PosterPDF Available

Neural Agent-based Models To Study Language Contact Using Linguistic Data


Abstract and Figures

In this paper, we propose an outline for linguistic research on language change, as observed in the languages of the world, using neural agent-based models of emergent communication. We describe how such models could be used to study morphological simplification, using a case study of language contact in Eastern Indonesia. A neural architecture is used to represent hypothesized cognitive mechanisms of language change: a generalization mechanism, the procedural/declarative model, and a phonological mechanism, the hyper & hypo articulation model, that involves a theory of mind of the listener.
Content may be subject to copyright.
Neural Agent-based Models to Study Language
Contact using Linguistic Data
Peter Dekker
AI Lab
Vrije Universiteit Brussel
Brussels, Belgium
Bart de Boer
AI Lab
Vrije Universiteit Brussel
Brussels, Belgium
In this paper, we propose an outline for linguistic research on language change,
as observed in the languages of the world, using neural agent-based models of
emergent communication. We describe how such models could be used to study
morphological simplification, using a case study of language contact in Eastern
Indonesia. A neural architecture is used to represent hypothesized cognitive mecha-
nisms of language change: a generalization mechanism, the procedural/declarative
model, and a phonological mechanism, the hyper & hypo articulation model, that
involves a theory of mind of the listener.
1 Introduction
What happens to language, when an established population of speakers of a language, comes into
contact with a community of strangers, who try to learn their language? Agent-based computer
simulations of interactions between speakers can be effective models to study this question [
]. In
this paper, we will outline how agent-based models with a neural model of production and perception
can be used to study linguistic questions about language change, based on real-world data from the
languages of the world. We will provide the desiderata a neural agent model should fulfill to be able to
study these questions, and give a sketch of a possible model, which we plan to implement in the future.
We want to infer general factors behind language change, by seeing how these general factors surface
in case studies on real languages. Central in this paper is a case study of language contact in Eastern
Indonesia. We will study the hypothesis that morphological simplification is caused by contact
between native speakers of a language (L1) and adult learners, who learn the language as a second
language (L2). We use neural networks as an architecture to implement two cognitive mechanisms that
could lead to morphological simplification: a generalization mechanism, the procedural/declarative
model, and a phonological mechanism for simplifying or clarifying utterances, the hyper & hypo
articulation model, that involves a theory of mind of the listener.
2 Previous work
Agent-based models are used in many areas describing social or cultural processes, for example
to describe social segregation or the spread of religion [
]. Agent models have been used to
describe the evolution of language [
] and to describe concrete instances of language change, for
example change in word order in Dutch [
]. More abstract models from language evolution have
been used to study the influence of social factors (like population size and language contact) on
linguistic structure [
]. In addition to this work on agents for linguistic modelling, agent-based
approaches of emergent communication have been developed in the community of natural language
processing (NLP). Agents in these models use an abstract language to designate objects or images
34th Conference on Neural Information Processing Systems (NeurIPS 2020), Vancouver, Canada.
and have to agree on a name for a certain object [
]. In recent agent models from NLP, deep neural
networks are used as the comprehension and production model [
]. Interesting analyses can be
made about the language neural agents learn, such as the frequency distribution of symbols and word
order [
]. [
] model contact between communities of deep neural agents, and the formation of a
creole language. Some of these models ultimately have the goal of constructing conversational agents
in mind. We will use the same types of neural models, but apply them to study linguistic questions
about how real languages change. [
] suggest that among others linguistics and cognitive science
could contribute to hypotheses to test experimentally in neural models of emergent communication.
3 Method
3.1 Data
As a case study, we consider the linguistic situation at the Alor and Pantar islands in Eastern Indonesia.
Alorese, an Austronesian language, is spoken on the coasts of the islands, while landward, Papuan
Alor-Pantar languages are spoken. Many L1 speakers of Alor-Pantar languages, learn Alorese as a
second language. Alorese lost all of its morphology, when compared to closely related Lewoingu
Lamaholot [
], which has not been in contact with Alor-Pantar languages. It is assumed Alorese
lost its morphology due to adult language contact [22].
Figure 1: Case study: morphological simplification in Alorese, spoken on the Alor & Pantar islands in Indonesia.
