Pienemann 2015

Language Learning ISSN 0023-8333
An Outline of Processability Theory and Its

Relationship to Other Approaches to SLA
Manfred Pienemann
Paderborn University
In this article I make the point that there has been a continuous focus on second language
development in second language acquisition research for over 40 years and that there is
clear empirical evidence for generalizable developmental patterns. I will both summarize
some of the core assumptions of Processability Theory (PT) as an approach to explaining
developmental patterns and learner variation and compare the position assumed by PT
with the Dynamic Systems Theory approach proposed by de Bot, Lowie, and Verspoor
and with O’Grady’s processing-based approach to Emergentism. In addition, I will
summarize the Teachability Hypothesis and describe its limited relationship to PT in
order to respond briefly to the article on the same issue by Zhang and Lantolf in this
Special Issue.
Keywords Processability Theory; acquisition order; processing
Introduction
In this article I am concerned both with the idea of the existence of developmen-
tal sequences in second language (L2) development, the topic of the present
Special Issue, and with elucidating similar technical terms and concepts that
are being used differently in different approaches to second language acquisi-
tion (SLA) research. My goal is to clarify the historical and present context of
Processability Theory (PT; e.g., Pienemann, 1998a, 1998b, 1998c) as well as
to bridge some of the terminological gaps and conceptual differences that in
my opinion are at play in the Special Issue.
I would like to thank the following people for their invaluable support in writing this article: Anke
Lenzing, Michael H. Long, Jürgen M. Meisel, and Howard Nicholas.
Correspondence concerning this article should be addressed to: Manfred Pienemann,
Paderborn University, Department of English, Warburgerst. 100, Paderborn, Germany, 33098.
E-mail: [email protected]
Language Learning 65:1, March 2015, pp. 123–151 123

C 2015 Language Learning Research Club, University of Michigan
DOI: 10.1111/lang.12095
Pienemann An Outline of Processability Theory
In terms of PT, I will outline some of the key conceptual developments that
have taken place in SLA and in PT since the 1990s, taking care to be explicit
about the approach to the description of developing systems of morphosyntax
that is inherent in PT. In its contemporary formulation, PT is based on the inter-
play between a processing theory and a theory of linguistic knowledge, is fully
explicit, and applies to the full range of phenomena captured by the interacting
theories of language processing and linguistic knowledge. Furthermore, PT can
account for entire systems of morphosyntax rather than isolated morphosyn-
tactic aspects. With regard to terminological gaps and conceptual differences,
I will identify and critically analyze overlaps and distinctions between PT and
other key theoretical positions represented in the Special Issue.
The Continued Focus on Interlanguage Development

This Special Issue of Language Learning, titled “Orders and Sequences in the
Acquisition of L2 Morphosyntax, 40 Years On,” focuses on L2 development.
The terms “orders and sequences” and various other terms are used in this
issue and in the editors’ introduction to refer to what are currently known as
“trajectories of grammatical development.” The changed labels are significant
for a number of reasons. One key reason why the terminology has changed over
the past 40 years is the conceptual development that has taken place over this
period. In the early days of SLA research, that is, in the 1970s, there was a need
to establish basic empirical facts about SLA—whether interlanguages could be
demonstrated to be systematic languages in their own right. This was the period
of the morpheme order studies and extensive longitudinal and cross-sectional
studies, all of which established an empirical basis for orders and sequences
in L2 acquisition in support of the argument that interlanguages were neither
just translations of the fist language (L1) nor random in their organization.
In the 1980s, the focus of SLA research shifted from mere observation to
explanation—seeking to explain the logical and the developmental problems1
(Felix, 1984). It was in this period that the field diversified enormously as it
embraced epistemological approaches ranging from nativist to functionalist.
During all this time, many approaches to SLA maintained a keen interest in
the development of L2 grammatical systems and their empirical basis, including
Meisel’s research group in Germany, which focused on bilingual language
development (e.g., Meisel, 1994; 2011); Florence Myles’s group in the United
Kingdom (e.g., Myles & Towell, 2005); and the research groups in Lund (e.g.,
Håkansson & Norrby, 2010). This body of research has a strong empirical

basis, as reflected in the collection and increasing accessibility of interlanguage

corpora (e.g., Myles, 2005).
Despite the continuity of interest in the development of grammatical sys-
tems, what changed was the sophistication of the linguistic and psycholinguistic
theories employed and the sophistication of the formal accounts of L2 gram-
matical systems within these theories. The initially fairly open descriptions
have evolved into elaborated (and differing) formal accounts of developing
linguistic systems. A shared theme in the diverse formal accounts is the idea
that L2 systems develop in an orderly manner (while allowing for a degree of
variability).
In their introduction to this Special Issue the editors make the following
observation:
It appears that not very many empirical studies on acquisition orders in

SLA have been published since 1996, with the exception of: (1) the
research conducted by Pienemann and his associates in the framework of
Processability Theory [ . . . ]; (2) the research conducted in the framework
of the European Science Foundation project on untutored L2 acquisition,
conducted by Klein and Perdue and their many associates in various
countries, proposing the so-called Basic Variety of SLA [ . . . ]; and (3)
papers by Jia and Fuse (2007) and Luk and Shirai (2009). (Hulstijn, Ellis,
& Eskildsen, 2015, pp. 2–3)
Indeed, it is hard to find the initial terms “orders and sequences of L2

acquisition” in recent SLA research. However, this does not at all imply that only
two or three research groups have shown an interest in developing L2 systems
since the mid 1990s. A quick look at the articles published in the last 10 years
in the journals Second Language Research and Bilingualism: Language and
Cognition reveals quite the opposite, namely a strong and continuous interest
in L2 developmental regularities, their empirical basis, and the developmental
problem, for instance, Bonnensen (2009), Donaldson (2011), Garcı́a Mayo and
Olaizola (2011), Jaensch (2011), Schulz (2011), and van de Craats and van
Hout (2010).
These are just a few examples taken from two journals. Many of these
articles are based on corpus data. Apart from corpus data, empirical data are also
collected in online and offline experiments such as reaction time experiments,
self-paced reading experiments, grammaticality judgment tasks, and various
studies of neural activities (e.g., Felser, 2005; Roberts, 2012).
125 Language Learning 65:1, March 2015, pp. 123–151

While SLA research with a focus on the representation of linguistic knowl-

edge has grown since the mid-1990s (e.g., White, 2003), there is an uninter-
rupted tradition of research focusing on L2 development. Processability Theory
(PT) is not the only approach, but it is firmly rooted in this tradition.
In other words, the notion of developmental sequences, which was attributed
the status of a well-established finding in SLA by Long (1990) and later by
VanPatten and Williams (2006), grew into the more refined notion of develop-
mental trajectories, which is the object of a great deal of current research and
debate.
Let me add a comment on Ellis’s (2015) quotation from Long’s (1990)
article:
[L]earners of different ages, with or without instruction, in foreign and

second language settings, follow similar developmental sequences for
such items as English negation. (p. 659)
In his contribution, Ellis refers to this statement as “Long’s law.” However,

Long did not construe the above statement as a law. Instead, he attributed to it the
status of a generalization. This distinction is crucial for both general arguments
about scientific evidence and central themes in this Special Issue. In scientific
research, the term “law” has a particularly privileged status, because one of
the challenges in science is to establish whether any generalization based on
observations can (ever) be awarded the status of a law. Ellis acknowledges the
more limited status of Long’s claims by placing “law” in quotation marks in his
first reference to Long’s article and including the terms “generalizations” and
“idealizations” as characterizations of the meaning alongside the word “law.”
In doing so he acknowledges the limits on the claims that Long was making.
However, a few pages further on, he resumes repeated references to this concept
as “Long’s law” rather than maintaining his inclusion of the terms “generaliza-
tions” and “idealizations.” By not maintaining the acknowledgement of Long’s
claim as a generalization, Ellis creates space for misinterpretation by others.
As with many of the issues that I have addressed in this article, this slippage
in terminology creates an ambiguity where it is important to be clear and
consistent.
The status of the above statement as generalization is evident in the follow-
ing quotations from Long’s (1990) article:
It becomes very dangerous to claim that X is an established fact or that Y

has attained the status of a generalization or perhaps even of a law when
there is disagreement over what constitutes legitimate data and when

researchers and textbook writers are not reading or respecting the same
literature. (p. 650)
Generalizations are only possible when empirical observations are
repeatedly supported. Generalizations are crucial to, and must be
accounted for by, credible theories. It is not just the existence of
consensus that makes generalizations important for theory construction,
but the rationality of the scientific process that underlies them [. . .] the
following are a few examples (by no means intended as a comprehensive
list) of what I would claim are well-established findings about learners,
environments, and interlanguages, along with some of the challenges they
pose to current SLA theories [. . .] For each generalization, references are
provided to recent reviews and/or to key studies of the phenomenon
concerned. (pp. 656–657)
In other words, the object of our discussion here are “well-established find-
ings in SLA,” generalizations, not a law. Theoretically explicit and empirically
based debate is an ongoing requirement.
The Rationale Behind PT

