Strange sounds, familiar words. Interlingual decoding from a CxG
Version of record published in: Timothy Colleman, Frank Brisard, Astrid De Wit,
Renata Enghels, Nikos Koutsoukos, Tanja Mortelmans & María Sol Sansiñena (eds.),
The wealth and breadth of construction-based research (Belgian Journal of Linguistics
34 (2020)), 122–134.
When communicating across closely related languages or varieties (e.g., in interdialectal
communication or in regions such as Mainland Scandinavia), speakers have to learn how
to decode words that show partial phonological differences from the equivalents in their
L1. Although contact situations like these are rather common, interlingual decoding has
scarcely been addressed in the CxG literature. As a contribution to this field of research,
the paper discusses how (a particular stage in) emerging receptive multilingualism can be
modelled from a CxG perspective. Specifically, it deals with the idea that repeated inter-
lingual decoding generates partially schematic cross-linguistic constructions mirroring
the speaker’s knowledge about sound correspondences, as suggested by Diasystematic
Construction Grammar (Höder 2019).
1 Learning to decode across linguistic boundaries
When speakers of different languages talk to each other and each is using his or her own
language in the conversation, this can work in two different ways: (a) They know their
interlocutor’s language or (b) they do not know it, but their languages are so similar that
they are – to a certain extent – mutually intelligible. While the structural similarity of the
languages is irrelevant in the former case, the mode of communication presented in the
latter relies on cross-linguistic similarities, most importantly on cognates, and would
therefore only work for typologically close language pairs with sufficient overlaps in the
That said, even cognates from closely related languages often show more or less ex-
treme phonological differences, which makes intercommunication of that sort a challenge
for the untrained listener. When the differences are systematic and speakers are repeatedly
exposed to input from the other language, however, it seems reasonable to assume that
speakers would acquire the respective sound correspondence rules (Weinreich 1954, 390;
Braunmüller 1995, 45–46; Höder 2019, 346). Knowing these, they could exploit given
cross-linguistic similarities better and increase their decoding success.
Given enough input, a speaker without prior knowledge of the other language (as in
scenario (b)) would eventually switch over to a mode of communication based on full
receptive knowledge of the other language (as in scenario (a)). This transition amounts to
a learning process in the course of which different decoding strategies become available.
Fig. 11 shows how these strategies can be classified regarding the receptive proficiency
level necessary for their application, with the leftmost side of the continuum illustrating
the absence of receptive knowledge and the rightmost side a high level of receptive
Fig. 1: Decoding strategies at different levels of receptive proficiency
When speakers do not have any prior receptive knowledge, the only resource they can
access in decoding is the phonological similarity between the input and stored lexical
representations from their L1 or additional languages/varieties. They can, so to speak,
only make ‘uninformed’ guesses. The central decoding process at this stage would be
gestalt recognition (cf. Braunmüller 1995, 44; Pulvermüller 1996, 318), i.e., recognising
patterns as representations of already known forms despite partial differences, potentially
facilitated by prior experience with phonological variability in the speakers’ L1 (through
dialects or registers). If speakers are repeatedly confronted with the other language, how-
ever, they would presumably work out the rules relating to specific sound correspond-
ences sooner or later and be able to apply them to new input. This intermediate stage
would allow for informed guesses. Eventually, at a high level of receptive knowledge
acquired through extensive contact with the new language, decoding increasingly works
via whole word recognition, because a speaker will have heard more and more words
often enough to have formed directly accessible representations (Bannert 1981). Guessing
would thus be superseded by knowing. Note that although there is a general progress from
less to more receptive knowledge, this knowledge is also item-based and different
decoding strategies might be applied simultaneously in many cases once the knowledge
accumulation process has started.
This paper is concerned with the intermediate stage and discusses how the acquisition
of sound correspondence rules as a type of receptive multilingual knowledge can be mod-
elled from a CxG perspective, more precisely from the perspective of Diasystematic Con-
struction Grammar (DCxG) (cf., e.g., Höder 2019). It picks up suggestions for the analy-
sis of multilingual knowledge organisation made by Höder (2018; 2019) and applies them
to a model case of receptive knowledge acquisition step by step.
The paper is structured as follows: Section 2 explains the linguistic relevance of the
phenomenon of intercommunicative decoding. Section 3 outlines the perspective adopted
and the research questions to be discussed. After a brief introduction to the basic assump-
tions of DCxG in Section 4, Section 5 presents an example analysis of the establishment
of a sound correspondence construction. Section 6 gives a conclusion.
