Journal of Logic, Language and Information (2024) 33:9–20

https://doi.org/10.1007/s10849-023-09409-x

Assessing the Strengths and Weaknesses of Large

Language Models

Shalom Lappin1,2,3

Accepted: 16 October 2023 / Published online: 11 November 2023

© The Author(s) 2023

Abstract
The transformers that drive chatbots and other AI systems constitute large language
models (LLMs). These are currently the focus of a lively discussion in both the scien-
tific literature and the popular media. This discussion ranges from hyperbolic claims
that attribute general intelligence and sentience to LLMs, to the skeptical view that
these devices are no more than “stochastic parrots”. I present an overview of some of the
weak arguments that have been presented against LLMs, and I consider several of the
more compelling criticisms of these devices. The former significantly underestimate
the capacity of transformers to achieve subtle inductive inferences required for high
levels of performance on complex, cognitively significant tasks. In some instances,
these arguments misconstrue the nature of deep learning. The latter criticisms identify
significant limitations in the way in which transformers learn and represent patterns
in data. They also point out important differences between the procedures through
which deep neural networks and humans acquire knowledge of natural language. It is
necessary to look carefully at both sets of arguments in order to achieve a balanced
assessment of the potential and the limitations of LLMs.

Keywords Deep learning · Transformers · Artifical intelligence · Natural language

processing

1 Introduction
The introduction of transformers (Vaswani et al., 2017) with multiple attention heads,
and pre-trained with large scale word embeddings, has revolutionised NLP. They have

B Shalom Lappin
[email protected]

1 School of Electronic Engineering and Compuer Science, Queen Mary University of London,
London, England
2 Centre for Linguistic Theory and Studies in Probability, University of Gothenburg, Gothenburg,
Sweden
3 Department of Informatics, King’s College London, London, England

123
10 S. Lappin

yielded near or above human performance on a variety of core NLP tasks, which
include, among others, machine translation, natural language generation, question
answering, image captioning, text summarisation, and natural language inference
(NLI). Transformers operate as autoregressive token predictors (GPT-1-GPT4, Ope-
nAI), or as bidirectional token predictors for masked positions in contexts, such as
BERT (Devlin et al., 2019). They significantly outperform sequential deep learning
networks, like Long Short Term Memory Recurrent Neural Networks (LSTMs) and
Convolutional Neural Networks (CNNs), on most natural language tasks, and for many
non-linguistic applications, such as image recognition.
Transformers provide the LLMs that drive chatbots. The rapid success of these bots
in generating extended sequences of coherent, human like discourse in response to
prompts has produced vigorous debate in both the scientific literature and the popular
media. Some of this discussion consists of exaggerated claims on the capabilities of
LLMs. Other comments offer pat dismissals of these systems as nothing more than
artificial parrots repeating training data. It is important to consider LLMs in a critical
and informed way, in order to understand their abilities and their limitations.
In Sect. 2 I take up several of the more prominent weak arguments that have been
brought against LLMs. These include
(i) the view that they simply return their training data,
(ii) the claim that they cannot capture linguistic meaning due to the absence of seman-
tic grounding,
(iii) the assertion that they do not acquire symbolic representations of knowledge, and
(iv) the statement that they do not learn in the way that humans do.
In Sect. 3 I consider some of the strong arguments concerning the limitations of
LLMs. These involve
(i) important constraints on LLMs as sources of insight into human cognitive pro-
cesses,
(ii) the lack of robustness in LLM performance on NLI tasks,
(iii) the unreliability of LLMs as a source of factually sound information,
(iv) the inability of LLMs to identify universal patterns characteristic of natural lan-
guages,
(v) the consequences of the large data required to train LLMs, for control of the
architecture and development of these systems, and
(vi) the opactiy of these systems.
Section 4 draws conclusions concerning the capacities and limitations of LLMs. It
suggests possible lines of future research in deep learning in light of this discussion.