The rationale behind PT is outlined in the opening lines of my 1998 book,
where I stated that a perspective on learnability that defines it as a purely
logico-mathematical problem (e.g., Berwick & Weinberg, 1984)
ignores the fact that this problem has to be solved, not by an

unconstrained computational device, but by a mind that operates within
human psychological constraints. In this book I propose a theory which
adds to learnability theory the perspective of processability [ . . . ] In my
view, the logico-mathematical hypothesis space is further constrained by
the architecture of human language processing. (Pienemann, 1998a, p. 1)
PT was designed to address the developmental problem and later also

the logical problem (see Pienemann, Di Biase, & Kawaguchi, 2005). This is
achieved by framing an L2 learner’s specific language acquisition challenges
in terms of their processing requirements as defined within Levelt’s (1989,
1993) model of language generation based on their representational correlates
as defined within Lexical-Functional Grammar (LFG; Bresnan, 2001).
In other words, PT takes a psycholinguistic approach and its representational
counterpart as its point of departure, and observations about development are
tested within this framework. We have used the PT framework to propose

Information Exchange
Locus of exchange Example Illustration

Sentence Within sentence he talk-s S
NPsubj VP
Pro V
[3rd pers sg] [pres, cont, 3rd pers sg]
Phrase Within phrase only two kids NP
Det N
[pl] [pl]
V
Category No exchange talk-ed [past]
Figure 1 A simplified account of the Processability Hierarchy illustrating information

exchange required for the insertion of English morphemes.
a developmental hierarchy in syntax and morphology for all languages. We

have demonstrated that the hierarchy can be applied to many target langu-
ages (TLs) by careful analysis of those TLs in terms of specific processing
aspects formalized in LFG. This implies that, in the PT framework, SLA
research starts with theory, and predictions derived from the theory are tested
empirically.
The developmental hierarchy is the centerpiece of PT. It has two aspects:
(1) transfer of information within and between constituents and (2) a number of
mapping processes that connect the semantic roles, constituent structure, and
grammatical functions of the constituents in a sentence.
A simplified account of the processability hierarchy (with a focus on con-
stituent structure) is given in Figure 1, where three examples of levels of
constituent structures are listed in the left-hand column. The second column
specifies the type of information transfer possible at each level. Morphologi-
cal structures for English L2 are given in the next column to exemplify types
of structures that are compatible with the processing operations at each level,
and the information transfer involved in the generation of these structures is il-
lustrated in the column on the right-hand side. For instance, for the noun phrase
“two kids” (e.g., in the sentence “he has two kids”) the information “plural”
only has to be exchanged between the determiner and the noun. In other words,
this information does not need to be exchanged with any other constituent in
the sentence. In this article, I will not elaborate the specific features of the
developmental stages further because they are not central to my argument here.
However, details of the morphological, syntactic, and functional aspects of L2
development are available in multiple texts (e.g., Pienemann, 2007; Pienemann
& Keßler, 2011, 2012). It is crucial to bear in mind that, as an approach to
SLA, PT can account for entire systems of morphosyntax rather than isolated
morphosyntactic aspects.
Currently, all applications of the universal PT processing hierarchy are
based on six overall levels that can be operationalized for a large number of
morphological and syntactic structures for any language. There is no standard
time a learner requires to traverse these levels. Primary school students have
been demonstrated to reach level 2/3 of this framework (on average) by the
end of 2 years of English L2 teaching (Lenzing, 2015). University students (of
German) typically reach level 4/5 in their spontaneous language production
after 1 year of study (see Pienemann, 1998a).
Applying the PT hierarchy to a specific TL will not result in all grammatical
features of the TL being lined up in a tight sequence like pearls on a string—
reflecting a hierarchical step from one feature to the next. Arrangements other
than a tight linear sequence are reflected in multiple aspects of the model. One
aspect that challenges the notion of a single linear sequence is that several
features may be placed at the same level of the hierarchy. Another aspect is that
learners may develop their own interlanguage versions of the TL—introducing
features that are neither part of the L1 nor of the L2, but these features may
be found to operate at the same level of processing as other features. As a
result, learners may progress through universal levels of acquisition (in terms
of processing), yet the shape of their interlanguages at any one stage may vary.
In other words, the presence of different linguistic features at any one level
of the PT hierarchy gives rise to interlanguage variability as defined in this
approach. On the one hand the PT hierarchy constrains learners’ grammars
according to the stage of their processing capacity. On the other hand the
nature of the formalization of the processing capacity allows for sufficient
structural leeway for L2 learners to opt for different solutions of language
learning problems. For instance, before learners are able to place “do” in
second position (a level 5 operation) they may produce Wh-questions either
by relying on canonical word order (e.g., *Why he go home?) or by deleting
the subject pronoun (e.g., *Why go home?). In both cases the learner opts
for a solution that avoids the processing of “do” in a position preceding the
grammatical subject.

D
E T1 T2
V
E
L
O
P
M
E
N
VARIATION
T
Figure 2 Different developmental trajectories.
Both structures illustrated in the above example are processable at level 3

and above (see Pienemann, 1998a, pp. 239 and following for detail). In conse-
quence, this example illustrates the explanatory power of PT for the variational
dimension, as its generic psycholinguistic framework delineates precisely
which structural options are available to learners who do not have level
5 processing procedures at their disposal. In other words, all structural
elements of a developmental trajectory permit a range of solutions of the
structural problem using developmentally earlier processing procedures, and
this structural leeway gives rise to variable solutions of structural problems for
any given level of processing and for every structural learning problem (see
Pienemann, 1998a, for further details).
The different developmental trajectories with common underlying levels
of processability are illustrated in Figure 2—with T1 and T2 representing
different varieties of the interlanguage emerging as the learners progress along
their different trajectories.
Summing up, the processability hierarchy determines universal levels of
processability for specified morphological and syntactic features, which serve to
predict developmental trajectories of L2s through a shared set of common stages
for these features. At the same time, many other linguistic features of the inter-
language are available at many different developmental stages and take up an
orderly position within a nondevelopmental (variational) dimension, quite sim-
ilar to linguistic variation in fully developed languages. As will be shown in the
following section, all of this is part of the dynamics of the acquisition process.
PT as a Framework for Dynamic Linguistic Systems

PT includes a refined approach to the theory–data interface, which is based
on an extended tradition in linguistics that views language as a dynamic
system. This dynamic linguistic framework utilizes concepts developed by

