<s>
Neurocomputational	B-Application
speech	I-Application
processing	I-Application
is	O
computer-simulation	O
of	O
speech	O
production	O
and	O
speech	O
perception	O
by	O
referring	O
to	O
the	O
natural	O
neuronal	O
processes	O
of	O
speech	O
production	O
and	O
speech	O
perception	O
,	O
as	O
they	O
occur	O
in	O
the	O
human	O
nervous	O
system	O
(	O
central	O
nervous	O
system	O
and	O
peripheral	O
nervous	O
system	O
)	O
.	O
</s>
<s>
Neurocomputational	B-Application
speech	I-Application
processing	I-Application
is	O
speech	O
processing	O
by	O
artificial	B-Architecture
neural	I-Architecture
networks	I-Architecture
.	O
</s>
<s>
important	O
structures	O
within	O
artificial	B-Architecture
neural	I-Architecture
networks	I-Architecture
.	O
</s>
<s>
An	O
artificial	B-Architecture
neural	I-Architecture
network	I-Architecture
can	O
be	O
separated	O
in	O
three	O
types	O
of	O
neural	O
maps	O
,	O
also	O
called	O
"	O
layers	O
"	O
:	O
</s>
<s>
like	O
a	O
self-organizing	B-Algorithm
map	I-Algorithm
;	O
see	O
also	O
Fig	O
.	O
</s>
<s>
Thus	O
,	O
each	O
"	O
model	O
neuron	O
"	O
or	O
"	O
artificial	B-Algorithm
neuron	I-Algorithm
"	O
within	O
this	O
2D-map	O
is	O
physiologically	O
represented	O
by	O
a	O
cortical	B-General_Concept
column	I-General_Concept
since	O
the	O
cerebral	O
cortex	O
anatomically	O
exhibits	O
a	O
layered	O
structure	O
.	O
</s>
<s>
A	O
neural	O
representation	O
within	O
an	O
artificial	B-Architecture
neural	I-Architecture
network	I-Architecture
is	O
a	O
temporarily	O
activated	O
(	O
neural	O
)	O
state	O
within	O
a	O
specific	O
neural	O
map	O
.	O
</s>
<s>
from	O
syllable	B-General_Concept
to	O
syllable	B-General_Concept
)	O
.	O
</s>
<s>
the	O
planning	O
of	O
speech	O
articulation	O
for	O
a	O
specific	O
syllable	B-General_Concept
or	O
for	O
a	O
longer	O
speech	O
item	O
(	O
e.g.	O
</s>
<s>
Neural	O
mappings	O
(	O
in	O
contrast	O
to	O
neural	B-General_Concept
pathways	I-General_Concept
)	O
store	O
training	O
information	O
by	O
adjusting	O
their	O
neural	O
link	O
weights	O
(	O
see	O
artificial	B-Algorithm
neuron	I-Algorithm
,	O
artificial	B-Architecture
neural	I-Architecture
networks	I-Architecture
)	O
.	O
</s>
<s>
Neural	O
mapping	O
between	O
two	O
neural	O
maps	O
are	O
compact	O
or	O
dense	O
:	O
Each	O
neuron	O
of	O
one	O
neural	O
map	O
is	O
interconnected	O
with	O
(	O
nearly	O
)	O
each	O
neuron	O
of	O
the	O
other	O
neural	O
map	O
(	O
many-to-many-connection	O
,	O
see	O
artificial	B-Architecture
neural	I-Architecture
networks	I-Architecture
)	O
.	O
</s>
<s>
In	O
contrast	O
to	O
neural	O
mappings	O
neural	B-General_Concept
pathways	I-General_Concept
can	O
connect	O
neural	O
maps	O
which	O
are	O
far	O
apart	O
(	O
e.g.	O
