  what  of  the  poets  themselves ?sx   If  my  reconstruction  and  
dating  are  correct  , surely  our  poets  should  have  detected  the  
similarities  between  their  English  , and  the  theories  of  Latin  
prosody  they  had  available  to  them  in  the  mid-  to  late  16th  
century ?sx   Although  the  evidence  is  again  difficult  to  reconstruct  , 
this  seems  to  be  what  happens .sx   A  link  can  be  made  ( and  for  
testament  to  the  possibility  of  linking  poetic  form  to  
( phonological  ) characteristics  of  the  language  , see  Allen  ) .sx  
But  to  understand  the  link  between  the  Latinate  stress  phonology  
entering  the  language  and  the  favoured  Latin  verse  forms  of  the  
later  16th  century  , I  must  first  , having  reconstructed  the  
phonological  background  , reconstruct  the  metrical  background .sx   
  While  most  discussion  on  the  metrical  background  has  focused  on  
syllabic  quantity  ( see  below  ) , stress  itself  is  not  wholly  
irrelevant  to  the  quantitative  enterprise .sx   This  is  so  for  two  
reasons :sx   first  , syllabic  quantity  in  Latin  is  directly  relevant  to  
the  placement  of  stress  , and  so  there  is  a  prima  facie  case  that  
stress  and  quantity  may  be  linked  in  the  verse  line .sx   ( However  that  
relationship  may  be  exploited  by  poets  , it  is  still  a  
relationship .sx   ) Second  , and  consequently  , one  way  of  reading  and/or  
performing  the  verse  line  ( specifically  , the  hexameter  ) seems  to  
have  been  accentual  , i.e.  whatever  the  underlying  patterns  of  
quantity  determining  the  abstract  structure  of  the  line  , one  
possible  exponent  of  that  pattern  is  an  accentual  reading  , one  
based  on  word  stress .sx   As  Allen  puts  it  ( 1973 :sx 339  ) , " it  has  
to  be  remembered  that  in  ordinary  Latin  speech   .sx ..  quantity  was  
closely  linked  with  accent  ; and  it  is  possible  that  the  untaught  
speaker  would  have  been  aware  of  quantitative  differences  only  in  
so  far  as  they  were  connected  with  the  placement  of  
accent  "   .sx   But  again  , matters  are  complicated  by  the  fact  
that  in  the  first  part  of  the  hexameter  , accent  , and  ictus  ( =  unit  
of  the  abstract  quantitative  pattern  ) did  not  necessarily  coincide  , 
and  only  , in  fact  , coincided  regularly  in  the  last  two  feet  of  the  
line .sx   Nevertheless  , the  link  between  stress  and  quantity  is  
important :sx   for  counterpoint  to  exist  at  all  in  the  first  part  of  
the  line  does  not  mean  that  stress  ( or  accent  ) is  irrelevant  , only  
that  the  link  is  differently  exploited .sx   
  Allen  ( 1973 :sx 345ff.  ) argues  convincingly  that  " in  
antiquity  the  scanning  pronunciation  [i .sx e.  the  ictic  reading  of  a  
line :sx   McC]  was  the  less  normal  way  of  reading  [than  an  accentual  
one :sx   McC]  "  and  suggests  that  such  a  " two-fold  
tradition  -  pedagogical  scanning  and  accentual  reading  "  
persists  " from  the  Latin  grammarians  to  the  present  
day  "  ( ) .sx  Attridge  ( 1974 :sx 35  ) corroborates  this :sx   
  " It  seems  certain   .sx ..  that  the  accepted  practice  in  England  
( and  elsewhere  ) was  to  read  Latin  verse  with  normal  
word-accents  "   .sx   Yet  the  tradition  of  'scanning'  was  not  dead  
in  England :sx   " We  may  conclude  that  the  best-educated  men  in  
England  at  the  end  of  the  sixteenth  century  would  have  held  that  
the  correct  way  of  reading  Latin  verse  was  with  prose  stresses  , but  
that  even  they  would  be  accustomed  to  using  the  stressed-ictus  
method  for  learning  by  heart  or  for  scanning  "  ( Attridge  
1974 :sx 40  ; see  also  Pulgram  ) .sx  The  upshot  of  this  discussion  
is  , then  , that  stress  was  relevant  to  the  quantitative  enterprise  
in  England  , even  if  in  unexpected  ways .sx   The  tradition  of  accentual  
reading  , moreover  , helps  to  explain  the  demise  of  the  movement  ( see  
section  4  below) .sx   It  remains  , however  , for  us  to  look  in  more  