Bailey (1973), DeCamp (1973), and Labov (1972a, 1972b). These authors
developed approaches to the description and representation of language that
account for the variability of language over time, region, and social strata,
viewing varieties of a language as interconnected systems. Some of the basic
descriptive aspects of this approach to language acquisition were developed by
Meisel et al. (1981) as part of what is known as the Multidimensional Model
(MM). This approach was extended in Pienemann (1998a) to form the de-
scriptive component of PT. The dynamic linguistic approach inherent in PT
utilizes implicational scaling inspired by DeCamp’s work in order to model
linguistic development, and it utilizes probabilistic weights for linguistic struc-
tures inspired by Labov’s work in order to model optionality and aspects of
variability—thus adding a probabilistic component to PT.
The dynamic descriptive component of PT relies on a clear and explicit
requirement that distributional analyses are carried out for all corpus data—
with linguistic contexts analyzed in relation to the linguistic hypotheses being
tested and the form–function relationships that these contexts entail. In PT’s
descriptive component, longitudinal and cross-sectional data are subjected to
implicational analysis, which works back from the data to claims about re-
lationships, and the probability of rule application is specified for linguistic
contexts as a result of this analysis.
A specific acquisition criterion is part of this dynamic linguistic approach
(Meisel et al., 1981). This is the emergence criterion, which is used to deter-
mine the point of acquisition of a specific linguistic structure in L2 development.
Building on Meisel et al.’s (1981) approach, I demonstrated that specifying the
point of emergence of a specific interlanguage structure—based on a careful dis-
tributional analysis of a linguistic corpus—is far less arbitrary than applying an
accuracy criterion to the use of TL morphemes (Pienemann, 1998a). Consider
Figure 3 where three fictitious curves have been placed in a coordinate sys-
tem with the horizontal axis representing time and the vertical axis representing
grammatical accuracy. Each curve (a, b, and c) refers to a different grammatical
feature. Note that a, b, and c have different gradients—as they would in real data.
It is easy to see from Figure 3 that different quantitative acquisition criteria
will yield different orders of acquisition: For the 50% criterion the order is
c-b-a; For the 90% criterion the order is c-a-b. The order depends on the
unpredictable gradient of the curve and the arbitrary choice of the quantitative
acquisition criterion. In contrast, the point of emergence is not subject to such
unpredictability, as the point of emergence is fixed for all linguistic features,
using the same steps in the distributional analysis. Focusing on the point of

c a b
100%
50% c b a
0%
a b c time
Figure 3 Accuracy and developmental trajectories, reprinted from Pienemann (1998a).
emergence is motivated by the desire to model the learner’s in-principle capacity

to process the corresponding procedure and is, therefore, not arbitrary in the
way that 50% or 70% or 90% are all arbitrary in the context of SLA (see
Andersen, 1977).
The dynamic linguistic approach inherent in PT has been utilized in a large
number of empirical studies of SLA (see Pienemann, 2005, for an overview),
most of which are based on large corpora of natural or elicited interlanguage
discourse. The operationalization of this dynamic approach follows logically
from its goal of accounting for the way in which the TL grammar unfolds as the
learner’s interlanguage processing capacity develops. The focus is on emerging
interlanguage forms. The emergence criterion avoids a one-to-one comparison
with the TL not only in terms of accuracy but also in terms of TL forms, because
it is based on an analysis of form in relation to function in the interlanguage.
Table 1 illustrates this approach. It shows the percentage of occurrence of
the morphemes –s, 0 (i.e., no morpheme) and –x (unspecified morpheme) on
nouns in two different functional contexts, plural and singular.
Table 1 Form-function analysis
N-s N-0 N-x
Plural context 40 50 10
Singular context 10 80 10

For an imaginary L2 sample, Table 1 shows that in 40% of the plural

contexts the morpheme –s (on nouns) is used, in 50% the zero morpheme
is used, and in 10% an unspecified morpheme (e.g., -x) is used. For the
singular context the corresponding figures are 10%, 80%, and 10%. The 40%
use of plural –s would not satisfy most accuracy criteria. However, Table 1
shows that the morpheme –s is used far more often in plural contexts than
in singular contexts, and this distribution suggests that there is a regularity
in the form–function relationship between plural and –s in the sample that
needs to be acknowledged and theorized. A full distributional analysis would
also include a list of all nouns and of all morphemes contained in the sample.
The emergence criterion requires the same noun to be used both with the
plural morpheme in plural contexts and with the singular morpheme (zero) in
singular contexts and the percentage of usage associated with plural contexts
to be significantly higher than in the other contexts (see Pienemann, 1998a).
Ensuring that a morpheme appears with different lexical items and that a given
item is also used with a different morpheme is one measure to ensure that
formulaic chunks are not counted as instances of morpheme insertion.
It goes without saying that the emergence criterion and distributional anal-
ysis are tools of the linguist and totally separate from the learner’s intuitive
discovery procedures that he or she employs in order to unravel the structure
of the TL.
This approach to distributional analysis is a standard tool in field linguistics.
In the context of SLA it turned out that for morphology the emergence criterion
needs to be applied in the context of a distributional analysis of the type
summarized above (see Pienemann, 1998a). This approach builds from the
individual to the group. It is never assumed that groups describe individuals.
Based in the theoretical framework, the empirical question is consistently asked
of whether there are patterns that go beyond individuals.
Hulstijn (2015) advocates a different approach to the analysis of
interlanguage data:
A good example of the type of research I advocate is the study conducted

by Klein Gunnewiek (2000) [ . . . ] Using multilevel analyses
(distinguishing target structure, time of task, type of task, and individual
learners as independent variables), the researcher was able to investigate
the question of whether a common developmental pattern could be
observed despite minor individual differences. The development of the
five structures hardly seemed interdependent and no stage-like theory was
supported by the data. (2015, p. 226)

As Hulstijn (2015) indicates, Klein Gunnewiek (2000) was unable to iden-

tify developmental patterns using quantitative methods based on average cor-
rectness scores and group mean scores. As I argued above, this demonstrates the
weakness of the analytical approach rather than constituting empirical evidence
of the absence of developmental patterns.
Summing up, the emergence criterion can pick up regularities in developing
interlanguages far more systematically than accuracy-based criteria and when
combined with repeated distributional analyses it describes the interlanguage
in its own dynamic terms—starting at the point of emergence and from there
all the way to ultimate attainment.
Generative Entrenchment: The Logical Problem and the Initial

State
A dynamic view of language acquisition requires a view of what is being ac-
quired, how the learning task is constrained, and how a learner progresses from
one point to the next. As I pointed out above, PT is built on the premise that
learnability is a logico-mathematical problem but crucially that “the logico-
mathematical hypothesis space is further constrained by the architecture of
human language processing” (Pienemann, 1998a, p. 1). This claim is signifi-
cantly different from claims made in innatist approaches to language. In PT,
learners are constrained in entertaining hypotheses about the structure of the
TL by what they can process. Hence the focus is on the effect of processing
constraints on possible structural hypotheses rather than on access to universal
principles of language (cf. White, 2003).
PT makes minimal assumptions about innate linguistic structures. Rather
than assuming sets of universal principles, it assumes only that the basic notion
of constituency and the one-to-one mapping of semantic roles (such as agent,
patient, etc.) is a given, and all other formal aspects of grammar follow from this.
To explain the highly constrained nature of learner language with this minimal
assumption about the initial state, PT has recourse to two sets of constraints:
(1) constraints on human language processing and (2) a mathematical aspect of
the dynamics of language acquisition processes. In Pienemann (1998a, 1998b,
1998c), I have shown that earlier structural decisions influence the later devel-
opment of the interlanguage system. This percolation of structural properties
in developmental processes is a familiar phenomenon in science and philoso-
phy and has been termed “generative entrenchment” by Wimsatt (1986, 1999).
A prime example is embryonic development where sections of the fertilized

egg take on more and more specialized structures (e.g., Coen & Carpenter,
1993; Wolpert, 1992). When things go wrong in embryonic development, early
problems have a more widespread effect than later problems. This implies that
earlier structural decisions are carried forward in the developmental process.
In other words, structures that have been developed earlier do not need to be
developed again when the structure of the system is further refined. This conser-
vative behavior of dynamic systems amounts to a computational saving when
compared with a system in which all structural features have to be developed
anew for every structural change. Wimsatt (1986, 1999) demonstrated that the
computational advantage of generative entrenchment is considerable and thus
limits the amount of information that needs to be given in the initial state. He
further demonstrated that generative entrenchment is active in a wide range of
dynamic processes, from population development to genetic processes.
In Pienemann (1998a, 1998b), I have applied this framework to L1 and
L2 development, and I have demonstrated that the differential developmental
trajectories found in the acquisition of German as an L1 and as an L2 can be
explained on the basis of the different initial hypotheses that learners make in
the two types of acquisition processes (SOV for German as L1 and SVO for
German as L2). I have further demonstrated that all structural features found
at every level of L1 and L2 development follow from the initial structural
hypothesis and that at all stages of development the interlanguage systems
are in line with the constraints imposed on the system by the processability
hierarchy.
It should be clear from this brief summary that PT implies a very limited
position on innateness, relying on a minimum of innate linguistic knowledge—
namely the notion of constituency and the one-to-one mapping of semantic
roles onto the corresponding constituents. All other formal aspects of language
are assumed to develop out of the engagement with the language environment
that is shaped by the initial hypotheses and within the narrow space remaining
through constraints modeled by PT and generative entrenchment. This stance on
innateness differs substantially from the assumptions of nativist SLA theorists,
who assume a highly explicit initial state specified by universal grammar (see
White, 2003).
The Role of the L1

