What really intrigues me is when you look at languages which have really complex morphology, which English doesn't really have, like Hungarian, Turkish and Russian - I'm going to be looking at how children learn Polish morphology, which is very hard if you've ever tried to learn it as a foreigner!
The theories of what goes on in children's minds are really interesting and hotly debated, and I summarise the two 'biggies' within in this essay. However, the first half of the essay looks more at what morphology is and does in different languages of the world, which of course are amazingly diverse.
Later in the essay I go into the big head-to-head: examining the 'dual-route' and 'single-route' theories, and how these can (or cannot) cater for languages like Polish.
Any questions, comments or criticisms are welcome!
.................................................................................................................................
A cross-linguistic review of inflectional morphology and the dual- / single-route theories of acquisition
Of the various
aspects of language that children acquire during development, one which has
attracted much attention for child language acquisition (CLA) research is
morphology. Definitions vary, though for the purposes of this essay, morphology
is “the branch of
linguistics that deals with words, their internal structure, and how they are
formed” (Aronoff, & Fudeman, 2005, pp. 1-2). This branch is usually broken
down into two sub-systems: inflection, i.e. how a single word concept
(‘lexeme’) can take a number of specific forms based on rules, and word formation,
i.e. production of a new lexeme from an older one, either by derivation or
compounding (Booij, 2007; Spencer, 1993; Stump; 2001). Booij (2007) uses the
idea of a dictionary to explain the two types of morphology; inflections are a
set of words a user is expected to know under a single dictionary entry (e.g. cat implies cats; start implies starting, starts etc. (my examples)), whereas word formation gives rise to
separate entries (e.g. cat-like (adjective)
is a derivation of cat (noun); jump-start is a compound related to start). Regarding children’s acquisition
of morphology, it is inflection which is theoretically more important, as one
of its more pertinent properties distinguishing it from word formation is its
role in the grammar, as summarised by the aphorism “inflectional morphology is
what is relevant to the syntax” (Anderson, 1982, p. 587). The rest of this
essay, therefore, focusses specifically on acquisition of inflectional morphology.
Inflectional
morphology is an almost ubiquitous component of grammar in the world’s
languages, and as one might expect, it exhibits a diverse range of form and
function across and within languages. While English has a relatively
impoverished inflectional morphology, relying more on overt syntactic
structures to mark grammatical relations and thereby convey meaning, other
languages such as Russian have a much larger system of inflections available to
achieve this (Ambridge, & Lieven, 2011); some languages such as Greenlandic
have such rich morphology that a single ‘word’ can convey what could be a full
sentence in English (Kelly, Wigglesworth, Nordlinger, & Blythe, 2014).
Morphological typology will be returned to shortly in more detail; the main
point here is that this wide range of inflectional morphology is something that
all children have the potential to acquire in early life, and one fundamental
aim of the CLA field is to determine how they do this. Following over two
decades of research and debate in this pursuit, two major theories predominate:
the dual-route and single-route accounts (Rowland, 2014).
In
essence, these two rival theories explain how inflectional processes occur in
the adult human brain and how children acquire them. The dual-route model
posits two distinct mechanisms to process regular and irregular word forms
respectively, the first involving a (possibly innate) development of a default
rule to inflect words (e.g. In English: ‘add -s to a noun to make it plural’), and the second storing irregular
forms as a complete word in the mental lexicon (Ambridge, & Lieven, 2011);
this model is also known as the ‘words and rules’ account (e.g. Pinker, 1998).
In contrast, the single-route model rejects the notion of a default
rule-forming facility and instead proposes that all word forms (regular and irregular) are stored and processed in
an associative memory (e.g. Bybee, & Moder, 1983). However, the two
theories as briefly introduced here have undergone much revision and
development since their early initial English-based formulations of the 1980s
and 1990s – the current state of the two accounts will be returned to in detail
shortly.
This essay,
therefore, seeks to address the key question of how children acquire
inflectional morphology by evaluating current dual-route and single-route
accounts, with a special focus on cross-linguistic research.