We will specifically look at the verb morphology of Alorese: compared to Lamaholot, all verb suffixes,
signifying the subject of the sentence on the verb, have been lost. For example, the third person plural
for the verb lodo ‘to go down’ is lodo-ka (with the 3PL marker -ka) in Lewoingu Lamaholot, while
it is lodo (without person marker) in Alorese. In Alorese, only verb prefixes, on a small number
of vowel-initial verbs, have been retained. As initialization of our model, we use verb forms, and
their affixes from a grammar of Lewoingu Lamaholot [
], which can be seen as the starting state of
Alorese, before it underwent morphological simplification. We use 56 verbs, these are the only verbs
in the grammar explicitly classified as prefixing or suffixing: 17 prefixing and 39 suffixing verbs. As
this grammar is descriptive, there is no distributional data over forms: per verb concept, one verb form
and one prefix/suffix per grammatical person is available. We compare the outcome of the language
in the model to the current state of Alorese from a grammar of Alorese [
] and demographic data,
about the proportion of L2 speakers in Alorese communities [
]. For possible future research on the
lexicon, the digital dataset LexiRumah, which includes Alorese and Alor-Pantar languages, could be
used [13].
As data representation, we choose to stay close to the real language: agents communicate using real
word forms, prefixes and suffixes. Every phoneme in these forms and affixes, is represented as a string
of phonetic features. By staying close to the real language, language-specific factors which make the
case study interesting, such as the phonological vowel-initial retention of prefixes, can be included
in the study. However, we want to test the validity of our model to describe general mechanisms
of language change. Therefore, we plan to evaluate our model, with as much as possible the same
method and data representation, on other case studies of morphological simplification, such as
language contact between Scandinavian languages and Low German (possible dataset: NorthEuraLex
3.2 Task
The task for agents is to successfully communicate about concepts in the world, roughly inspired
by a Lewis signalling game [
] or naming game [
]. Every iteration, every agent in a population
speaks: it picks a concept, produces a form based on that concept and sends it to the listener. In
our case study, we look at verb morphology. We want to analyze the interaction between different
grammatical persons (e.g. 1st person singular, 1SG) in the verb paradigm, and we want to look at
transitivity (a verb having an object), because this determines the affixes being used. Therefore, a
concept consists of a combination of a lexical concept (e.g. verb to go), a person (e.g. 1SG) and a
transitivity (containing object or not) of the sentence. The listener tries to infer the concept from the
received form and points to this object. The speaker then points to the correct object. Based on this
feedback, either only the listener, or the speaker and the listener, update their internal model. We
create a simulation with both L1 and L2 agents. We initialize L1 agents with (train on) a concept-form
mapping from Lewoingu Lamaholot, a precursor of Alorese which has not undergone morphological
simplification. The L2 agents, with a neighboring Papuan language as first language, are initialized
with a random model. We do not initialize the agents with a Papuan language model, since the
literature presupposes no L1 effect on morphology, but rather a general effect of L2 learning at an
adult age [
]. By running a model with and without adult language contact, and comparing the
results to the current state of Alorese (where morphological simplification has taken place), we can
test the hypothesis that adult language contact was responsible for morphological simplification.
3.3 Investigating cognitive factors
In our model we will implement two cognitive mechanisms: Ullman’s declarative/procedural model
of language learning [
], and a phonological component, Lindblom’s H&H model [
] (Figure
concept form
Figure 2: Cognitive mechanisms: the declarative/procedural model, used during comprehension and production,
and the H&H model, used during production, re-entering the produced utterance in the comprehension system.
A number of theories account for the differences between L1 and L2 language processing, leading
to morphological simplification during adult language contact, including: a critical threshold age
for learning languages [
], missing surface inflection [
] – which assumes that adult learners have
knowledge of inflection, but cannot realize it –, a noise channel-based approach (where L2 speakers
have less information to decode) [
], and the role of the L1 knowledge when learning L2 [
We choose Ullman’s declarative/procedural model as cognitive mechanism of language learning and
generalization. According to the procedural/declarative model, grammar is produced by a procedural
cognitive system, while the lexicon is memorized in a declarative cognitive system. In L2 learners,
linguistic forms which are normally produced in the procedural system, such as morphology, are
memorized in the declarative system. We hypothesize this computationally heavy memorization
step in L2 speakers leads to simplification. According to the theory, based on age of acquisition and
experience, morphology may be produced more via the procedural system also in L2 speakers. We
do not claim that the declarative/procedural model is the only mechanism at play, for example, some
of the aforementioned mechanisms may play a role, but Ullman’s mechanism is the perspective we
will use to approach our problem.