detail  at  theories  of  syllabic  quantity .sx   
  The  16th  century  had  inherited  a  theory  of  Latin  prosody  
( Attridge  1974 :sx 9  ) where  syllables  could  be  long  'by  nature' :sx   these  
were  typically  lexical  monosyllables  (  nox  , 'night'  , 
  frons  , 'foliage'  ) , or  syllables  containing  long  vowels  or  
diphthongs  ( e.g.  the  first  syllable  of  via  , 'way') .sx   Syllables  
could  also  be  long  'by  position'  ( see  below) .sx   In  both  cases  , 
syllabic  length  ( or  heaviness  ) may  imply  stress .sx   The  position  rule  , 
for  example  , the  more  difficult  of  the  two  rules  , implied  that  a  
( non-final  ) syllable  was  stressed  if  it  was  heavy  ( e.g.  first  
syllables  of  cul .sx pa  , 'blame'  , fal .sx sus  , 'false') .sx   It  also  
claimed  a  syllable  was  long  ( but  not  necessarily  implying  stress  ) 
if  it  was  followed  , in  the  same  or  different  words  , by  two  
consonants  ( Attridge  1974 :sx 9  , based  on  Raven  1965 :sx 23-25  ) ; so  the  
first  syllable  of  et  d  unch  na  is  'long'  by  this  
principle .sx   There  were  also  a  group  of  syllables  falling  within  the  
purview  of  the  position  rule  which  could  be  scanned  as  either  short  
or  long  ( light  or  heavy  ) ; these  were  syllables  followed  by  plosive  
+  /l  , r/ .sx   Thus  the  first  syllable  of  atrox  , 'fierce'  , could  be  
scanned  as  light  (  a.trox  -  a  later  age  would  call  this  
initial-maximal  syllabification  ) , or  heavy  (  at .sx rox  ) -  see  
again  Attridge  ( ) .sx  These  points  are  detailed  under  ( 13) .sx   
  ( 13  ) Latin  theories  of  syllable-structure  and  versification :sx   
  a.  Syllables  long  by  nature   :sx   ( i  ) lexical  
monosyllables  ( e.g.  nox  , frons  etc.  ; see  also  the  
discussion  of  XVC#  syllables  below  ) ; ( ii  ) syllables  containing  long  
vowels  or  diphthongs  ( e.g.  via) .sx   
  b.  Syllables  long  by  position   :sx   ( i  ) first  syllables  of  
e.g.  cul .sx pa  , fal .sx sus  ;  ( ii  ) first  syllable  of  e.g.  et  
d  unch  na  ,  where  is  classed  as  'long  by  position'  
since  it  precedes  two  consonants .sx   
  c.  Syllables  classifiable  as  long  or  short :sx   at .sx rox  or  
  a.trox  , pat .sx rius  or  pa .sx trius  etc.  
  Let  me  try  to  reinterpret  this  in  terms  of  non-linear  theory .sx   
What  had  developed  in  the  history  of  English  was  a  new  kind  of  
trading  relationship  between  syllabic  heaviness  and  stress .sx   Of  
course  , it  is  arguable  that  syllabic  heaviness  has  always  traded  
with  stress  in  English  -  it  certainly  does  in  OE  , where  the  
rightward  migration  of  the  OE  Stress  Rule  is  quantity-sensitive  at  
the  left  edges  of  domains :sx   consider  here  the  contrast  between  
  t  i  mbr  o-grave  de  ( 'he  built'  , with  
secondary  stress  , since  the  first  syllable  is  heavy  ) and  
  b  i  fode  ( 'it  trembled'  , with  no  secondary  
stress  since  the  first  syllable  is  light  ; McCully  and  Hogg  
1990) .sx   
  Recall  now  what  the  new  , Latin-based  English  Stress  Rule  ( ESR  ) 
does :sx   it  assigns  foot-headedness  -  stress  -  at  the  right  edges  of  
words  ( not  the  left  edges  , as  in  OE  ) on  the  basis  of  syllable  
weight .sx   This  relationship  is  easily  observed  in  English  
monosyllables  , which  are  all  syllabically  heavy  and  all  
  " inherently  stressed  "  ( Hogg  and  McCully  ) .sx  
In  disyllables  , too  , heavy  initial  syllables  ( as  well  as  light  
ones  , a  fact  which  puzzled  some  16th-century  theorists  ) are  
primary-stressed .sx   When  our  16th-century  poets  studied  Latin  
prosody  , they  found  a  similar  relationship :sx   heavy  syllables  are  
equated  with  'length'  , and  length  implies  ( or  increases  the  
expectation  of  ) stress .sx   And  although  the  16th  century  was  to  
misinterpret  part  of  this  implication  ( by  taking  it  to  mean  not  
that  syllables  were  'long  by  position'  but  that  their  vowels  
'  lengthened  by  position'  ) , what  is  crucial  here  is  the  idea  of  
syllable  structure .sx