In order fully
examine the two theories, it is first necessary to take a closer look at the
diverse ways in which inflectional morphology works in the world’s languages. Inflection
performs a considerable range of communicative functions in language, and its
possible forms are even more numerous. Languages can be grouped into four
categories according to their overall inflectional properties (Spencer, 1993): isolating languages, such as Mandarin Chinese,
do not use inflections at all, grammatically relying on syntax; agglutinating languages, such as
Turkish, concatenate stems with (potentially multiple) morphemes in which a
single morpheme has a single meaning or function; inflectional or fusional languages,
such as Latin, are similar to agglutinating languages except for their tendency
to combine multiple meanings or functions into a single morpheme (hence the
term ‘fusional’); and polysynthetic languages,
such as Greenlandic referred to in the introduction, allow nouns to be
incorporated (similarly to compounding) with morphemes such that a single
‘word’ can express whole sentence-like meaning. However, a valid criticism of the
above system of morphological typology is that some languages belong in more
than one category, and therefore it is limited in its theoretical usefulness (Spencer,
1993). In reality, morphological richness is a continuum, with linguists such
as Haspelmath and Sims (2010) using the single dimension of analytic-synthetic to describe it; in this approach, isolating and
polysynthetic languages are at the extremes, respectively. A given language’s
approximate degree of synthesis can be quantified by calculating its
morpheme-to-word ratio based on corpus data. A now dated but influential study used
text samples to estimate Modern English at 1.68, making it more synthetic than
the isolating Vietnamese (1.06) but more analytic than German (1.92), Old
English (2.12), Swahili (2.55) and the polysynthetic Greenlandic (3.72)
(Greenberg, 1959).
A proper
evaluation of the dual-single debate must not only acknowledge the full gamut
of morphological richness, but it must also recognise the whole range of forms
and functions that inflection presents in human language. In broad terms, there
is little that inflectional morphology does not encode in one language or
another; some of these are functions and concepts that are entirely absent in
the English grammar, such as the evidential mood inflection on Latvian verbs to
mark a statement as ‘hearsay’ (Booij, 2007).
The general grammatical functions of
inflections are to mark agreement (two lexemes take particular forms that
correspond to one another), government (one lexeme has its form imposed on it
morphosyntactically) – both of which are considered ‘contextual’ inflections – and,
in contrast, ‘inherent’ properties (where a particular lexeme, or a member of
its paradigm, has a particular non-contextual intrinsic morphology) (Bauer,
2001; Stump, 2001). The parts of speech which languages typically inflect in
these ways are nouns, verbs and adjectives.
Noun
inflections (declensions) often occur to mark gender and number, which fusional
languages often combine as multifunctional morphemes. Number is often declined
as singular or plural, such as in Finnish, which overtly marks both (Booij,
2007), though some languages such as Slovene (Lingvopedia, 2018) have a dual
form, and other languages even mark trial (Booij, 2007). Case is another
frequent use of contextual inflection, of which ‘direct cases’ (e.g. dative,
nominative, accusative) are syntactically determined, and ‘oblique cases’ (e.g.
locative, instrumental) are semantic. Further noun inflections are for
possession, definiteness, and diminution and augmentation (Stump, 2001).
Languages
use verb inflections (conjugations) to mark person, tense, aspect, polarity
(i.e. affirmation or negation), voice (chiefly active and passive, but also
middle as in Polish (Swan, 2002)), mood (e.g. indicative, imperative,
subjunctive and evidential), and finiteness (e.g. telicity, or ‘completeness’).
Adjectives have generally fewer morphemes, though many languages inflect for
degree (positive, comparative and superlative), as well as definiteness (e.g.
in Syrian Arabic) and number (Booij, 2007; Stump, 2001).
As
the above examples have shown, there are several overlaps where two parts of
speech inflect for the same grammatical or semantic feature. When looking at a
single language, such overlaps bring about government and agreement relations.