PT contains an explicit position on the role of L1 transfer, known as the
Developmentally Moderated Transfer Hypothesis (DMTH; Pienemann, 1998a,

1998b; Pienemann et al., 2005). The DMTH is a component of PT (see also

Pienemann & Keßler, 2011). The underlying logic of the DMTH is that transfer
is constrained by processability, in particular by the capacity of the L2 learner’s
language processor, whose development provides the underpinning capacity
for the L2 learner’s stage of acquisition (Pienemann, 1998a, 2005). This means
that only those features can be transferred which can be processed within the
current capacities of the L2 processor; in other words, the L2 learners can only
transfer features from the L1 when they are developmentally ready to acquire
them (see Pienemann et al., 2005, p. 85).
Pienemann, Di Biase, Kawaguchi, and Håkansson (2005) review several
extensive studies on L1 and L2 transfer that include a wide range of typolog-
ically different languages supporting the DMTH. Lenzing (2013) provides an
up-to-date review of research on L1 and L2 transfer and of the theory debate
behind it. She also demonstrates empirically that chunk learning occurs along-
side rule-based learning and that one needs to differentiate between the two
when testing predictions of theories of SLA.
PT, the MM, Strategies, and the Predictive Framework

The foundation of PT was set out in Pienemann (1998a). Many publications by
an international network of researchers have followed. In the 1998a publication
as well as in other publications on PT (e.g., Pienemann, 2005) I sketched out the
history of ideas that led up to the construction of PT. However, this may have
resulted in a blurred perception of ideas preceding PT and concepts contained
in it. Therefore, it may be useful to point out the main differences between PT
and preceding frameworks.
The MM (Meisel et al., 1981) is a purely descriptive framework. Its main
features are the emergence criterion and the differentiation of two dimensions
in SLA (development and variation). The MM is operationalized using aspects
of dynamic linguistic paradigms as explained above.
The MM was complemented by Clahsen’s (1984) strategies approach (based
inter alia on Bever, 1970). This approach was intended to explain the devel-
opmental word order patterns that were found for German as L2 using three
processing strategies. The main idea behind the strategies approach was that
processing strategies constrain movement transformations.
The Predictive Framework (Pienemann & Johnston, 1987a2 ) took both sets
of ideas one step further and applied them to English word order and morphol-
ogy. This was conceptualized about 30 years ago in the 1980s. This framework

was not published because I soon realized that it had severe limitations, as it
was based on transformational assumptions, had no typological or psychologi-
cal plausibility, and was limited to a very narrow linguistic domain.
To overcome these limitations, I developed PT, which is based on an ex-
plicit psycholinguistic account of language processing and an explicit and
typologically plausible linguistic theory (i.e., LFG). In this way, typological
and psychological plausibility can be achieved, and the linguistic domain can
be extended to any phenomenon that can be captured by LFG. At the same
time, all predictions can be fully operationalized.
In other words, the creation of PT marks a complete turnaround in the
conceptualization of the explanatory framework. PT has no transformational
connection; it is not a limited explanatory fragment. Instead, it is based on the
interplay between a processing theory and a theory of linguistic knowledge, is
fully explicit, and applies to the full range of phenomena captured by the inter-
acting theories of language processing and linguistic knowledge. As mentioned
above, PT can account for entire systems of morphosyntax rather than isolated
morphosyntactic aspects.
Teachability
The Teachability Hypothesis (Pienemann, 1984) was put forward long before
PT was conceptualized. Basically, it assumes that the effect of teaching inter-
vention is constrained by the learner’s current state of development. When the
Teachability Hypothesis was put forward in the 1980s, it was based on two basic
premises: (1) natural SLA and formal SLA are not fundamentally different and
(2) processing strategies (as assumed at the time and therefore before PT was
formalized) are implicationally ordered. The logical conclusions were: (1) that
learners cannot circumnavigate the next developmental state (through formal
instruction) and (2) that formal instruction may be beneficial if timed correctly
in developmental terms—assuming a weak interface between procedural and
declarative knowledge. It goes without saying that all of this applies to sponta-
neous spoken language under normal time constraints. The latter concept was a
key point of the strategies approach (Clahsen, 1984) on which the Teachability
Hypothesis was based.
The Teachability Hypothesis has been supported by a range of empir-
ical studies which demonstrate that formal L2 learners follow the same
developmental trajectories as natural learners even if the focus of formal
teaching is on entirely different aspects of the language (e.g., Pienemann,
1984, 1989).

As described above, PT is a formalized psycholinguistic framework that

formally delineates developmental trajectories of L1s and L2s and that can ex-
plain several core aspects of the logical problem of language acquisition. When
PT was conceptualized (Pienemann, 1998a), I pointed out that the Teachability
Hypothesis could be integrated into this new framework. However, it is not
a necessary component of PT, and it does not follow from PT. The Teacha-
bility Hypothesis is therefore not a corollary of PT. Instead, the Teachability
Hypothesis can be formalized within PT. Statements about the effect or non-
effect of formal intervention are applications of the theory rather than central
to the framing of the theory. Nevertheless, since the formulation of PT, a num-
ber of studies have tested the Teachability Hypothesis and found it to have
high explanatory value (for further details, see Pienemann, 1998a, pp. 254 and
following).
Zhang and Lantolf (2015) address the Teachability Hypothesis specifically.
Due to space constraints, I can comment only on selected issues raised by
Zhang and Lantolf.
Zhang and Lantolf (2015) claim that, “According to Pienemann (1987)
PT and its corollary, TH, were inspired by Piaget’s developmental theory”
(p. 154). Pienemann (1987) is the article in which the Predictive Framework
is summarized. Obviously, Zhang and Lantolf lumped PT together with the
Predictive Framework, although both are based on very different premises
and are separated by more than a decade. Also, as I pointed out above, the
Teachability Hypothesis is not a corollary of PT.
Zhang and Lantolf (2015) also claim that “The TH allows us to compare
the claims of Piaget [...] and Vygotsky [...] regarding the relationship between
instruction and development, with specific focus on the development of L2 pro-
cessing ability” (p. 155). This claim is untenable because the only connection
between Piaget’s work and the TH is “the implicational nature of processing
prerequisites” (Pienemann, 1987, p. 92). However, at the time the implicational
relationship was operationalized using the strategies approach and not any com-
ponent of Piaget’s theory (e.g., “adaptation” or “accommodation”).3 In their
article, Zhang and Lantolf quote the full extent of my reference to Piaget from
my 1987 publication. In this 30-page paper, Piaget’s work is referred to merely
in a single paragraph (Pienemann, 1987, p. 92). This was long before PT was
conceptualized. In contrast to the strategy framework underpinning the Teach-
ability Hypothesis, PT is based on a coherent psycholinguistic framework that
includes, among many other things, an implicational processing hierarchy—
albeit motivated in quite a different manner from Piaget’s notion. Therefore, it
does not enable a comparison between Piaget’s and Vygotsky’s claims.