</s>
<s>
From	O
the	O
functional	O
or	O
modeling	O
viewpoint	O
,	O
neural	B-General_Concept
pathways	I-General_Concept
mainly	O
forward	O
information	O
without	O
processing	O
this	O
information	O
.	O
</s>
<s>
A	O
neural	B-General_Concept
pathway	I-General_Concept
in	O
comparison	O
to	O
a	O
neural	O
mapping	O
need	O
much	O
less	O
neural	O
connections	O
.	O
</s>
<s>
A	O
neural	B-General_Concept
pathway	I-General_Concept
can	O
be	O
modelled	O
by	O
using	O
a	O
one-to-one	O
connection	O
of	O
the	O
neurons	O
of	O
both	O
neural	O
maps	O
(	O
see	O
topographic	O
mapping	O
and	O
see	O
somatotopic	O
arrangement	O
)	O
.	O
</s>
<s>
Example	O
:	O
In	O
the	O
case	O
of	O
two	O
neural	O
maps	O
,	O
each	O
comprising	O
1,000	O
model	O
neurons	O
,	O
a	O
neural	O
mapping	O
needs	O
up	O
to	O
1,000,000	O
neural	O
connections	O
(	O
many-to-many-connection	O
)	O
,	O
while	O
only	O
1,000	O
connections	O
are	O
needed	O
in	O
the	O
case	O
of	O
a	O
neural	B-General_Concept
pathway	I-General_Concept
connection	O
.	O
</s>
<s>
Furthermore	O
,	O
the	O
link	O
weights	O
of	O
the	O
connections	O
within	O
a	O
neural	O
mapping	O
are	O
adjusted	O
during	O
training	O
,	O
while	O
the	O
neural	O
connections	O
in	O
the	O
case	O
of	O
a	O
neural	B-General_Concept
pathway	I-General_Concept
need	O
not	O
to	O
be	O
trained	O
(	O
each	O
connection	O
is	O
maximal	O
exhibitory	O
)	O
.	O
</s>
<s>
The	O
model	O
accounts	O
for	O
a	O
wide	O
range	O
of	O
phonetic	O
and	O
neuroimaging	B-Algorithm
data	O
but	O
-	O
like	O
each	O
neurocomputational	O
model	O
-	O
remains	O
speculative	O
to	O
some	O
extent	O
.	O
</s>
<s>
The	O
speech	O
sound	O
map	O
-	O
assumed	O
to	O
be	O
located	O
in	O
the	O
inferior	O
and	O
posterior	O
portion	O
of	O
Broca	O
's	O
area	O
(	O
left	O
frontal	O
operculum	O
)	O
-	O
represents	O
(	O
phonologically	O
specified	O
)	O
language-specific	O
speech	O
units	O
(	O
sounds	O
,	O
syllables	B-General_Concept
,	O
words	O
,	O
short	O
phrases	O
)	O
.	O
</s>
<s>
Each	O
speech	O
unit	O
(	O
mainly	O
syllables	B-General_Concept
;	O
e.g.	O
</s>
<s>
the	O
syllable	B-General_Concept
and	O
word	O
"	O
palm	O
"	O
/pam/	O
,	O
the	O
syllables	B-General_Concept
/pa/	O
,	O
/ta/	O
,	O
/ka/	O
,	O
...	O
)	O
is	O
represented	O
by	O
a	O
specific	O
model	O
cell	O
within	O
the	O
speech	O
sound	O
map	O
(	O
i.e.	O
</s>
<s>
Each	O
model	O
cell	O
(	O
see	O
artificial	B-Algorithm
neuron	I-Algorithm
)	O
corresponds	O
to	O
a	O
small	O
population	O
of	O
neurons	O
which	O
are	O
located	O
at	O
close	O
range	O
and	O
which	O
fire	O
together	O
.	O
</s>
<s>
Each	O
neuron	O
(	O
model	O
cell	O
,	O
artificial	B-Algorithm
neuron	I-Algorithm
)	O
within	O
the	O
speech	O
sound	O
map	O
can	O
be	O
activated	O