For example, the dual number in Slovene is overtly inflected on nouns, verbs
and adjectives in order to mark agreement (Lingvopedia, 2018). This
multiplicity of possible inflections is most apparent in the more synthetic
languages such as Nishnaabemwin, spoken in part of Canada, in which there are
890 theoretical forms for intransitive inanimate verbs alone (Valentine, 2001,
as cited in Lieber, 2010). Although noun, verb and adjective inflections can indeed
yield complex and extensive classes of paradigms, a certain degree of
inflectional homonymy is common, whereby the phonological form of different
members of a lexeme’s paradigm are indistinguishable (e.g. English you went vs she went).
Nevertheless,
it is clear that the possible breadth and depth of productive inflectional
morphology in human languages is considerable. This potential begs the key
question which the essay now turns to: how do children acquire inflectional
morphology? The next section looks at the agreed empirical phenomena observed
in children’s development, before returning to the dual-single debate in
greater detail.
Between the
ages of 12 and 20 months, children’s first whole recognisable words typically
appear. Within the following year, children begin to use certain morphemes
systematically in their speech; for children learning more synthetic languages
like Russian, this development can begin as soon as two months after their
first words (Clarke, 2001). Children then gradually progress to a more advanced
level of morphology, showing increasing competence in whichever inflectional
paradigms feature in their language (see previous section). An important
emphasis here is the word gradually –
it may take many years for a child to master the inflections of his/her
language, owing to either irregularities in form or to difficult semantic
distinctions, both of which will depend on the typology of the language in
question (ibid.). A theoretically important phenomenon at this point is ‘U-shaped
learning’: after early accuracy in the fewer (probably rote-learned) forms that
children use, an overgeneralisation phase occurs when children erroneously
apply a regular inflection where the form should actually be irregular (e.g.
plural inflection *three sheeps).
This phase later ends when the children have learnt the irregular forms, and
their overall accuracy becomes more adult-like (Rowland, 2014).
Overgeneralisation suggests unconscious generalised inflectional processes,
albeit immature, in the child’s mind (Ambridge, & Lieven, 2011).
Eventually,
most children will reach a stage of full productivity whereby they will mark
contextual inflections accurately and consistently on novel verbs. It is
well-established phenomenon in CLA literature, for example, that a six-year-old
child, when faced with a nonce verb such as mot
and prompted to say what happened yesterday, will produce the past tense
form motted (Berko, 1958). Since it
is certain that the child has never before heard such a word and therefore has
not rote-learned the form motted, as
with overgeneralisation, there must be also an underlying generalised process
that the child uses when marking past tense, at least on novel verbs.
The
essential question here, which goes to the core of CLA debates in general, is
whether the above ‘generalised processes’ that children (and adults) use for
inflection can be regarded as instances of a formal grammatical rule (e.g. mot: ‘add -ed to denote the past’) or rather an analogy to similar known
paradigms (e.g. mot: ‘copy the
pattern spot-spotted, dot-dotted etc.’).
As will be shown, these two explanations are not exactly unrelated, but they
are associated with generativist and constructivist approaches, respectively.
Generative linguists argue that rules are essential for children to acquire language,
in this case, rules of regular inflectional
morphology. (Irregular morphology is by definition not rule-like and therefore
viewed as separate from the core grammar.) Furthermore, the so-called nativist
view is held by a particular group of generativists who argue further that
these formal rules – or the capacity to develop such rules – must be innate
(e.g. Rowland, 2014). On the other hand, constructivists argue against the need
for, and theoretical viability of, formal rules; it is claimed instead that
children use analogy and pattern-finding skills to develop language, including
inflectional morphology. (Grammar-like regular forms and lexis-like irregular
forms are treated as one and the same, with no specific ‘core grammar’ in the
generative sense.) (Tomasello, 2003).