2 An ordinary (but underinvestigated) phenomenon
The communicative scenario described above might seem exceptional at first sight, but
there are many language constellations that – in principle – allow for some form of inter-
communication based on mutual intelligibility, and in some regions this mode of commu-
nication is indeed rather common, e.g., in Mainland Scandinavia (Braunmüller 2008).
One has to bear in mind, of course, that the languages involved in such scenarios are
usually closely related and do not differ from each other much more than do some dialects
of one and the same language. Once the restriction to languages is discarded, it becomes
apparent that the scenario is actually far from exceptional and that speakers quite regu-
larly have to decode phonologically divergent input, e.g., in interdialectal communication
(especially in speech communities where this is common practice, e.g., in Switzerland or
Norway) or when dealing with foreign-accented speech.
Despite this, intercommunicative decoding has scarcely been addressed in the CxG
literature. The subject in fact touches upon several somewhat underresearched aspects,
namely (a) receptive knowledge (in contrast to knowledge used for production), (b) the
phonological side of constructions (especially on the submorphemic level), and (c) the
decoding of phonologically divergent input in general. Although some constructionist re-
searchers have attended to these issues (e.g., Vihman and Croft 2007), the variation al-
lowed for in previous accounts of spoken language decoding is often moderate, ranging
within the boundaries of a relatively homogenous ‘monolingual’ system, or it is already
fully integrated as part of the hearer’s (multilingual) linguistic knowledge. Under these
circumstances, the full range of challenges hearers are confronted with in intercommuni-
cation naturally does not come to the fore.
At the same time, hardly any researchers on intercommunication/intercomprehension
have adopted a cognitivist view on their field of research either. Interlingual comprehen-
sion is a hot topic not just among Scandinavian language researchers, but also, e.g., among
Romance language researchers (cf., e.g., Klein and Stegmann 2000), and, accordingly, an
extensive body of literature is dedicated to the subject. However, previous research has
primarily been concerned with the assessment of comprehension skills and the develop-
ment of teaching strategies, but not with the cognitive processes behind comprehension.
Furthermore, studies from intercomprehension research often focus exclusively on writ-
ten language comprehension.
Although there are some theoretical contributions that do address single cognitive pro-
cesses involved in intercommunicative decoding in spoken contexts (e.g., Braunmüller
1995; van Heuven 2008), no comprehensive model of intercommunicative decoding has
been developed so far that approaches the phenomenon from a usage-based perspective.
With their commitment to psychological realism and the descriptive tools they provide,
Cognitive Linguistics and usage-based CxG could contribute to modelling intercommu-
nicative decoding in a systematic and cognitively realistic way and thereby help close the
cognitive gap in intercommunication research.
3 Perspective and research questions
The perspective adopted in this paper is strictly usage- and acquisition-based. Acquisition
in this context must be understood as the acquisition of constructions (Ellis and Wulff
2019), but not as (additional) language acquisition in the traditional sense, i.e., as a pro-
cess in which speakers learn how to speak an additional language. Instead, the learning
process is primarily geared towards knowing enough about the new language to be able
to decode the input. This does not mean that speakers would not eventually learn to use
the language for production as well, but the stage of receptive knowledge discussed here
deals with constructions that emerge from decoding for decoding and on the basis of rel-
atively little input. It thus focusses on a transitionary stage all learners of a closely related
language go through in implicit learning, but which – for some – represents their linguistic
knowledge more permanently, namely if they only use it sporadically in intercommuni-
From such a perspective, the following questions arise:
1) Which prerequisites must be fulfilled in order for a speaker to work out cross-
linguistic sound correspondences?
2) Which constructions would correspond to this linguistic knowledge?
Before these questions are discussed on the basis of an example from an Interscandina-
vian context, the following chapter briefly explains the basic assumptions of Diasystem-
atic Construction Grammar.
4 Diasystematic Construction Grammar
Diasystematic Construction Grammar (e.g., Höder 2018; 2019) is a usage-based CxG ap-
proach to language contact phenomena and multilingual/-lectal knowledge organisation.
DCxG proposes that linguistic knowledge about different languages/varieties is (poten-
tially) interconnected rather than stored and processed separately. If, as CxG suggests, all
linguistic knowledge can be subject to generalisation processes, there is no reason to as-
sume that speakers do not also detect and exploit form-meaning correspondences across
linguistic boundaries. Höder (2019, 341), therefore, suggests that besides constructions
that are assigned to a specific language/variety (so-called idioconstructions), a multilin-
gual/-lectal speaker’s constructicon also comprises language-unspecific diaconstructions.