Furthermore, we add a phonological component, because in the data of our Alorese case study,
morphological simplification is also phonologically conditioned. We use Lindblom’s hyper and
hypo articulation theory: speakers produce a form more clearly or less clearly (e.g., drop an affix),
depending on their estimation of intelligibility by the listener. This requires the speaker to have a
theory of mind about the listener. One option is re-entrance [
]: the speaker interprets its utterance
using its own language comprehension system, as if it was the listener. Another option would be to
take the characteristics of the listener (e.g. L1 or L2) into account, but this assumes that these listener
characteristics are available to the speaker. The H&H articulation component turns the problem of
zero-shot learning upside down. Instead of burdening an L2 listener, who hears a form for the first
time, with the task to infer an concept, the speaker will try to adjust his pronunciation to the L2
3.4 Neural model
We want to implement the proposed cognitive mechanisms (section 3.3) in a neural network, which
serves as a language comprehension and production model of every agent in the simulation. It is a
challenging task to implement cognitive mechanisms in a neural model, since it is not trivial how a
certain cognitive trait (e.g. generalization versus no generalization) can be translated into a neural
network architecture. Therefore, we will propose some possible ideas, and draw from the literature
on neural emergent communication. We want to develop a model that is cognitively plausible: it
should consist of cognitive modules which can be postulated in humans, and the model should be
able to exhibit to some extent human language processing behaviour in an agent-based simulation.
At the same time, the implementation does not have to be neurally plausible: the structure of the
neural network does not have to mimic the structure of the brain. A deep neural network is merely a
powerful and robust model, to be able to implement our cognitive mechanisms. The network will
perform a reinforcement learning task, where the communicative success is the reward. As in [
we think a modular structure of the network can well represent our cognitive mechanisms. A specific
challenge is the relatively small amount of data in our setting, which may call for specific network
architectures or data augmentation.
We will implement a declarative and a procedural module, in both L1 and L2 agents. The procedural
system facilitates generalization over different concepts, grammatical persons and sentence transi-
tivities. The declarative system performs no generalization, but instead memorizes concept-form
mappings. It is an open question how this generalization versus no generalization can be implemented
in a neural network. A possible approach could be to model the procedural model using a smaller
number of nodes, and add dropout and regularization, forcing it to generalize over training examples.
The declarative system could consist of a larger number of nodes and layers, nudging it to overfit.
In L1 agents, there is a stronger bias towards using the procedural module, while L2 agents use the
declarative module more, but are able to shift to the procedural system after gaining more experience.
The weights which modulate procedural versus declarative system usage, should thus be initialized
differently for L1 and L2 agents, but be able to be change by experience.
The other cognitive mechanism, the H&H articulation model, could involve a step of re-entrance:
how would a speaker’s utterance be perceived if the speaker had to interpret it himself? This could
possibly be implemented using a game setting like self-play [
], where an agent plays against itself.
Based on the estimated intelligibility of the utterance, the utterance is produced clearer or sloppier.
This post-processing step of an utterance, can be performed in a separate module of the network.
This module will need to have knowledge of how exactly a word form can be pronounced clearer or
sloppier (e.g. dropping an ending), possibly by pre-training on language data.
4 Conclusion
We have described a dataset, task and model that show how neural models of language production
and comprehension can be used in an agent-based setting to study language change, using data from
languages of the world. We sketched how the declarative-procedural model, responsible for linguistic
generalization, and the hyper & hypo articulation model, which involves a theory of mind through
re-entrance, could be implemented in a neural model. Further research is needed to determine the
precise architectures to implement these mechanisms, specifically in a small data setting. We hope
the techniques proposed help to develop models that can better explain language change, and by
doing this, eventually shed light on human (pre)history.
Acknowledgments and Disclosure of Funding
This work was supported by funding from the Flemish Government under the Onderzoeksprogramma
Artificiële Intelligentie (AI) Vlaanderen programme. PD was supported by a PhD Fellowship funda-
mental research (11A2821N) of the Research Foundation – Flanders (FWO).
BAPTI STA, M., GELMAN, S. A., AND BECK, E. Testing the role of convergence in language
acquisition, with implications for creole genesis. International Journal of Bilingualism 20, 3
(June 2016), 269–296.