The specific theories that this
essay concerns, as introduced earlier, are the dual-route and single-route
models of inflectional morphology. As stated above, the two theories are not entirely
incompatible with one another; the dual-route model incorporates the role of
memory storage and analogy (irregular inflection only) that constitutes all of
the single-route model, and the rule-based process for regular inflection need
not be innate or based on universal grammar, a concept fundamentally refuted by
constructivists (e.g. Tomasello, 2003). Leaving these overlaps temporarily to
one side, the dual-route account will now be considered in isolation.
The dual-route
account, sometimes referred to as ‘words and rules’, was originally developed
in large part by the well-known linguist Steven Pinker (1998). The theory,
which began with a focus on the English past tense inflection, states that “irregular
past-tense forms are stored in the lexicon, a division of declarative memory,
whereas regular forms can be computed by a concatenation rule, which requires the
procedural system” (Pinker, & Ullman, 2002, p. 456). The terms ‘declarative
memory’ and ‘procedural system’ are separate parts of the mind which are said
to be located in two distinct regions of the brain; this dissociation is
consistent with an earlier neurological study by Jaeger et al. (1996) where
positron emission tomography (PET) scans of subjects asked to read or speak
past-tense verbs revealed a difference in speed and distribution of neuronal
activation in the brain when processing either regular or irregular forms. In
addition, the phrase ‘computed by a concatenation rule’ refers to the
application of formal morphological rules described earlier. Note that concatenation
(here, past tense suffixation) is only one of a number of inflectional forms
possible in language; the theory does not directly incorporate others such as
infixes, circumfixes and transfixes not present in English inflection (Bauer,
2001), though they could be included in principle. Pinker and Ullman (2002)
counter criticisms that this part of the dual-route theory is based only on
English and fails to explain inflectional processes in other languages by
citing studies into the German plural as further support for the theory. In
addition, an important clarification for the theory is that the term ‘rule’ may
be misleading: the dual-route theory does not posit an explicit rule in the
sense which a language teacher may use it, but rather it is an unconscious
default affixation of a morpheme (e.g. -ed)
to a part of speech which is variable (e.g. a verb), which occurs in a
consistent rule-like way (ibid.). How exactly any given specific inflectional
rule is acquired is not clear.
This default rule, however, is only
one half of the ‘words and rules’ mechanism, because it only applies to regular
forms. The rule can be overridden, or ‘blocked’, when a known irregular form is
retrieved from the lexicon instead, hence the ‘words’ part of the theory
(Pinker, 1998). This is what occurs when a mature speaker says three sheep, knowing (often implicitly)
that the plural sheep is irregular,
whereas a young child overgeneralising to *three
sheeps has not learnt this exception and therefore is simply applying the
default rule ‘add -s’ that he/she has
acquired; there is no entry for sheep (plural)
in the child’s lexicon strong enough to block this process. Note that strength,
based on frequent of exposure to a particular word, is a relevant factor here.
As the child experiences the correct irregular form more often, a point will be
reached where he/she begins to correctly block the default inflection with the
correct form (Stemberger, 2001). This may explain why children do not suddenly
begin consistently using the correct irregular form, and they sometimes use
both incorrect and correct forms during the same developmental period. The fact
that the English irregular verbs comprise the majority of the most frequent
verbs is compatible with this claim (Pinker, & Ullman, 2002); a blocking
mechanism would not work reliably if certain irregular forms were rarely
encountered.
Another way in which blocking may
not predictably occur is for pairs of verbs that have both irregular and
regular forms, such as dreamed / dreamt, where the regular form dreamed may or may not be blocked by dreamt, depending on the strength of dreamt in the speaker’s own lexicon. The
same holds true for regulars that have a phonological similarity to irregulars,
as in the regular blink, which does
not conjugate similarly to the irregular-but-frequent drink, though the high strength of drink-drank-drunk may exert a partial blocking effect, encouraging
a perhaps younger speaker to possibly form blank
or blunk instead of applying the
default to correctly form blinked (Pinker,
& Ullman, 2002).
