Short Summaries of the Uniform Structure Principle, the MLF
Model and the 4-M Model
(See curriculum vitae for information
about the references cited.)
The Uniform Structure Principle (USP)
Although the USP is introduced only in Myers-Scotton, 2002
(Contact Linguistics), it is obvious that it offers an
explanation for the distributions of morpheme types in many
types of bilingual speech discussed in earlier works
(including classic CS under the MLF model).
Even though the USP clearly operates in monolingual
linguistic data (i.e., uniform structure is maintained in
a given constituent type), how uniform structure applies in
bilingual data is not so obvious. Thus, there is a need to
recognize the USP when dealing with bilingual data:
Here is the USP: A given constituent type in any language
has a uniform structure and requirements of well-formedness
for this constituent type must be observed whenever that
In bilingual speech, the structures
of the Matrix Language (ML) are always preferred; that is,
following the ML's structures satisfies the Uniform
Embedded Language (EL) islands, phrases from other
varieties participating in the clause, are allowed if they
meet EL well-formedness conditions, but also ML conditions
applying to the clause as a whole (e.g. phrase placement).
See curriculum vitae for information
about Myers-Scotton, 2002; 2005c for evidence that the USP
applies in classic CS.
The Matrix Language Frame (MLF) Model Summarized
The MLF model applies only to bilingual clauses.
(Codeswitching, of course, also occurs between clauses or
sentences and across discourse turns.) The model was
intended to apply to varieties that are not mutually
intelligible; if it applies to dialects, that is a bonus.
Myers-Scotton, 1993 (
) introduces the
model, but later publications give a truer picture of the
model as it stands today (for example, see the "Afterword"
in the 1997 edition of
, or—better, see
, 2002 and later journal articles or
chapters in edited volumes. (See curriculum vitae for
In summary -- Under the MLF model, (a) Participating
do not play equal roles in structuring the bilingual
clause; (b) not all morpheme types in bilingual
constituents within this clause can come equally from the
ML and EL;
(c) Most importantly, the System Morpheme Principle limits
the occurrence of the morphemes that build clausal
structure to coming only from the ML (the outsider late
The key word characterizing the MLF model is asymmetry.
Asymmetry is obvious in (a) the roles of the participating
varieties (languages) and (b) in differences in
distribution of morpheme types.
One participating language dominates in the sense that it
supplies the morpho-syntactic frame for the bilingual
clause. This language is identified as the Matrix
Language (ML). The ML is identified by the empirical
support for the Morpheme Order Principle and the System
Morpheme Principle (see below).
Having only one language supplying the frame for the
bilingual clause is the feature that differentiates
classic CS from composite CS.
The MLF model only applies unconditionally to classic CS.
In composite CS, more than one language can contribute
abstract structure that affects the surface structure of
the morpho-syntactic frame. (See Myers-Scotton, 2002 for
Asymmetry also differentiates morpheme types under the MLF
model as either content morphemes or system morphemes.
In classic CS (under the MLF model), the role of the
non-ML, which is called the Embedded Language (EL), largely
is limited to supplying content morphemes in mixed
constituents, or EL phrase-level constituents (called EL
islands), or both.
(Note: In some data sets, early system morphemes or even an
occasional bridge system morpheme do come from the EL, but
these are not counter-examples to any provisions of the
Content morphemes are defined as assigning or receiving
thematic roles; for example, nouns receive such roles and
most verbs assign them.
System morphemes do not assign or receive thematic roles.
Under the 4-M model system morphemes are further divided
into three types.
The MLF model contains two principles that operate as
hypotheses about the participation of the different
languages involved in CS:
The Morpheme Order Principle
states that morpheme order
within the bilingual clause comes from only one language
and this language is identified as the ML.
(Note that within EL islands, which may occur in the
bilingual clause, EL morpheme order is present.)
The System Morpheme Principle
states that one type of
system morpheme must come from only one of the
participating language and this language is identified as
Note that the newer 4-M model names this type of morpheme
outsider late system morpheme
The 4-M Model Summarized
The 4-M model is not a revision of the MLF model—even
though its features help explain why the MLF model is
supported. Instead, the 4-M model is a model of morpheme
classification that should apply to language in general
(cf. Myers-Scotton & Jake, 2000; Myers-Scotton, 2002;
Myers-Scotton and Jake forthcoming).
Under the 4-M model, morphemes are classified in terms of
their empirically-evident syntactic roles, as well as
hypotheses about how they are activated in language
There are four types of morphemes: content morphemes, and
three types of system morphemes: early SMs and two types of
late SMs, bridges and outsiders.
As in the MLF model, content morphemes assign and receive
thematic roles; in this feature, they continue to differ
from all system morphemes.
But the 4-M model emphasizes a further division:
Content morphemes and early SMs group together as
conceptually-activated, as opposed to late system
morphemes, which are structurally-assigned.
means that speaker pre-linguistic
intentions activate the language-specific
semantic-pragmatic feature bundles that become lemmas in
the mental lexicon. (Lemmas are abstract features that
underlie surface level morphemes.) In this sense, content
morphemes are "directly elected" and early SMs that may
accompany them on the surface are "indirectly elected" (cf.
Levelt, 1989 and Bock and Levelt, 1994). See
Myers-Scotton, 2002 for details.
means that late SMs are not salient
until the level of the formulator (in language production).
It is at this level that larger phrases and clauses are
assembled; late SMs do this work. See Myers-Scotton, 2002
What do early SMs do?
They occur with content morphemes
(their heads) and they flesh out the meaning of their
heads. Examples: plural affixes, determiners.
In CS corpora, most early SMs come from the ML, but note
that the System Morpheme Priciple does not apply to early
SMs or bridge SMs; it only specifies that outsiders must
come from the ML.
How do bridges differ from outsiders?
Bridges are, in effect, "bridges" between elements that
make up larger constituents. Bridges depend on their form
from information within the maximal projection in which
they occur; that is, they occur to satisfy well-formedness
conditions within that projection. Examples include
elements that join together two NPs, with an associative,
equivalent or partitive meaning (e.g. "of" in English in
bone of Bora or Bora's bone
. Most bridges come from the
ML in CS.
Outsiders depend on information outsider the elements with
which they occur; this information can come from an element
in another constituent in the clause, or from the
Outsiders are the primary way that argument
structure is indicated and agreement relations are
maintained in any clause
. Examples of outsiders: morphemes
marking subject or object-verb agreement; case affixes in
Observations of how different types of morpheme have
different distributions in actual surface level
constructions give rise to
The Differential Access
This hypothesis suggests that the different types of
morphemes under the 4-M model are differentially accessed
in the abstract levels of the production process. Content
morphemes and early SMs are salient at the level of the
mental lexicon, but late SMs do not become salient under
the level of the formulator (cf. Myers-Scotton, 2002:78;
This hypothesis is an attempt to explain differences in
distributions of morpheme types. Data from codeswitching,
but also other language phenomena, support the hypothesis.