BLOEM , J., VER SLOOT, A., AND WEERMAN, F. An agent-based model of a historical word
order change. In Proceedings of the Sixth Workshop on Cognitive Aspects of Computational
Language Learning (Lisbon, Portugal, 2015), Association for Computational Linguistics, pp. 22–
CHAABOUNI, R., KH ARITONOV, E., DUPOUX, E., AND BARONI, M. Anti-efficient encoding
in emergent communication. In Advances in Neural Information Processing Systems (2019),
pp. 6290–6300.
order biases in deep-agent emergent communication. arXiv:1905.12330 [cs] (June 2019).
DAGA N, G., HU PKES, D., AND BRUN I, E. Co-evolution of language and agents in referential
games. arXiv preprint arXiv:2001.03361 (2020).
03n04 (May 2012), 1150017.
DE BOER , B. Self-organization in vowel systems. Journal of Phonetics 28, 4 (Oct. 2000),
J., AND JÄGER , G. NorthEuraLex: A wide-coverage lexical database of Northern Eurasia.
Language Resources and Evaluation (Nov. 2019).
EVTIM OVA, K., DROZDOV, A., KIELA, D., AND CHO , K. Emergent Communication in a
Multi-Modal, Multi-Step Referential Game. arXiv:1705.10369 [cs, math] (Apr. 2018).
FUTRE LL, R., A ND GIBSON, E. L2 processing as noisy channel language comprehension.
Bilingualism: Language and Cognition 20, 4 (Sept. 2016), 683–684.
GRAES SER, L., CHO, K., AND KIEL A, D. Emergent Linguistic Phenomena in Multi-Agent
Communication Games. arXiv:1901.08706 [cs] (Jan. 2019).
IANNACCO NE, L. R., AND MAKOW SKY, M. D. Accidental Atheists? Agent-Based Explana-
tions for the Persistence of Religious Regionalism. Journal for the Scientific Study of Religion
46, 1 (Mar. 2007), 1–16.
KAIPING, G. A., A ND KLAMER, M. LexiRumah: An online lexical database of the Lesser
Sunda Islands. PLOS ONE 13, 10 (Oct. 2018), e0205250.
KLAMER, M. A. F. A Short Grammar of Alorese (Austronesian). No. 486 in Languages of the
World. Materials. LINCOM Europa, Muenchen, 2011.
LAZARIDOU, A., AND BARONI, M. Emergent Multi-Agent Communication in the Deep
Learning Era. arXiv:2006.02419 [cs] (July 2020).
LAZARIDOU, A., PEYSAKHOVICH, A., AND BARONI, M. Multi-Agent Cooperation and the
Emergence of (Natural) Language. arXiv:1612.07182 [cs] (Mar. 2017).
LENNE BERG, E. H. Response to reviews of biological foundations of language. Journal of
Communication Disorders 1, 4 (Oct. 1968), 320–322.
[18] LEWIS, D. Convention. Harvard University Press, Cambridge, MA, 1969.
LINDBLOM, B. Explaining Phonetic Variation: A Sketch of the H&H Theory. In Speech
Production and Speech Modelling, W. J. Hardcastle and A. Marchal, Eds. Springer Netherlands,
Dordrecht, 1990, pp. 403–439.
LOU-MAGNUSON, M., AN D ONNIS, L. Social Network Limits Language Complexity. Cogni-
tive Science 42, 8 (Nov. 2018), 2790–2817.
LOWE , R., GUP TA, A., FOER STER, J., KIEL A, D., AN D PINEAU, J. On the interaction
between supervision and self-play in emergent communication. arXiv:2002.01093 [cs, stat]
(June 2020).
MORO , F. R. Loss of Morphology in Alorese (Austronesian): Simplification in Adult Language
Contact. Journal of Language Contact 12, 2 (Aug. 2019), 378–403.
NISHIYAMA, K., AND KE LEN, H. A Grammar of Lamaholot, Eastern Indonesia. Lincom
Europa, 2007.
PVOST, P., AN D WHITE, L. Missing Surface Inflection or Impairment in second language
acquisition? Evidence from tense and agreement. Second Language Research 16, 2 (Apr. 2000),
REALI , F., CHATER, N., AND CHRISTIANSEN, M. H. Simpler grammar, larger vocabulary:
How population size affects language. Proceedings of the Royal Society B: Biological Sciences
285, 1871 (Jan. 2018), 20172586.
SCHELLING, T. C. Dynamic models of segregation
.The Journal of Mathematical Sociology
1, 2 (July 1971), 143–186.