It is clear from above that the dual-route
theory acknowledges the relevance of phonological similarity – observing that
verbs (in this case) form so-called phonological clusters (Ambridge, &
Lieven, 2011), and that this plays a role in encouraging the production of
certain inflections. However, the theory specifically states that this only
occurs in the lexicon, for irregular forms. In contrast, the default rule function
will always apply to any stem regardless of its phonological form (unless, of
course, in the event of blocking by the irregular), hence its status of being
‘default’, or what Pinker and Ullman (2002) refer to as “a productive default
that does not critically depend on the statistics of patterns in memory” (p. 458).
If this was not the case, and the rule was influenced by phonological factors,
then it would no longer be
a rule – the essence of the dual-route model.
If the
application of an inflectional rule was subject to phonological forces, the
‘rule’ would be better characterised as a ‘tendency’ constrained by phonological
patterns present in the afore-mentioned clusters, such as swim-swam / ring-rang and
sleep-slept / weep-wept. The way in which regular forms are inflected would
actually mirror the way in which irregular forms are produced. In short, the
‘words’ and ‘rules’ would, essentially, become one and the same (McClelland
& Patterson, 2002), perhaps under the better description of ‘phonological
analogies’. This is the essence of the single-route model.
The single-route model posits that
all inflections – be it English past tense or Slovene dual – are produced by
the same mechanism of phonological analogy across known forms stored in the
lexicon (Ambridge, & Lieven, 2011). Because words that can be inflected
(nouns, verbs, adjectives) are semantic in nature, the single-route theory
claims that semantics is a factor in determining all inflections, regardless of
regularity (Ramscar, 2001). As discussed earlier, an inflectional paradigm
corresponds to a complete, meaningful lexeme; the single-route theory presupposes
an associative memory of whole lexemes according to similarities in form and
semantics. The dual-route theory, to reformulate it here, claims that those
lexemes whose inflectional paradigms are regular have no associations with one
another, neither by form nor by semantics – they only relate inasmuch as they
inflect in the same, regular way.
Where the dual-route theory found arguable
empirical support in PET scans of the brain, the single-route theory found it
in the pioneering ‘connectionist model’ (Rumelhart, & McClelland, 1986, as
cited in Bauer, 2001). In this experiment, a computer was trained on learning
material consisting of pairs of present and past English verb forms, in which
it searched for phonological patterns. Semantic features were not built into
the model. After the training, the computer was inputted with new verbs and,
using its learned network (a ‘pattern associator’), it outputted past forms.
The authors calculated a 91% success rate, though a recalculation of the data
by Pinker and Prince (1988) suggested 62%; in either case, it is compelling
evidence that phonological analogy alone can go a long way towards producing
accurate past tense inflections, and it lends support for the single-route
mechanism.
The
similarities and differences of the single-route and dual-route model of
inflectional morphology have been presented. The final part of this essay
examines the cross-linguistic validity of these theories by reviewing studies
into children’s acquisition of inflection in languages other than English.
In terms of studies whose results
support the dual-route theory, the majority is still based on English, usually
the past tense inflection. These studies generally look for evidence in the
form of neurological dissociation of how and where regular and irregular
inflections are used, either via direct brain imaging similar to the PET scans described
earlier (Bakker, MacGregor, Pulvermüller, & Shtyrov, 2013; Jaeger et al.,
1996), or by training and testing children with specific language impairment
(SLI) (Smith-Lock, 2015). The evidence for different processing mechanisms for
English regular and irregular forms is quite convincing, which raises the
question as to why there are relatively few studies comparing regular and
irregular inflection in more synthetic languages.