and	O
subsequently	O
activates	O
a	O
forward	O
motor	O
command	O
towards	O
the	O
motor	O
map	O
,	O
called	O
articulatory	O
velocity	O
and	O
position	O
map	O
.	O
</s>
<s>
The	O
Maeda	B-Application
articulatory	I-Application
speech	I-Application
synthesizer	I-Application
is	O
used	O
in	O
order	O
to	O
generate	O
articulator	O
movements	O
,	O
which	O
allows	O
the	O
generation	O
of	O
a	O
time-varying	O
vocal	O
tract	O
form	O
and	O
the	O
generation	O
of	O
the	O
acoustic	O
speech	O
signal	O
for	O
each	O
particular	O
speech	O
item	O
.	O
</s>
<s>
In	O
terms	O
of	O
artificial	B-Application
intelligence	I-Application
the	O
articulatory	O
model	O
can	O
be	O
called	O
plant	O
(	O
i.e.	O
</s>
<s>
isolated	O
speech	O
sounds	O
,	O
syllables	B-General_Concept
,	O
words	O
,	O
short	O
phrases	O
)	O
.	O
</s>
<s>
In	O
the	O
case	O
of	O
a	O
frequent	O
syllable	B-General_Concept
,	O
a	O
co-activation	O
occurs	O
at	O
the	O
level	O
of	O
the	O
phonetic	O
map	O
,	O
leading	O
to	O
a	O
further	O
co-activation	O
of	O
the	O
intended	O
sensory	O
state	O
at	O
the	O
level	O
of	O
the	O
sensory	O
state	O
maps	O
and	O
to	O
a	O
co-activation	O
of	O
a	O
motor	O
plan	O
state	O
at	O
the	O
level	O
of	O
the	O
motor	O
plan	O
map	O
.	O
</s>
<s>
In	O
the	O
case	O
of	O
an	O
infrequent	O
syllable	B-General_Concept
,	O
an	O
attempt	O
for	O
a	O
motor	O
plan	O
is	O
generated	O
by	O
the	O
motor	O
planning	O
module	O
for	O
that	O
speech	O
item	O
by	O
activating	O
motor	O
plans	O
for	O
phonetic	O
similar	O
speech	O
items	O
via	O
the	O
phonetic	O
map	O
(	O
see	O
Kröger	O
et	O
al	O
.	O
</s>
<s>
The	O
Birkholz	B-Application
3D	I-Application
articulatory	I-Application
synthesizer	I-Application
is	O
used	O
in	O
order	O
to	O
generate	O
the	O
acoustic	O
speech	O
signal	O
.	O
</s>
<s>
5	O
indicates	O
those	O
neural	O
maps	O
and	O
processing	O
modules	O
,	O
which	O
process	O
a	O
syllable	B-General_Concept
as	O
a	O
whole	O
unit	O
(	O
specific	O
processing	O
time	O
window	O
around	O
100ms	O
and	O
more	O
)	O
.	O
</s>
<s>
The	O
two	O
mirrored	O
locations	O
of	O
the	O
phonetic	O
map	O
are	O
connected	O
via	O
a	O
neural	B-General_Concept
pathway	I-General_Concept
(	O
see	O
above	O
)	O
,	O
leading	O
to	O
a	O
(	O
simple	O
)	O
one-to-one	O
mirroring	O
of	O
the	O
current	O
activation	O
pattern	O
for	O
both	O
realizations	O
of	O
the	O
phonetic	O
map	O
.	O
</s>
<s>
This	O
neural	B-General_Concept
pathway	I-General_Concept
between	O
the	O
two	O
locations	O
of	O
the	O
phonetic	O
map	O
is	O
assumed	O
to	O
be	O
a	O
part	O
of	O
the	O
fasciculus	O
arcuatus	O
(	O
AF	O
,	O
see	O
Fig	O
.	O
</s>
<s>
This	O
signal	O
is	O
preprocessed	O
,	O
passes	O
the	O
auditory	O
map	O
,	O
and	O
leads	O
to	O