The second point concerns Zhang and Lantolf ’s (2015) data elicitation, data
analysis, and the conclusions drawn from them. They claim that they “used the
same spontaneous elicitation procedures and emergence criteria utilized by
Pienemann and his colleagues in their research on PT and TH” (p. 174). How-
ever, they use data from elicited imitation and spontaneous speech production
and amalgamate these data for the purpose of their study. This means that they
are not comparing like with like. Pienemann, Keßler, and Lenzing (2013) have
demonstrated experimentally that L2 learners perform at a significantly higher
level in elicited imitation than in spontaneous speech production. Therefore,
data obtained through elicited imitation cannot be compared one to one with
spontaneous speech production data. In terms of language processing, the two
types of data tap into different psycholinguistic mechanisms.
The third point concerns the nature of the features investigated. As Zhang
and Lantolf (2015) note, the topicalization of adjuncts or objects is not oblig-
atory in Chinese. In other words, it does not follow constituent structure con-
straints (like do-insertion in English) but discourse-pragmatic constraints. One
may be inclined to assume that the presence of topicalized constituents indi-
cates the acquisition of “topicalization.” The problem is that, in interpreting
their data, Zhang and Lantolf apply the reverse logic, that is, if learners do
not topicalize, they interpret this as evidence that they are unable to do it.
In Pienemann (1998a, pp. 145 and following), I demonstrated that there is a
difference between: (1) no evidence, (2) insufficient evidence, (3) evidence
for the nonapplication of a rule, and (4) evidence for rule application. The
case of nonapplication of adjunct topicalization that is presented by Zhang and
Lantolf falls into the category of no evidence. This means we cannot draw
any conclusions from this observation about the learners’ ability to topicalize
adjuncts. And this is the core of the empirical evidence against the Teachability
Hypothesis presented by Zhang and Lantolf.
As Anke Lenzing pointed out to Zhang and Lantolf (personal communica-
tion, March 24, 2014), the problem of the lack of evidence could be overcome
by including other Chinese structures from level 3 of the PT hierarchy. How-
ever, the authors dismiss this option because they claim that Pienemann, M.,
Di Biase, B., & Kawaguchi, S. (2005) assume that “for any given stage mor-
phology is likely to emerge before syntax and it may bootstrap syntax” (Zhang
& Lantolf, 2015, p. 173) and that this assumption had been called into ques-
tion. I need to state here that Pienemann et al. (2005) do not make such an
assumption and that, within PT, there is a theoretically justified link between
syntax and morphology for levels 1–3 of the hierarchy that could be explored
in order to resolve this issue.

Comparing PT and Dynamic Systems Theory (DST)

On the surface it may seem that de Bot, Lowie, and Verspoor’s (2007) DST
approach is based partly on claims similar to those assumed in the PT frame-
work. The seeming similarity starts with the term “dynamic.” As I pointed
out above, PT is based on a tradition of describing and accounting for dy-
namic linguistic systems. In addition, PT builds upon generative entrenchment
as a notion that captures a key aspect of developmental dynamics. However,
this coincidence in the use of terminology disguises some deeper differences.
de Bot et al. (2007, p. 8), as well as Lowie and Verspoor (2015), assume that all
variables affecting SLA are interconnected and that this interconnectedness is
in no way constrained.4 In contrast, the dynamic processes of PT are based on
the assumption that there are several layers of constraints that operate on SLA,
specifically at the cognitive level and within the dynamics of the developmental
processes. De Bot et al. (2007) also state that
[g]rowth is conceived of as an iterative process, which means that the

present level of development depends critically on the previous level of
development (Van Geert, 1994), and variation is not seen as NOISE but
as an inherent property of a changing system. (p. 14)
Again, these two statements are reminiscent of features of PT. However,

de Bot et al. have something quite different in mind. The conditioning of the
following levels by the previous levels appears to be conceptualized merely as a
metaphor inspired by de Bot et al.’s understanding of the modeling of dynamic
systems in science and the role played by the initial state. This metaphorical
conceptualization has not been operationalized. Rather, it is seen as being
manifested in the various descriptions of the acquisition of diverse linguistic
features. For the claim to be able to be tested empirically, the distinction between
“NOISE” and “an inherent property” would have to be operationalized and the
consequences of the operationalization tested. It was because of this lack of
operationalization that I inquired at the Language Learning Roundtable 2013
in Amsterdam how the DST approach presented by de Bot et al. (2007) could
be falsified. I pointed out that their approach would not have the status of a
theory if it could not be falsified. In their article in this Special Issue, Lowie and
Verspoor (2015) state that “DST is not very different from other approaches
in this sense. UG and Behaviorism cannot be falsified either” (p. 80). This
reply does not resolve the issue of operationalization and testability. Viewing
other approaches as not operationalized does not remove the requirement that

a theory be empirically testable. This applies to the above version of DST as

much as to any other work that claims the status of a theory.
PT has been fully operationalized within a psycholinguistic framework,
and all constraints on processability are readily testable and thus falsifiable.
In fact, they come with a wealth of empirical support. PT has been tested
extensively using large sets of data to trace developmental trajectories and vari-
ational features in a wide range of L2s, including Arabic (Mansouri, 2005), En-
glish (Pienemann, 1998a), German (Pienemann, 1998a), Swedish (Håkansson
& Norrby, 2010; Glahn et al., 2001; Salameh, Håkansson, & Nettelbladt.,
2004), Chinese (Zhang, 2005), Italian (Di Biase, Bettoni, & Nuzzo, 2009),
Japanese (Kawaguchi, 2005), and Turkish (Özdemir, 2004). For all these L2s,
developmental trajectories have been identified, and they are supported by full
distributional analyses conducted as outlined above—applying the emergence
criterion. PT readily accounts for variation in accuracy as part of the varia-
tional dimension of Hypothesis Space, which complements the developmental
dimension (see Lenzing, 2015). This data set is available for the alternative
operationalizations of terms within the DST framework to be tested.
The DST position espoused by de Bot et al. (2007) and Lowie and Verspoor
(2015) repeatedly insists that variation is an inherent property of a changing
system. This is also a form of words that could equally well be used in a PT
context, albeit with a rather different meaning. In the PT context, interlan-
guage variation is conceptualized in the tradition of variationist linguistics—
referring to alternative manifestations of a linguistic system. In the context of
de Bot et al.’s version of DST, variation is illustrated mostly by variable accuracy
rates in morpheme order studies and as such reflects a system of right/wrong
rather than variation within underlying systems such as alternative manifesta-
tions of, say, plurality or grammatical person. Despite the similar terminology,
the constructs that are being referred to in the different frameworks are signifi-
cantly different. Therefore, in attempting to understand the relationship between
the two frameworks, we must go beyond the surface of the words used. When
we do, we discover both similarities and more precisely definable differences
in claims.
The fundamental difference in the approaches taken by PT and that taken
by de Bot et al. (2007) is particularly visible in the notion of development,
which is conceptualized by de Bot and colleagues in terms of frequency of
insertion of morphemes in learner utterances as evidenced in their definition of
grammatical development: “grammatical development as measured by MLU
data” (de Bot et al., 2007, p. 13). In their perception, the graphs used to
depict average sentence length in L2 learners represent “developmental curves”

(p. 16). In other words, they equate changes in frequency with development,
which means that any change can be referred to as development, in their
overly general sense. This contrasts with the PT perspective. As I showed
above, frequency and accuracy rates are invalid measures of development when
development is understood as increased complexity of the overall system (for
further detail, see Pienemann, 1998a).
In their 2007 article and in other publications (including Lowie & Verspoor,
2015), de Bot and colleagues make the point that group mean scores—as used
in morpheme order studies (they quote Krashen, 1977)—obscure the detail of
individual language use. It needs to be noted here that this observation was made
repeatedly several decades ago (e.g., Andersen, 1977; Meisel et al., 1981) by
contemporaries of the morpheme order researchers. And it was for this reason
that more suitable research designs were developed over the years. Nevertheless,
de Bot and colleagues have amassed large amounts of data which they and
others (e.g., Murakami, 2013) analyze using the morpheme order approach
with the intention of demonstrating that the linear order that was claimed to
follow from a morpheme order analysis is not substantiated at the level of the
individual learner. The keen observer of SLA studies is not surprised. Despite
their groundbreaking place in the history of SLA, morpheme order studies do
not stand up to the requirements of current research.
Nevertheless, de Bot and colleagues (2007) go on to conclude from the
observed concealment of individual accuracy differences by group mean scores
in morpheme order studies (Murakami, 2013) that “languages do not develop
according to a predetermined sequence” (Lowie, 2013, p. 1809), and Lowie and
Verspoor (2015) claim that “predetermined stages in language development are
an artifact of the method of investigation” (p. 79).
However, this bold conclusion about the total absence of developmental
regularities in SLA does not follow logically from the observed individual
differences in accuracy. In their framing of the issue, de Bot and colleagues
(2007) equate the level of accuracy in learner samples with the learners’ level
of acquisition. I have shown above that this is not warranted. Instead, the
notions of acquisition, variation, and development all need to be carefully op-
erationalized before empirical data can be used to test hypotheses, such as their
no-developmental regularities hypothesis. De Bot and colleagues offer a defini-
tion of acquisition (increasing or ultimate mastery), but a view of development
(change) that does not permit shared developmental features to be distinguished
from variation. PT contains an alternative operationalized descriptive frame-
work, as outlined above (acquisition = emergence; development = predictable
increases in complexity of processing prerequisites; variation = alternative