However,
there are some studies that have attempted to do this. Clahsen, Alvedo and Roca
(2002) analysed children’s errors in Spanish regular and irregular verbs, in
which they found that overgeneralisation occurred in one direction only:
regular inflections being applied to irregular stems. Furthermore, there was
some evidence that frequency distribution in the children’s vocabulary did not
affect which verbs were overgeneralised, suggesting rule-based behaviour impervious
to certain patterns. Another study focused on Russian, whose verbs comprise 11
classes of distinct inflectional paradigms which do not group as neatly into
‘regular’ and irregular’ as English verbs do. The authors had to represent
‘regular’ verbs as those in the most frequent and productive of these classes,
and ‘irregular’ from the smallest and most unproductive class. This study was
another which used neuroimaging, again with findings that pointed to a physical
dissociation in the processing of ‘regular’ and ‘irregular’ verb inflections,
this time for a highly synthetic and fusional language; however, an important additional
observation was of the high level of connectivity between the two regions of
the brain, especially when subjects had to produce more morphologically complex
forms (Kireev, Slioussar, Koroktov, Chemigovskaya, & Medvedev, 2015). This
would suggest an inflectional processing system – for Russian verbs, at least –
somewhere between a distinct dual route and a single route.
Studies
which have provided evidence for the single-route model encompass a wider range
of more synthetic languages, though the vast majority are Indo-European and
fusional rather than agglutinative or polysynthetic. Where the pro-dual-route
studies have tended to look for dissociation, these studies challenged the
notion of ‘default’ (i.e. ‘regular’) which is central to the rule-forming mechanism.
Studies into Dutch plurals (Keuleers et al., 2007), German plurals (Zaretsky,
Lange, Euler, & Neumann, 2016), Dinka noun numbers (Ladd, Remusen, &
Manyang, 2009) and Polish genitive, dative and accusative case-marked nouns (Dąbrowska, 2001; Dąbrowska, & Szczerbiński, 2005) concluded that there is no ‘default’ inflection for those
specific and relatively complex paradigms, which fundamentally contradicts the
dual-route model. Furthermore, these studies found that children’s accurate
production was significantly sensitive to frequency and phonological form,
directly conflicting with the conclusion of Clahsen, Alveledo and Roca’s (2002)
Spanish study above. Aguado-Orea and Pine (2015) also looked at Spanish verb
inflections (regular and irregular), finding that error rates were inversely
proportional to verb frequency, again casting doubt on the idea of a formal
inflectional rule insensitive to frequency and form. Finally, Smolik and Kříž (2015) found that the semantic property of
imageability facilitated Czech-speaking children’s acquisition of verb and noun
morphology, lending weight to the theory that inflections are somehow linked to
the lexicon rather than resulting from formal operations.
It may appear from the previous
section that the weight of cross-linguistic evidence lies in favour of the
single-route theory, but these studies’ conclusions are tentative, and none so
far have addressed the considerable evidence for dissociation from neurological
studies. A key point here is that evidence for ‘dissociation’ does not
necessarily entail evidence for ‘dual-route’, nor does evidence against a
‘default’ entail evidence for ‘single-route’. Studies which have approached a
particular inflectional paradigm with an open approach towards the dual-single debate
have been the most inconclusive (e.g. Nicoladis, & Paradis, 2012). Furthermore,
for both accounts there are still numerous languages and inflectional systems
for which children’s acquisition has had little or no research conducted into
it.
It
does, however, appear that both models in their current form have their own
insurmountable flaw: the dual-route theorists may need to concede that
semantics, frequency and phonological form play a role in all inflection,
albeit sometimes variably, even if it is not apparent in the rather
unrepresentative English past tense system. They may also need to acknowledge
that the notion of regularity and default is highly spurious when one looks at
how inflectional morphology works in some of the world’s languages.
Equally, the
single-route theorists may have to accept that the human brain does process
simpler, more regular inflections in a different way to those particular
‘exceptions’ which are more idiosyncratic and more lexical – though the these
two processes may be two ends of a continuum rather than discrete and isolated
in the mind (Kireev et al., 2015). They would also have to agree with Booij
(2007, p. 243) in that a lexicon which stores each inflection within a paradigm
for highly synthetic languages (such as Turkish or Nishnaabemwin) without a means to capture redundant learning would be “quite absurd”.