These are established when speakers generalise across idioconstructions from two (or
more) different varieties.
Diaconstructions can be fully lexically filled, but the generalisation often leads to par-
tially or fully schematic constructions (Höder 2019, 342–343). Partially schematic di-
aconstructions contain, on the one hand, the formal and/or functional overlap detected in
a process of interlingual identification (i.e., the parts that are treated as equivalent across
the two idioconstructions) and, on the other, schematic slots for the parts that are per-
ceived as different (Höder 2019, 341–342). The slots can be filled with material from
either variety, thereby specifying the forms in terms of communicative context (Höder
2019, 340). The specification of the communicative context is thus part of the slot-filling
idioconstructions’ function. For convenience, the communicative contexts will be sub-
sumed under the label of the respective glottonym in the same shorthand as used by Höder
(2019, 340): <Cglottonym>, e.g., <CDanish>.
Crucially, Höder (2019, 344–346) suggests that schematic slots are relevant also on
the submorphemic level, especially in interlingual identification processes taking place
between closely related varieties. He even explicitly mentions correspondence rules as a
likely example for cross-linguistic constructions. The following chapter explains what the
emergence of sound correspondence constructions could look like.
5 Example analysis
The analysis uses Danish-Swedish intercommunication as a model case. Danish and Swe-
dish are closely related, and their grammatical and lexical similarities are striking in writ-
ten form. However, Danish pronunciation has experienced some dramatic changes in the
recent past that are not mirrored in the orthography and which create a considerable lin-
guistic distance to spoken Swedish (Sandøy 2005). This has led to relatively low average
comprehension scores in Interscandinavian comprehension tests among Swedish partici-
pants (cf., e.g., Delsing and Lundin Åkesson 2005).
Danish and Swedish offer various more or less systematic sound correspondences. One
that is particularly puzzling to speakers of Swedish is that between Swedish long open
back rounded [ɒː] and Danish long open-mid front unrounded [ɛː] (orthographically rep-
resented by <a> in both languages)2, featured in cognate pairs like, e.g., hval/valDA/SV
‘whale’, aneDA/anaSV ‘ancestor’ (cf. Tab. 1).
Tab. 1: Danish and Swedish cognates compared
The correspondence’s potential to confuse speakers of Swedish lies in the fact that [ɛ(ː)]
is also part of the Swedish vowel system, but corresponds to what is orthographically
represented as <ä> (cf. Gooskens and van Bezooijen 2013). Knowing the correspondence
rule can thus minimise the risk of misinterpreting input.
The point of departure for the analysis is a scenario where a speaker of Swedish who has
never heard spoken Danish before enters into an intercommunicative conversation with a
speaker of Danish. In the conversation, s/he is confronted with different words that con-
tain Danish [ɛː] and correspond to a Swedish cognate equivalent featuring [ɒː] in the same
position each – e.g., among others, the three examples specified in Fig. 2.
Fig. 2: Establishment of cross-linguistic sound correspondence constructions
In decoding, speakers can arrive at the concept associated with a lexeme’s sound pattern
in different ways, e.g., by exploiting context cues or based on repairs from their commu-
nication partners. However, given space limitations, the analysis concentrates on phono-
logical word recognition in the following, i.e., the comparison of input with already stored
representations based on formal (phonological) similarities.
Broadly speaking, three steps are necessary for a sound correspondence construction
to emerge in such a scenario: (a) word recognition, (b) repeated interlingual identification
(either in word recognition or later on the representational level), and (c) abstraction and
generalisation. These, in turn, lead to the activation and/or establishment of three different
kinds of links: (a) symbolic links (in word recognition), (b) cross-linguistic links (in in-
terlingual identification), and (c) inheritance links (in generalisation).