manifestations of an underlying system that can be described using the PT

apparatus). Because the terms that are being used are not operationalized in
the same way, it is dangerous to compare the constructs as if they were the
same. Let me add that massive evidence of fixed stages of L2 acquisition has
been produced in a large number of research projects based on longitudinal
data using research frameworks similar to the one outlined in this paper (cf.
Pienemann, 2005).
Finally, let me clarify a misunderstanding. Lowie and Verspoor (2015)
state that “[PT] assumes a genetically programmed sequence of predetermined
stages” (p. 82). It should be clear from what I have said above that PT does
not make recourse to a genetic program. Quite the opposite. As the name of
the theory indicates, the assumption underlying PT is that language processing
determines the development of L2 grammar—complemented by the dynamics
inherent in the developmental process.
Processing in Emergentism and PT

There is a further terminological distinction that needs to be clarified so that
the positions entailed by the approach espoused by de Bot et al. (2007) and
Lowie and Verspoor (2015) and PT can be understood. I have outlined above
the descriptive role of emergence in defining a powerful view of acquisition. It
is, however, important to differentiate this use of the term emergence from the
larger epistemology of emergentism. In language acquisition research, emer-
gentism is an approach to epistemology that assumes that complex linguistic
forms emerge from less complex forms without recourse to innate linguistic
knowledge (Bates, 1979). Emergentism is an important feature of de Bot et al.’s
arguments for the nature of learning and against the idea that there needs to be a
formal representation of language in the learner’s mind. O’Grady (2012, 2015)
adds language processing to this view. He proposes that “language acquisition
is a secondary effect of processing amelioration: attempts by the processor
to facilitate its own functioning by developing routines of particular sorts”
(2012, p. 116).
On the surface, O’Grady’s (2015) position appears similar to the basic
stance implied in the PT framework. As in his 2013 presentation at EuroSLA,
he is concerned with processing as a key source for explaining the why
behind the observed course of language development. However, the internal
logic of the two approaches is quite different. O’Grady’s approach is based
on the claim that “learners are able to infer at least some form-meaning

mappings without prior knowledge of a language’s syntax” (2012, p. 118).

This is his version of the bootstrapping assumption. Similar assumptions were
made, for instance, by Bever (1970), Pinker (1984), and Slobin (1982). O’Grady
illustrates his view with the development of word order in English as a first
language. His illustration of first language word order development is based
on a hypothetical learner who has no previous knowledge of syntax (neither
learned nor innate) and who “is familiar with a couple of noun-type words
(say ‘Charlie’ and ‘ball’) and a verb-type word (‘kick’)” (2012, p. 118). When
this learner hears the sentence Charlie kicked the ball in a meaningful context,
O’Grady assumes that,
[u]nder such circumstances, the learner should be able to map the input
onto meaning by proceeding one word at a time in the following manner.
[...]
1. Interpret the first nominal (“Charlie”)

CHARLIE
2. Access the meaning of the transitive verb “kick”; find its agent argument
(“Charlie”) to the left.
KICK <agent: CHARLIE>
3. Interpret the nominal to the right (“the ball”); treat it as the verb’s patient
argument.
KICK <agent: CHARLIE; patient: BALL>
As illustrated here, the end result of these operations is the creation of a

link, or mapping, between a particular word order (“Charlie kicked the
ball”) and a particular meaning involving a kicking event. (2012, p. 118)
In other words, O’Grady (2012) assumes that L1 learners are capable of

mapping words onto semantic roles (such as agent, patient) based on their posi-
tion in the sentence and the category of the respective words. At the same time
he assumes that this capacity of L1 learners does not require any kind of linguis-
tic knowledge. However, additionally and in a potentially contradictory manner,
O’Grady (2011) assumes that the syntactic structures that emerge from such
mapping processes (i.e., SVO word order) constitute symbolic representations.
PT does not make the assumption that mental representations can emerge
from no linguistic knowledge. The mapping processes implied in PT are
those spelled out by the formalism of LFG and described in Lenzing (2015).

The basic principle behind this is similar to that implied in O’Grady’s (2012)
example, except that the processes involved are more formalized in LFG, and
they affect more than the lexicon and semantic roles. PT and O’Grady’s (2012)
approach are fundamentally different with respect to the following points:
1. The innate basis: PT assumes basic notions of constituency and the one-to-
one mapping of semantic roles to be present in the initial state. All other
formal aspects of language develop from this. Also, the basic notion of
predicate-argument structure is assumed to be part of the initial state.
2. PT contains a hierarchy of mapping processes. This hierarchy predicts ex-
plicitly the sequence in which mapping processes and the required grammar
develop in the learner.
In other words, PT contains a minimal, but explicit set of assumptions about

the initial state, and formal aspects of grammar are assumed to develop out of
the minimal components contained in the initial state—shaped by processing
constraints and generative entrenchment.
In my view, O’Grady’s (2011, 2015) assumption of a syntax-free initial
state is at loggerheads with his mapping processes, which rely on the knowl-
edge/recognition of formal word classes and positions in sentences (left and
right of the verb). Also, his mapping processes rely on the recognition of seman-
tic roles and predicate-argument structures. Although these are not syntactic in
nature, they are nevertheless specific to language. Therefore they either need to
be assumed to be present in the initial state, or it must be shown how they can
evolve out of general cognitive structures. It is for these reasons that PT was
designed with the assumption of a minimal set of formal assumptions about
the initial state that can serve as the seeds for further structural development.
Here the issues are not just interpretations of terms, but alternative positions
on the nature of language processing, with alternative testable claims.
Conclusion
In this article I have tried to show that similar technical terms are being used
differently in different approaches to SLA research and that these terms differ in
the degree to which they are operationalized. This makes it difficult to compare
global statements about such fundamental issues as whatever each researcher
and approach might understand as universal regularities in L2 development.
The focus of PT-based research and many other approaches to SLA devel-
opment is not on factors external to SLA but on the inner mental processes

that drive key aspects of the dynamics of SLA. As an approach to SLA, PT can
account for entire systems of morphosyntax rather than isolated morphosyntac-
tic aspects, and it contains clear and operationalized criteria for developmental
and variational aspects of these L2 systems.
In the rich literature available on PT-based SLA research, there is over-
whelming empirical evidence supporting universal regularities of L2 develop-
ment based on hundreds of informants and thousands of hours of recorded
speech. Any test of theoretical predictions derived from PT needs to be carried
out with explicit acknowledgement of its descriptive and explanatory frame-
work and the associated acquisition criterion. If that is done, there is scope for
productive exchange between tight and focused perspectives such as PT and
perspectives that seek to build connections across multiple linguistic systems.
Final revised version accepted 30 October 2014
Notes
1 The logical problem concerns the issue of where grammatical knowledge originates,
and the developmental problem addresses the question of why there are universal
trajectories of development.
2 The “predictive framework” was summarized briefly in Pienemann and Johnston
(1987b). However, the paper on the predictive framework itself was never published.
3 Apart from everything else, Piaget’s theory had not been applied to language in the
first place.
4 At the same time, de Bot et al. (2007) assume that “Every cognitive system is
constrained by limited resources” (p. 14). However, they do not operationalize this
claim. So it cannot be tested empirically.
References
Andersen, R. (1977). The impoverished state of cross-sectional morphology
acquisition/accuracy methodology (or the leftovers are more nourishing than the
main course). Working Papers on Bilingualism, 14, 49–82.
Bailey, C.-J. N. (1973). Variation and linguistic theory. Arlington, VA: Center for
Applied Linguistics.
Bates, E. (1979). The emergence of symbols. New York: Academic Press.
Berwick, R. C., & Weinberg, A. S. (1984). The grammatical basis of linguistic
performance: Language use and acquisition. Cambridge, MA: MIT Press.
Bever, T. G. (1970). The cognitive basis for linguistic structures. In J. R. Hayes (Ed.),
Cognition and the development of language (pp. 279–362). New York: Wiley.