The possible reconciliation here,
combining the merits of the dual- and single-route theories, which would
accommodate the morphological diversity of language summarised earlier in this
essay, would be to take the notion of an inflectional ‘rule’ and regard it rather
as a schema, or set of schemas (Ambridge, & Lieven, 2011), whose form is
based on the distributional frequency of inflections in a given language. The
schema is input-based and therefore constrained by semantics, frequency and
phonology, as described in the usage-based theory of language acquisition
(Tomasello, 2001). Such a unification of the dual-route and single-route
mechanisms, with the respective concessions highlighted above, may lead to a
better all-round theory of morphological acquisition that stands the
cross-linguistic test. There are signs that this development is in progress,
such as recent research into Lithuanian noun morphology, which supported a
usage-based mechanism that was neither fully ‘dual’ nor ‘single’ in their form
described in this essay (Savičiutė, Ambridge, &
Pine, 2018). As Pinker himself wrote (2002, p. 462), “the adversarial nature of scientific debate might sometimes have
prevented both sides from acknowledging that features of one model may
correspond to constructs of the other”.
List of References
Aguado-Orea, J.,
& Pine, J. M. (2015). Comparing different models of the development of verb
inflection in early child Spanish. PLoS
ONE, 10(3): 1-12. https://doi.org/10.1371/journal.pone.0119613
Ambridge, B. &
Lieven, E. V. M. (2011). Child language
acquisition: Contrasting theoretical approaches. Cambridge, United Kingdom:
Cambridge University Press.
Anderson, S. R.
(1982). Where's morphology? Linguistic
Inquiry, 13: 571-612.
Aronoff, M., &
Fudeman, K. (2005). What is morphology? Oxford, United Kingdom: Blackwell
Publishing
Bakker, I.,
MacGregor, L. J., Pulvermüller, F., & Shtyrov, Y. (2013). Past
tense in the brain’s time: Neuropsychological evidence for dual-route
processing of past-tense verbs. NeuroImage,
71: 187-195. http://dx.doi.org/10.1016/j.neuroimage.2012.12.065
Bauer, L. (2001). Morphological productivity. Cambridge,
United Kingdom: Cambridge University Press.
Berko, J. (1958).
The child’s learning of English morphology. Word,
14: 150-177. https://doi:10.1080/00437956.1958.11659661
Booij, G. (2007). The grammar of words: An introduction to
morphology. (2nd ed.). Oxford, United Kingdom: Oxford University
Press.
Bybee, J. L., &
Moder, C. L. (1983). Morphological classes as natural categories. Language, 59(2): 251-270.
https://doi.org/10.2307/413574
Clahsen, H., Aveledo,
F., & Roca, I. (2002). The development of regular and irregular verb
inflection in Spanish child language. Journal
of Child Language, 29: 591-622. https://doi.org/10.1017}S0305000902005172
Clark, E. V. (2001).
Morphology in language acquisition. In A. Spencer & A. M. Zwicky (Eds.), The handbook of morphology (pp.
374-389). Oxford, United Kingdom: Blackwell Publishers Ltd.
Greenberg, J.
(1959). A quantitative approach to the morphological typology of language. International Journal of American
Linguistics, 26: 178-194.
Haspelmath, M.,
& Sims, A. D. (2010). Understanding morphology.
(2nd ed.). London, United Kingdom: Hodder Education.
Jaeger, J. J.,
Lockwood, A. H., Kemmerer, D. L., van Valin, R. D., Murphy, B. W., &
Khalak, H. G. (1996). A positron emission tomographic study of regular and
irregular verb morphology in English. Language,
72: 451-497. https://doi.org/10.2307/416276
Kelly, B., Wigglesworth,
G., Nordlinger, R., & Blythe, J. (2014). The acquisition of polysynthetic
languages. Language and Linguistics
Compass, 8(2): 51-64. https://doi.org/10.1111/lnc3.12062
Keuleers, E.,
Sandra, D., Daelemans, W., Gillis, S., Durieux, G, & Martens, E. (2007).