The first step – successful word recognition – forms the prerequisite for any further
knowledge acquisition regarding the sound correspondence. In successful phonological
word recognition, the symbolic link between the Swedish form (as the stored representa-
tion with the greatest overlap) and the concept associated with it will be activated. Sim-
ultaneously, the Danish form and the linguistic and extra-linguistic context it has been
encountered in leaves a first memory trace in the form of an exemplar (Goldberg 2019,
13–14), which can be the first step towards a robust symbolic link in a new Danish lexeme
Drawing a connection to a known Swedish word in this case coincides with what
Weinreich (1964, 7; cf. also Höder 2018, 54) calls interlingual identification, i.e., the
process of recognising two elements from different languages/varieties to be functional
and/or formal equivalents. On the word level, the whole phonological form of the Danish
input word is associated with the representation of the Swedish cognate equivalent in the
speaker’s constructicon. Below the word level, however, interlingual identification can
take place, too, and this does not seem too far-fetched, either, since the on-line compari-
son of an incoming sound pattern with established representations in word recognition
already includes the identification of similarities and differences, i.e., matching vs. non-
matching elements. First submorphemic cross-linguistic associations are thus already be-
ing made in the process of phonological word recognition.
Understanding the utterances the cognates occur in reassures the speaker that s/he
matched them with the correct Swedish equivalent, which in turn validates the cross-lin-
guistic mapping of matching and non-matching sounds on the submorphemic level. If the
same non-matching sounds (i.e., the sounds of the sound correspondence) occur consist-
ently in the same position in words that are identified as equivalents across the languages,
they will sooner or later be identified as submorphemic equivalents, too.
These associations might then be further developed on a long-term representational
level, forming the basis for more permanent cross-linguistic links (Höder 2018, 57–58).
It is also likely that the sounds of the sound correspondence would be perceived as salient,
because they naturally stand out in a linguistic environment that is characterised by cross-
linguistic similarities, which gives them a certain surprisal value.
This is the point where the association between the sounds could be stored as a sound
correspondence rule in the form of a Danish-Swedish DCxn that can be used productively
in processing (in this case in decoding) (Bybee 2010, 94). The reoccurring element [ɛː] is
consistently associated with the corresponding sound in Swedish ([ɒː]), while the sur-
rounding phonological context varies as the sounds are embedded in different cognates.
The overall frequency of the sound correspondence as a submorphemic cross-linguistic
pattern is therefore higher than the frequency of each instantiation (Zeldes 2012, 204),
thus promoting a schematisation process.
Crucially, the repeated identification of a reoccurring element across variable types
along with repeated activation of cross-linguistic links leads to a parallel schematisation
process on the basis of both the witnessed Danish exemplars and the co-activated Swedish
representations. While whole word forms are connected via symbolic links with their lex-
ical meaning and relevant pragmatic information, the single sounds are stripped of all
lexical meaning after the abstraction process. The only information left would be their
association with the communicative context. This makes them what Höder (2019, 344)
calls phonological language markers (PLMs), whose function is to mark the linguistic
forms they occur in as belonging to a specific communicative context (as opposed to
(1) a. Danish PLM: […-[ɛː]-… <CDanish>]
b. Swedish PLM: […- [ɒː]-… <CSwedish>]
Their relation to each other, i.e., their being part of a correspondence, is in turn expressed
in the Danish-Swedish DCxn connecting the two PLMs in Figure 2. The DCxn basically
contains the information that there is a Danish-Swedish sound correspondence with the
following form-function assignment: [ɛː] = <CDanish> vs. [ɒː] = <CSwedish>. It is thus a
generalisation based on the submorphemic cross-linguistic links between the concrete
The interlingual nature of the DCxn is what makes it useful for decoding. If speakers
are confronted with a new cognate including the sound known from the sound corre-
spondence, the Danish PLM is activated, which in turn activates the Danish-Swedish
DCxn and the Swedish PLM. The comparison of input and stored exemplars in word
recognition would now be two-pronged: Not only exemplars that are perceived as similar
to the input sound pattern are activated, but also exemplars which contain the equivalent
Swedish sound and otherwise conform to the sound pattern.
The interplay between overlap comparison in the phonological word recognition pro-
cess and activation of PLMs would increase the likelihood that word recognition is suc-
cessful. Potentially allowing a speaker to decode a significant amount of unknown forms,
the use of sound correspondence constructions presents not only a useful, but also a very
economical decoding strategy (Höder 2019, 346).
One might, of course, ask whether higher-order schemas like the DCxn are psycholog-
ically realistic. In order to judge this, it might be helpful to apply the same reasoning to a
similar structural setup in a monolingual context (cf. also Höder 2019, 342). When, for
example, a language regularly marks singular forms of nouns with one type of suffix and
plural forms with another, there would be a singular construction and a plural construc-
tion, each containing their specific suffix on the form side. There might, however, also be
a more abstract construction subsuming the two suffix forms under the functional cate-
gory number. Its function to mark number would correspond to the diaconstruction’s
function to mark communicative context. The category itself captures the single construc-
tions as members of the same class with different values. If we consider the existence of
a morphological higher-order schema like this plausible, then a phonological diaconstruc-
tion is just as conceivable. Ultimately, the question is whether one wants to grant such
associations constructional status in general – a question which informs an ongoing de-
bate (Hilpert 2019).