SCHEPENS, J., VAN DER SLI K, F., AN D VAN HOUT, R. The effect of linguistic distance across
Indo-European mother tongues on learning Dutch as a second language. In Approaches to
Measuring Linguistic Differences, L. Borin and A. Saxena, Eds. DE GRUYTER, Berlin, Boston,
Jan. 2013.
SMITH, A. D. Models of language evolution and change. Wiley Interdisciplinary Reviews:
Cognitive Science 5, 3 (2014), 281–293.
STEEL S, L. Synthesising the origins of language and meaning using co-evolution, self-
organisation and level formation. In Approaches to the Evolution of Language. Cambridge
University Press, 1998, pp. 384–404.
STEEL S, L. Language re-entrance and the ‘inner voice’. Journal of Consciousness Studies 10,
4-5 (2003), 173–185.
ULLMAN, M. T. The neural basis of lexicon and grammar in first and second language:
The declarative/procedural model. Bilingualism: Language and Cognition 4, 2 (Aug. 2001),
ULLMAN, M. T. A neurocognitive perspective on language: The declarative/procedural model.
Nature Reviews Neuroscience 2, 10 (Oct. 2001), 717–726.
ResearchGate has not been able to resolve any citations for this publication.
Full-text available
This article describes the first release version of a new lexicostatistical database of Northern Eurasia, which includes Europe as the most well-researched linguistic area. Unlike in other areas of the world, where databases are restricted to covering a small number of concepts as far as possible based on often sparse documentation, good lexical resources providing wide coverage of the lexicon are available even for many smaller languages in our target area. This makes it possible to attain near-completeness for a substantial number of concepts. The resulting database provides a basis for rich benchmarks that can be used to test automated methods which aim to derive new knowledge about language history in underresearched areas.
Full-text available
This paper discusses historical and ongoing morphological simplification in Alorese, an Austronesian language spoken in eastern Indonesia. From comparative evidence, it is clear that Alorese lost almost all of its morphology over several hundred years as a consequence of language contact (Klamer, 2012, to appear). By providing both linguistic and cultural-historical evidence, this paper shows that Alorese has historically undergone morphological simplification as a result of second language (L2) learning. The first part of the paper presents a case study comparing the use of subject agreement prefixes in Alorese L1 speakers (n=6) and Alorese L2 speakers (n=12). The results show that L2 speakers deviate from the native norm, and tend to use one prefix as default agreement. The variation found among L2 speakers reveals an ongoing change possibly leading to the restructuring of the Alorese agreement system. The second part of the paper applies models of linguistic change (Kusters, 2003; Trudgill, 2011) to the Alorese community and shows that Alorese has been, and still is, spoken in a community with a large number of L2 speakers, where morphological simplification is expected to occur.
Full-text available
The Lesser Sunda Islands in eastern Indonesia cover a longitudinal distance of some 600 kilometres. They are the westernmost place where languages of the Austronesian family come into contact with a family of Papuan languages and constitute an area of high linguistic diversity. Despite its diversity, the Lesser Sundas are little studied and for most of the region, written historical records, as well as archaeological and ethnographic data are lacking. In such circumstances the study of relationships between languages through their lexicon is a unique tool for making inferences about human (pre-)history and tracing population movements. However, the lack of a collective body of lexical data has severely limited our understanding of the history of the languages and peoples in the Lesser Sundas. The LexiRumah database fills this gap by assembling lexicons of Lesser Sunda languages from published and unpublished sources, and making those lexicons available online in a consistent format. This database makes it possible for researchers to explore the linguistic data collated from different primary sources, to formulate hypotheses on how the languages of the two families might be internally related and to compare competing hypotheses about subgroupings and language contact in the region. In this article, we present observations from aggregating lexical data from sources of different type and quality, including fieldwork, and generalize our lessons learned towards practical guidelines for creating a consistent database of comparable lexical items, derived from the design and development of LexiRumah. Databases like this are instrumental in developing theories of language evolution and change in understud-ied regions where small-scale, pre-industrial, pre-literate societies are the majority. It is therefore vital to follow reliable design choices when creating such databases, as described in this paper.