Dutch plural inflection: the exception that proves the analogy. Cognitive Psychology, 54(4): 283-318.
https://doi.org/10.1016/j.cogpsych.2006.07.002
Kireev, M.,
Slioussar, N., Korotkov, A. D., Chemigovskaya, T. V., & Medvedev, S. V.
(2015). Changes in functional connectivity within the fronto-temporal brain
network induced by regular and irregular Russian verb production. Frontiers in Neuroscience, 9(36): 1-10.
https://doi.org/10.3389/fnhum.2015.00036
Ladd, D. R., Remusen,
B., & Manyang, C. A. (2009). On the distinction between regular and
irregular inflectional morphology: Evidence from Dinka. Language, 85(3): 659-670. https://doi.org/10.1353/lan.0.0136
Lieber, R. (2010). Introducing morphology. Cambridge,
United Kingdom: Cambridge University Press.
Lingvopedia. (2018).
Slovene. Retrieved from http://lingvo.info/en/lingvopedia/slovene
McClelland, J. L.,
& Patterson, K. (2002). ‘Words or rules’ cannot exploit the regularity in
exceptions. Trends in Cognitive Sciences,
6(11), 464-465. https://10.1016/S1364-6613(02)02012-0
Nicoladis, E., &
Paradis, J. (2012). Acquiring regular and irregular past tense morphemes in
English and French: Evidence from bilingual children. Language Learning, 61(1): 170-197. https://doi.org/10.1111/j.1467-9922.2010.00628.x
Pinker, S. (1998).
Word and rules. Lingua, 106(1-4):
219-242. https://doi.org/10.1016/S0024-3841(98)00035-7
Pinker, S., &
Prince, A. (1988). On language and connectionism: Analysis of a parallel
distributed processing model of language acquisition. Cognition, 28: 73-193. https://doi.org/10.1016/0010-0277(88)90032-7
Pinker, S., &
Ullman, M. T. (2002). The past-tense debate: The past and future of the past
tense. Trends in Cognitive Sciences, 6(11),
456-463. https://10.1016/S1364-6613(02)01990-3
Ramscar, M. (2001).
The role of meaning in inflection: Why the past tense does not require a rule. Cognitive Psychology, 45: 45-94.
Rowland, C. (2014). Understanding child language acquisition.
Abingdon, United Kingdom: Routledge.
Savičiutė, E., Ambridge, B., & Pine, J. M.
(2018). The roles of word-form frequency and phonological neighbourhood density
in the acquisition of Lithuanian noun morphology. Journal of Child Language, 45(3): 641-672. https://doi.org/10.1017/S030500091700037X
Smolik, F., & Kříž, A. (2015). The power of imageability: How the
acquisition of inflected forms is facilitated in highly imageable verbs and
nouns in Czech children. First Language,
35(6): 446-465. https://doi.org/10.1177/0142723715609228
Spencer, A. (1993). Morphological theory. Oxford, United
Kingdom: Blackwell Publishers.
Stemberger, J. P.
(2001). Morphology in language production with special reference to
connectionism. In A. Spencer & A. M. Zwicky (Eds.), The handbook of morphology (pp. 428-452). Oxford, United Kingdom:
Blackwell Publishers Ltd.
Stump, G. T. (2001).
Inflection. In A. Spencer & A. M. Zwicky (Eds.), The handbook of morphology (pp. 13-43). Oxford, United Kingdom:
Blackwell Publishers Ltd.
Swan, O. E. (2002). A grammar of contemporary Polish. Bloomington,
IN: Slavica Publishers.
Tomasello, M. (2003). Constructing
a language: A usage-based theory of language acquisition.
Cambridge, MA: Harvard University Press.
Zaretsky, E., Lange, B. P., Euler, H. A., &
Neumann, K. (2016). Factors considered and ignored in plural acquisition:
Frequency rules? Language and Cognition,
8: 283-313. https://doi.org/10.1017/langcog.2014.51