In Section 3, the following research questions were formulated:
1) Which prerequisites must be fulfilled in order for a speaker to work out cross-
linguistic sound correspondences?
2) Which constructions would correspond to this linguistic knowledge?
With regard to the first question, the paper suggests that there are both structural and
cognitive prerequisites. The sound correspondence must be (relatively) systematic and
instantiated by variable types (i.e., different cognates). The cognates must be successfully
decoded and recognised as cross-linguistic equivalents by the speaker. Finally, interlin-
gual identification on the submorphemic level must lead to the speaker’s identifying the
two sounds of the correspondence as functional equivalents (in the given context).
Attending to the second question in detail, the example analysis illustrates how sound
correspondence rules can be acquired and applied in the form of phonological language
markers connected to a corresponding DCxn.
The DCxG approach used for the example analysis proves to be particularly well suited
for the task because it is multilingually oriented, strictly usage-based and emergentist. As
such, it focusses on the speaker’s perspective and makes allowance for the commonalities
between receptive knowledge acquisition and more traditional forms of language acqui-
sition as well as for the specifics of the input exposure characteristic to intercommunica-
Crucially, there is no fundamental qualitative difference between the input speakers
are exposed to in intercommunication and in ‘monolingual’ L1 acquisition, and the mech-
anisms of knowledge acquisition are the same, too (Höder, Prentice and Tingsell forthc.).
The major difference is a quantitative one: Speakers are exposed to much less input in
intercommunication than they are in both L1 and AL acquisition. This presumably pro-
motes certain outcomes of knowledge acquisition. As mentioned in Section 5, sound cor-
respondences are likely to be noticed based on their relative frequency as patterns that
reoccur across many low frequent types. The low frequency input at the same time entails
that using sound correspondence constructions can stay an important (perhaps even the
dominant) decoding strategy for certain speakers for a long time or permanently, namely
for those whose exposure to the other language is limited to (sporadic) intercommunica-
The acquisition of sound correspondence constructions was discussed from a purely
theoretical point of view in this paper. The reason for this is a simple one: As of yet, there
remains a lack of empirical research both on the implicit learning of phonological
correspondence rules and on intercommunicative decoding in spoken language contexts
in general. While some promising studies on interlingual decoding have been conducted
in the past (Möller and Zeevaert 2015; Vanhove 2016), these worked with data from ex-
periments on written language comprehension. The results from these studies cannot
simply be transferred to a spoken language context, of course. However, the findings from
Vanhove’s (2016) study on the acquisition of graphemic correspondence rules between
Dutch and German suggest that speakers are indeed able to identify simple and systematic
correspondences on the submorphemic level without explicit instruction and use them
productively in receptive contexts. Future research must show whether the same holds
true for phonological correspondence rules – both simple and more complex ones.
1The visualisation focusses on phonological single word recognition and is highly sim-
plified. Linguistic and extra-linguistic context cues are deliberately not taken into ac-
2 For demonstration purposes, a simple and systematic sound correspondence has been
chosen for the example analysis. However, there are, of course, also more complex cor-
respondences that go beyond simple sound substitution and might include several changes
like, e.g., syllable reduction with lengthening differences.
The ideas discussed in this paper present part of the preliminary considerations to an on-
going PhD-project under the working title “Phonological schematicity in Interscandina-
Bannert, Robert. 1981. “Referat av diskussionen i sektionen Talperceptionsforskning
och nordisk hörförståelse.” In Internordisk språkförståelse: Föredrag och
diskussioner vid ett symposium på Rungstedgaard utanför Köpenhamn den 24-
26 mars 1980, anordnat av Sekretariatet för Nordiskt Kulturellt Samarbete vid
Nordiska Ministerrådet, ed. by Claes-Christian Elert, 37–45. Umeå: Universite-
tet i Umeå.
Braunmüller, Kurt. 1995. “Semikommunikation und semiotische Strategien. Bausteine
zu einem Modell für die Verständigung im Norden zur Zeit der Hanse.” In Nie-
derdeutsch und die skandinavischen Sprachen (Sprachgeschichte 4), vol. 2, ed.
by Kurt Braunmüller, 35–70. Heidelberg: Winter.