an	O
activation	O
pattern	O
for	O
each	O
syllable	B-General_Concept
or	O
word	O
on	O
the	O
level	O
of	O
the	O
auditory-phonetic	O
processing	O
module	O
(	O
ES	O
:	O
external	O
signal	O
,	O
see	O
Fig	O
.	O
</s>
<s>
The	O
phonetic	O
map	O
is	O
implemented	O
in	O
ACT	O
as	O
a	O
self-organizing	B-Algorithm
neural	I-Algorithm
map	I-Algorithm
and	O
different	O
speech	O
items	O
are	O
represented	O
by	O
different	O
neurons	O
within	O
this	O
map	O
(	O
punctual	O
or	O
local	O
representation	O
,	O
see	O
above	O
:	O
neural	O
representations	O
)	O
.	O
</s>
<s>
Three	O
examples	O
:	O
(	O
i	O
)	O
the	O
syllables	B-General_Concept
/p	O
@	O
/	O
,	O
/t	O
@	O
/	O
,	O
and	O
/k	O
@	O
/	O
occur	O
in	O
an	O
upward	O
ordering	O
at	O
the	O
left	O
side	O
within	O
the	O
phonetic	O
map	O
;	O
(	O
ii	O
)	O
syllable-initial	O
plosives	O
occur	O
in	O
the	O
upper	O
left	O
part	O
of	O
the	O
phonetic	O
map	O
while	O
syllable	B-General_Concept
initial	O
fricatives	O
occur	O
in	O
the	O
lower	O
right	O
half	O
;	O
(	O
iii	O
)	O
CV	O
syllables	B-General_Concept
and	O
CVC	O
syllables	B-General_Concept
as	O
well	O
occur	O
in	O
different	O
areas	O
of	O
the	O
phonetic	O
map	O
.	O
</s>
<s>
The	O
phonetic	O
map	O
is	O
hypermodal	O
or	O
multimodal	B-Application
:	O
The	O
activation	O
of	O
a	O
phonetic	O
item	O
at	O
the	O
level	O
of	O
the	O
phonetic	O
map	O
coactivates	O
(	O
i	O
)	O
a	O
phonemic	O
state	O
(	O
see	O
phonemic	O
link	O
weights	O
in	O
Fig	O
.	O
</s>
<s>
2009	O
)	O
,	O
(	O
ii	O
)	O
a	O
small	O
consonant	O
system	O
(	O
voiced	O
plosives	O
/b	O
,	O
d	O
,	O
g/	O
in	O
combination	O
with	O
all	O
five	O
vowels	O
acquired	O
earlier	O
as	O
CV	O
syllables	B-General_Concept
(	O
ibid	O
.	O
</s>
<s>
)	O
,	O
(	O
iii	O
)	O
a	O
small	O
model	O
language	O
comprising	O
the	O
five-vowel	O
system	O
,	O
voiced	O
and	O
unvoiced	O
plosives	O
/b	O
,	O
d	O
,	O
g	O
,	O
p	O
,	O
t	O
,	O
k/	O
,	O
nasals	O
/m	O
,	O
n/	O
and	O
the	O
lateral	O
/l/	O
and	O
three	O
syllable	B-General_Concept
types	O
(	O
V	O
,	O
CV	O
,	O
and	O
CCV	O
)	O
(	O
see	O
Kröger	O
et	O
al	O
.	O
</s>
<s>
2011	O
)	O
and	O
(	O
iv	O
)	O
the	O
200	O
most	O
frequent	O
syllables	B-General_Concept
of	O
Standard	O
German	O
for	O
a	O
6-year-old	O
child	O
(	O
see	O
Kröger	O
et	O
al	O
.	O
</s>
<s>
categorical	O
perception	O
and	O
the	B-Application
McGurk	I-Application
effect	I-Application
.	O
</s>
<s>
Furthermore	O
,	O
the	O
model	O
ACT	O
was	O
able	O
to	O
exhibit	O
the	B-Application
McGurk	I-Application
effect	I-Application
,	O
if	O
a	O
specific	O
mechanism	O
of	O
inhibition	O
of	O
neurons	O
of	O
the	O
level	O
of	O
the	O
phonetic	O
map	O
was	O
implemented	O
(	O
see	O
Kröger	O
and	O
Kannampuzha	O
2008	O
)	O
.	O
</s>