Bonnesen, M. (2009). The status of the “weaker” language in unbalanced

French/German bilingual language acquisition. Bilingualism: Language and
Cognition, 12, 177–192.
Bresnan, J. (2001). Lexical functional syntax. Malden, MA: Blackwell.
Clahsen, H. (1984). The acquisition of German word order: A test case for cognitive
approaches to second language acquisition. In R. Andersen (Ed.), Second
languages: A cross-linguistic perspective (pp. 219–242). Rowley, MA: Newbury
House.
Coen, E. S., & Carpenter, R. (1993). The metamorphosis of flowers. Plant Cell, 5,
1175–l181.
de Bot, K., Lowie, W., & Verspoor, M. (2007). A dynamic systems theory approach to
second language acquisition. Bilingualism: Language and Cognition, 10, 7–21.
DeCamp, D. (1973). What do implicational scales imply? In C.-J. N. Bailey & R. W.
Shuy (Eds.), New ways of analyzing variation in English (pp. 141–148).
Washington, DC: Georgetown University Press.
Di Biase, B., Bettoni, C., & Nuzzo, E. (2009). Postverbal subject in Italian L2: A
Processability Theory approach. In J.-U. Keßler & D. Keatinge (Eds.), Research in
second language acquisition: Empirical evidence across languages (pp. 153–173).
Newcastle upon Tyne, UK: Cambridge Scholars Publishing.
Donaldson, B. (2011). Left-dislocation in near-native French. Studies in Second
Language Acquisition, 33, 399–432.
Ellis, R. (2015). Researching acquisition sequences: Idealization and de-idealization in
SLA. [Special Issue]. Language Learning, 65, 181–209.
Felix, S. W. (1984). Maturational aspects of Universal Grammar. In A. Davies, C.
Criper, & A. Howatt (Eds.), Interlanguage (pp. 133–161). Edinburgh, UK:
Edinburgh University Press.
Felser, C. (2005). Experimental psycholinguistic approaches to second language
acquisition. Second Language Research, 21, 95–97.
Garcı́a Mayo, M. d. P., & Olaizola, V. (2011). The development of suppletive and
affixal tense and agreement morphemes in the L3 English of Basque-Spanish
bilinguals. Second Language Research, 27, 129–149.
Glahn, E., Håkansson, G., Hammarberg, B., Holmen, A., Hvenekilde, A., & Lund, K.
(2001). Processability in Scandinavian second language acquisition. Studies in
Second Language Acquisition, 23, 389–416.
Håkansson, G., & Norrby, C. (2010). Environmental influence on language
acquisition: Comparing second and foreign language acquisition of Swedish.
Language Learning, 60, 628–650.
Hulstijn, J. H. (2015). Discussion: How different can perspectives on L2 development
be? [Special Issue]. Language Learning, 65, 210–232.
Jaensch, C. (2011). L3 acquisition of German adjectival inflection: A generative
account. Second Language Research, 27, 83–105.

Jia, G., & Fuse, A. (2007). Acquisition of English grammatical morphology by native
mandarin-speaking children and adolescents: Age-related differences. Journal of
Speech, Language, and Hearing Research, 50, 1280–1299.
Kawaguchi, S. (2005). Acquisition of Japanese as a second language. Processability
Theory and Japanese as a Second Language, 8, 83–114.
Klein Gunnewiek, E. A. (2000). Sequenzen und Konsequenzen: Zur Entwicklung
niederländischer Lerner im Deutschen als Fremdsprache [Sequences and
consequences: Development of Dutch learners of German as a foreign language].
Amsterdam: Rodopi.
Krashen, S. (1977). Some issues relating to the monitor model. In H. Brown, C. Yorio,
& R. Crymes (Eds.). On TESOL ’77 (pp. 144–158). Washington, DC: TESOL.
Labov, W. (1972a). Sociolinguistic patterns. Philadelphia: University of Pennsylvania
Press.
Labov, W. (1972b). Language in the inner city. Philadelphia: University of
Pennsylvania Press.
Lenzing, A. (2013). The development of the grammatical system in early second
language acquisition: The multiple constraints hypothesis. Amsterdam: John
Benjamins.
Lenzing, A. (2015). Exploring regularities and dynamic systems in L2 development.
[Special Issue]. Language Learning, 65, 89–122.
Levelt, W. J. M. (1989). Speaking: From intention to articulation. Cambridge, MA:
MIT Press.
Levelt, W. J. M. (Ed.). (1993). Lexical access in speech production. Cambridge, MA:
Blackwell.
Long, M. H. (1990). The least a second language acquisition theory needs to explain.
TESOL Quarterly, 24, 649–66.
Lowie, W. (2013). Dynamic systems theory approaches to second language
acquisition. In C. Chapelle (Ed.), The encyclopedia of applied linguistics (pp.
1806–1813). Malden, MA: Blackwell.
Lowie, W., & Verspoor, M. (2015). Variability and variation in second language
acquisition orders: A dynamic re-evaluation. [Special Issue]. Language Learning,
65, 63–88.
Luk, Z. P., & Shirai, Y. (2009). Is the acquisition order of grammatical morphemes
impervious to L1 knowledge? Evidence from the acquisition of plural -s, articles,
and possessive ’s. Language Learning, 59, 721–754.
Mansouri, F. (2005). Agreement morphology in Arabic as a second language:
Typological features and their processing explanations. In M. Pienemann (Ed.),
Cross-linguistic aspects of processability theory (pp. 117–154). Amsterdam: John
Benjamins.
Meisel, J. M. (ed) (1994). Bilingual first language acquisition: French and German
grammatical development, John Benjamins Publishing Company.

Meisel, J. M. (2011). First and Second Language Acquisition: Parallels and

Differences, Series: Cambridge Textbooks in Linguistics.
Meisel, J. M., Clahsen, H., & Pienemann, M. (1981). On determining developmental
stages in natural second languages acquisition. Studies in Second Language
Acquisition, 3, 109–135.
Murakami, A. (2013). Individual variation and the role of L1 in the L2 development of
English grammatical morphemes: Insights from learner corpora. Unpublished
doctoral dissertation, Cambridge University, Cambridge, UK.
Myles, F. (2005). Interlanguage corpora and second language acquisition research.
Second Language Research, 21, 373–391.
Myles, F., & Towell, R. (Eds.). (2005). The acquisition of French as a second language.
Journal of French Language Studies, 14, 211–232.
O’Grady, W. (2011). Emergentism. In P. Hogan (Ed.), Cambridge encyclopedia of
language sciences (pp. 274–277). Cambridge, UK: Cambridge University Press.
O’Grady, W. (2012). Language acquisition without an acquisition device. Language
Teaching, 45, 116–130.
O’Grady, W. (2015). Processing determinism. [Special Issue]. Language Learning, 65,
6–32.
Özdemir, B. (2004). Language development in Turkish-German bilingual children and
the implications for English as a third language. Unpublished master’s thesis,
University of Paderborn, Paderborn, Germany.
Pienemann, M. (1984). Psychological constraints on the teachability of languages.
Studies in Second Language Acquisition, 6, 186–214.
Pienemann, M. (1987). Determining the influence of instruction on L2 speech
processing. Australian Review of Applied Linguistics, 10, 83–113.
Pienemann, M. (1989). Is language teachable? Psycholinguistic experiments and
hypotheses. Applied Linguistics, 10, 52–79.
Pienemann, M. (1998a). Language processing and second language development:
Processability Theory. Amsterdam: John Benjamins.
Pienemann, M. (1998b). Developmental dynamics in L1 and L2 acquisition:
Processability Theory and generative entrenchment. Bilingualism, 1, 1–20.
Pienemann, M. (1998c). A focus on processing. Bilingualism, 1, 36–38.
Pienemann, M. (Ed.). (2005). Cross-Linguistic aspects of Processability Theory.
Amsterdam: John Benjamins.
Pienemann, M. (2007). Processability theory. In B. VanPatten & J. Williams (Eds.),
Theories in second language acquisition: An introduction (pp. 137–154). Malden,
MA: Blackwell.
Pienemann, M., Di Biase, B., & Kawaguchi, S. (2005). Extending processability
theory. In M. Pienemann (Ed.), Cross-linguistic aspects of processability theory
(pp. 199–251). Amsterdam: John Benjamins.