Braunmüller, Kurt. 2008. On the Relevance of Receptive Multilingualism in a Global-
ised World: Theory, History and Evidence from Today’s Scandinavia (Arbeiten
zur Mehrsprachigkeit, Folge B, 90). Hamburg: Universität Hamburg, Sonderfor-
Bybee, Joan L. 2010. Language, Usage, and Cognition. Cambridge: Cambridge Univer-
Delsing, Lars-Olof, and Katarina Lundin Åkesson. 2005. Håller språket ihop Norden?
En forskningsrapport om ungdomars förståelse av danska, svenska och norska.
Kopenhagen: Nordiska Ministerrådet.
Ellis, Nick C., and Stefanie Wulff. 2019. “Cognitive Approaches to L2 Acquisition.” In
The Cambridge Handbook of Language Learning, ed. by John W. Schwieter,
and Alessandro G. Benati, 41–61. Cambridge: Cambridge University Press.
Goldberg, Adele. 2019. Explain Me This. Creativity, Competition, and the Partial
Productivity of Constructions. Princeton, Oxford: Princeton University Press.
Gooskens, Charlotte, and Renée van Bezooijen. 2013. “Explaining Danish-Swedish
Asymmetric Word Intelligibility. An Error Analysis.” In Phonetics in Europe:
Perception and Production, ed. by Charlotte Gooskens, and Renée van Bezooi-
jen, 59–82. Frankfurt am Main: Peter Lang.
Hilpert, Martin. 2019. “Higher-order Schemas in Morphology: What They Are, How
They Work, and Where to Find Them.” Word Structure 12.3: 261–273.
Höder, Steffen. 2018. “Grammar is Community-specific: Background and Basic Con-
cepts of Diasystematic Construction Grammar.” In Constructions in Contact:
Constructional Perspectives on Contact Phenomena in Germanic Languages
(Constructional Approaches to Language 24), ed. by Hans C. Boas, and Steffen
Höder, 37–70. Amsterdam: Benjamins.
Höder, Steffen. 2019. “Phonological Schematicity in Multilingual Constructions: A Di-
asystematic Perspective on Lexical Form.” Word Structure 12.3: 334–352.
Höder, Steffen, Julia Prentice, and Sofia Tingsell. Forthc. “Acquisition of Additional
Languages as Reorganization in the Multilingual Constructicon.” In Construc-
tions in Contact 2. Language Change, Multilingual Practices, and Additional
Language Acquisition (Constructional Approaches to Language), ed. by Hans C.
Boas, and Steffen Höder. Amsterdam/Philadelphia: Benjamins.
Klein, Horst G., and Tilbert D. Stegmann. 2000. EuroComRom – Die sieben Siebe. Ro-
manische Sprachen sofort lesen können. 2nd ed. Aachen: Shaker.
Möller, Robert, and Ludger Zeevaert. 2015. “Investigating Word Recognition in Inter-
comprehension: Methods and Findings.” Linguistics 53 (2): 313–352.
Pulvermüller, Friedemann. 1996. “Hebb’s Concept of Cell Assemblies and the Psycho-
physiology of Word Processing.” Psychophysiology 33 (4): 317–333.
Sandøy, Helge. 2005. “The Typological Development of the Nordic Languages I: Pho-
nology.” In The Nordic Languages. An International Handbook of the North
Germanic Languages (Handbooks of Linguistics and Communication Science
22), vol. 2, ed. by Oskar Bandle et al., 1852–1871. Berlin/New York: De Gruy-
van Heuven, Vincent J. 2008. “Making Sense of Strange Sounds: (Mutual) Intelligibil-
ity of Related Language Varieties. A Review.” International Journal of Humani-
ties and Arts Computing 2 (1–2): 39–62.
Vanhove, Jan. 2016. “The Early Learning of Interlingual Correspondence Rules in Re-
ceptive Multilingualism.” International Journal of Bilingualism 20 (5): 580–
Vihman, Marilyn, and William Croft. 2007. “Phonological Development: Toward a
‘Radical’ Templatic Phonology.” Linguistics 45 (4): 683–725.
Weinreich, Uriel. 1954. “Is a Structural Dialectology Possible?” Word 10: 388–400.
Weinreich, Uriel. 1964. Languages in Contact: Findings and Problems. 3rd ed. Lon-
Zeldes, Amir. 2012. Productivity in Argument Selection: From Morphology to Syntax.
Berlin/Boston: De Gruyter.