Full-text available
Languages with many speakers tend to be structurally simple while small communities sometimes develop languages with great structural complexity. Paradoxically, the opposite pattern appears to be observed for non-structural properties of language such as vocabulary size. These apparently opposite patterns pose a challenge for theories of language change and evolution. We use computational simulations to show that this inverse pattern can depend on a single factor: ease of diffusion through the population. A population of interacting agents was arranged on a network, passing linguistic conventions to one another along network links. Agents can invent new conventions, or replicate conventions that they have previously generated themselves or learned from other agents. Linguistic conventions are either Easy or Hard to diffuse, depending on how many times an agent needs to encounter a convention to learn it. In large groups, only linguistic conventions that are easy to learn, such as words, tend to proliferate, whereas small groups where everyone talks to everyone else allow for more complex conventions, like grammatical regularities, to be maintained. Our simulations thus suggest that language, and possibly other aspects of culture, may become simpler at the structural level as our world becomes increasingly interconnected.
The thesis in this paper is that L2 speakers differ from L1 speakers in their ability to do memory storage and retrieval about linguistic structure. We would like to suggest it is possible to go farther than this thesis and develop a computational-level theory which explains why this mechanistic difference between L2 and L1 speakers exists. For this purpose, we believe a noisy channel model (Shannon, 1948; Levy, 2008; Levy, Bicknell, Slattery & Rayner, 2009; Gibson, Bergen & Piantadosi, 2013) could be a good start. Under the reasonable assumption that L2 speakers have a less precise probabilistic representation of the syntax of their L2 language than L1 speakers do, the noisy channel model straightforwardly predicts that L2 comprehenders will depend more on world knowledge and discourse factors when interpreting and recalling utterances (cf. Gibson, Sandberg, Fedorenko, Bergen & Kiran, 2015, for this assumption applied to language processing for persons with aphasia). Also, under the assumption that L2 speakers assume a higher error rate than L1 speakers do, the noisy channel model predicts that they will be more affected by alternative parses which are not directly compatible with the form of an utterance.
The main objective of this paper is to test experimentally the role of convergence in language acquisition (second language acquisition specifically), with implications for creole genesis. Although there is ample evidence that similar features in languages in contact are enhanced both in second language acquisition and the creation of new languages, scholars are rarely explicit about the exact nature of that similarity and have not been in a position to observe convergence in progress. Our experiment is unique on two fronts as it is the first to use an artificial language to test the convergence hypothesis by making it observable, and it is also the first experimental study to clarify the notion of similarity by varying the levels and types of similarity that are expressed. We report an experiment with 94 English-speaking adults, designed to test how language learners acquire a new language in which the form and function (meaning) of linguistic items are either similar to or distinct from those in their L1. A miniature artificial language was created that included morphological elements to express negation and pluralization. Participants were randomly assigned to one of three conditions: congruent (form and function of novel grammatical morphemes were highly similar to those in English), reversed (negative grammatical morpheme was highly similar to that of English plural, and plural grammatical morpheme was highly similar to that of English negation), and novel (form and function were highly dissimilar to those of English). For each task, scores were entered into a one-way ANOVA with condition as the between-subjects factor. Participants in the congruent condition performed best, indicating that features that converge across form and function are learned most fully. More surprisingly, results showed that participants in the reversed condition acquired the language more readily than those in the novel condition, contrary to expectation. This experiment contributes to a more fine-grained understanding of what degree of similarity promotes acquisition and language creation and provides evidence that form-function mapping promotes learning over form or function alone; crucially, the study challenges the received assumption that acquiring features that are totally novel may be easier than acquiring "reversed" features whose form maps onto L1 but whose function is distinct from it.
In this article, two accounts of the variable use of inflection in adult second language (L2) acquisition are examined. The Missing Surface Inflection Hypothesis (MSIH) proposes that L2 learners have unconscious knowledge of the functional projections and features underlying tense and agreement. However, learners sometimes have a problem with realization of surface morphology, such that they resort to non-finite forms (e.g. Haznedar and Schwartz, 1997; Prévost and White, 1999). The Impaired Representation Hypothesis (IRH) claims that L2 inflection is essentially impaired, due to lack of functional categories, features or feature strength (e.g. Eubank, 1993/94; Meisel, 1997). These views make different predictions for adult L2 acquisition. Spontaneous production data from two adult learners of French and two adult learners of German are examined. The data show that finite forms do not occur in non-finite contexts, that learners exhibit syntactic reflexes of finiteness and that inflected forms largely show accurate agreement. These results suggest that adult L2 learners represent finiteness and agreement at an abstract level, rather than being impaired in this domain, supporting the MSIH.