Pienemann, M., Di Biase, B., Kawaguchi, S., & Håkansson, G. (2005). Processability,
typological distance and L1 transfer. In M. Pienemann (Ed.), Cross-linguistic
aspects of processablity theory (pp. 85–117). Amsterdam: John Benjamins.
Pienemann, M., & Johnston, M. (1987a). A predictive framework of second language
acquisition. Unpublished manuscript, Sydney University, Sydney, Australia.
Pienemann, M., & Johnston, M. (1987b). Factors affecting the development of
language proficiency. In D. Nunan (Ed.), Applying second language acquisition
research (pp. 45–141). Adelaide, Australia: National Curriculum Resource Centre.
Pienemann, M., & Keßler, J.-U. (Eds.). (2011). Studying processability theory:
Introductory textbook. Amsterdam: John Benjamins.
Pienemann, M., & Keßler, J.-U. (2012). Processability theory. In S. Gass & A. Mackey
(Eds.), Handbook of second language acquisition (pp. 228–246). Abingdon, UK:
Routledge.
Pienemann, M., Keßler, J.-U., & Lenzing, A. (2013). Developmentally moderated
transfer and the role of the L2 in L3 acquisition. In A. Flymann Mattsson & C.
Norrby (Eds.), Language acquisition and use in multilingual contexts: Theory and
practice (pp. 142–159). Lund, Sweden: Travaux de L’Institute Linguistique de Lund.
Pinker, S. (1984). Language learnability and language development. Cambridge, MA:
Harvard University Press.
Roberts, L. (2012). Psycholinguistic techniques and resources in second language
acquisition research. Second Language Research, 28, 113–127.
Salameh, E.-K., Håkansson, G., & Nettelbladt, U. (2004). Developmental perspectives
on bilingual Swedish-Arabic children with and without language impairment: A
longitudinal study. International Journal of Language & Communication
Disorders, 39, 65–91.
Schulz, B. (2011). Syntactic creativity in second language English: Wh-scope marking
in Japanese-English interlanguage. Second Language Research, 27, 313–341.
Slobin, D. I. (1982). Universal and particular in the acquisition of language. In E.
Wanner & L. Gleitman (Eds.), Language acquisition: The state of the art (pp.
128–172). Cambridge, UK: Cambridge University Press.
van de Craats, I., & van Hout, R. (2010). Dummy auxiliaries in the second language
acquisition of Moroccan learners of Dutch: Form and function. Second Language
Research, 26, 473–500.
Van Geert, P. (1994). Dynamic systems of development: Change between complexity
and chaos. New York: Harvester.
VanPatten, B., & Williams, J. (Eds.). (2007). Theories in second language acquisition:
An introduction. Mahwah, NJ: Lawrence Erlbaum.
White, L. (2003). Second language acquisition and Universal Grammar. Cambridge,
UK: Cambridge University Press.
Wimsatt, C. W. (1986). Developmental constraints, generative entrenchment, and the
innate-acquired distinction. In P. W. Bechtel (Ed.), Integrating scientific disciplines
(pp. 185–208). Dordrecht, Netherlands: Martinus Nijhoff.

Wimsatt, C. W. (1999). Generativity, entrenchment, evolution, and innateness. In V.

Hardcastle (Ed.), Where biology meets psychology: philosophical essays (pp.
139–179). London: MIT Press.
Wolpert, L. (1992). The shape of things to come. New Scientist, 134(18), 38–42.
Zhang, Y. (2005). Processing and formal instruction in the L2 acquisition of five
Chinese grammatical morphemes. In M. Pienemann (Ed.), Cross-linguistic aspects
of processability theory (pp. 155–178). Amsterdam: John Benjamins.
Zhang, X., & Lantolf, J. P. (2015). Natural or artificial: Is the route of L2 development
teachable? [Special Issue]. Language Learning, 65, 152–180.

Pienemann 2015

Uploaded by

Copyright:

Available Formats

Pienemann 2015

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pienemann 2015

Uploaded by

Copyright:

Available Formats

Language Learning ISSN 0023-8333

An Outline of Processability Theory and Its

Keywords Processability Theory; acquisition order; processing

Language Learning 65:1, March 2015, pp. 123–151 123

The Continued Focus on Interlanguage Development

Language Learning 65:1, March 2015, pp. 123–151 124

basis, as reflected in the collection and increasing accessibility of interlanguage

It appears that not very many empirical studies on acquisition orders in

Indeed, it is hard to find the initial terms “orders and sequences of L2

125 Language Learning 65:1, March 2015, pp. 123–151

While SLA research with a focus on the representation of linguistic knowl-

[L]earners of different ages, with or without instruction, in foreign and

In his contribution, Ellis refers to this statement as “Long’s law.” However,

It becomes very dangerous to claim that X is an established fact or that Y

Language Learning 65:1, March 2015, pp. 123–151 126

The Rationale Behind PT

ignores the fact that this problem has to be solved, not by an

PT was designed to address the developmental problem and later also

127 Language Learning 65:1, March 2015, pp. 123–151

Locus of exchange Example Illustration

Phrase Within phrase only two kids NP

Figure 1 A simplified account of the Processability Hierarchy illustrating information

a developmental hierarchy in syntax and morphology for all languages. We

129 Language Learning 65:1, March 2015, pp. 123–151

Figure 2 Different developmental trajectories.

Both structures illustrated in the above example are processable at level 3

PT as a Framework for Dynamic Linguistic Systems

system. This dynamic linguistic framework utilizes concepts developed by

131 Language Learning 65:1, March 2015, pp. 123–151

Figure 3 Accuracy and developmental trajectories, reprinted from Pienemann (1998a).

emergence is motivated by the desire to model the learner’s in-principle capacity

Table 1 Form-function analysis

N-s N-0 N-x

Language Learning 65:1, March 2015, pp. 123–151 132

For an imaginary L2 sample, Table 1 shows that in 40% of the plural

A good example of the type of research I advocate is the study conducted

133 Language Learning 65:1, March 2015, pp. 123–151

As Hulstijn (2015) indicates, Klein Gunnewiek (2000) was unable to iden-

Generative Entrenchment: The Logical Problem and the Initial

Language Learning 65:1, March 2015, pp. 123–151 134

The Role of the L1

135 Language Learning 65:1, March 2015, pp. 123–151

1998b; Pienemann et al., 2005). The DMTH is a component of PT (see also

PT, the MM, Strategies, and the Predictive Framework

Language Learning 65:1, March 2015, pp. 123–151 136

137 Language Learning 65:1, March 2015, pp. 123–151

As described above, PT is a formalized psycholinguistic framework that

Language Learning 65:1, March 2015, pp. 123–151 138

139 Language Learning 65:1, March 2015, pp. 123–151

Comparing PT and Dynamic Systems Theory (DST)

[g]rowth is conceived of as an iterative process, which means that the

Again, these two statements are reminiscent of features of PT. However,

Language Learning 65:1, March 2015, pp. 123–151 140

a theory be empirically testable. This applies to the above version of DST as

141 Language Learning 65:1, March 2015, pp. 123–151

Language Learning 65:1, March 2015, pp. 123–151 142

manifestations of an underlying system that can be described using the PT

Processing in Emergentism and PT

143 Language Learning 65:1, March 2015, pp. 123–151