2 Weak Arguments Against LLMs

2.1 Generalisation, Innovation, and Semantic Grounding

A common criticism of LLMs is that they do little more than synthesise elements
of their training data to produce the most highly valued response to a prompt, as
determined by their probability distribution over the prompt and the data. Bender et

123
Assessing the Strengths and Weaknesses of Large Language Models 11

al. (2021) and Chomsky et al. (2023) offer recent versions of this view. This claim
is false, given that transformers exhibit subtle and sophisticated pattern identification
and inferencing
This inductive capacity permits them to excel at medical image analysis and diag-
nostics (Shamshad et al., 2023). Transformers have revolutionised computational
biology by predicting properties of proteins and new molecular structures (Chandra et
al., 2023). This has opened the way for the use of deep learning for the development
of new medications and clinical treatments.
Dasgupta et al. (2023) use Reinforcement Learning (RL) to train artificial agents
with LLMs to respond appropriately to complex commands in a simulated visual
environment. These commands do not correspond to information or commands in
their training set.1
Bender and Koller (2020) argue that LLMs cannot capture meaning because they
are not semantically grounded, by virtue of the fact that their word embeddings are
generated entirely from text. Hence they cannot identify speakers’ references to objects
in the world or recognise communicative intensions. Sahlgren and Carlsson (2021),
Piantadosi and Hill (2022), Sørgaard (2023) reply to this argument by observing that
learning the distributional properties of words in text does provide access to central
elements of interpretation. These properties specify the topology of lexical meaning.
Even if one accepts Bender and Koller’s view that grounding is a necessary element
of interpretation, it does not establish their claim that LLMs are unable to represent
interpretations of natural language expressions. It is possible to construct multi-modal
word embeddings that capture the distributional patterns of expressions in visual and
other modalities. Multi-Model BERT (Lu et al., 2019) and GPT-4 (OpenAI, 2023)
are pre-trained on such word embeddings. Multi-modal transformers can identify
elements of a graphic image and respond to questions about them. The dialogue in
Fig. 1 from (OpenAI, 2023) illustrates the capacity of an LLM to reason about a
complex photographic sequence, and to identify the humour in its main image.
Bender and Koller (2020) do raise important questions about what a viable compu-
tational model of meaning and interpretation must achieve. It has stimulated a fruitful
debate on this issue. The classical program of formal semantics (Davidson, 1967;
Montague, 1974) seeks to construct a recursive definition of a truth predicate that
entails the truth conditions of the declarative sentences of a language. Lappin (2021)
observes that a generalised multi-modal deep neural network (DNN) achieves a major
part of this program by pairing a suitable graphic (and other modes of) representation
with a sentence that describes a situation.

2.2 Symbolic Representations, Grammars, and Hybrid Systems

Marcus (2022) and Chomsky et al. (2023) maintain that LLMs are defective because
they do not represent the symbolic systems, specifically grammars, which humans
acquire to express linguistic knowledge. They regard grammars as the canonical form
for expresing linguistic knowledge.

1 See Lappin (2021) for discussion of the problem of generalising to new linguistic input in deep learning.

123
12 S. Lappin

Fig. 1 ChatGPT-4 Multi-Modal Dialogue, from OpenAI (2023)

This claim is question begging. It assumes what is at issue in exploring the nature of
human learning. It is entirely possible that humans acquire and encode knowledge of
language in non-symbolic, distributed representations of linguistic (and other) regular-
ities, rather than through symbolic algebraic systems. Smolensky (1987), McClelland
(2016), among others, suggests a view of this kind. Some transformers implicitly iden-
tify significant amounts of hierarchical syntactic structure and long distance relations
(Lappin, 2021; Goldberg, 2019; Hewitt & Manning, 2019; Wilcox et al., 2023; Lasri,
2023). In fact Lappin (2021) and Baroni (2023) argue that DNNs can be regarded as
alternative theoretical models of linguistic knowledge.
Marcus (2022) argues that to learn as effectively as humans do DNNs must incor-
porate symbolic rule-based components, as well as the layers of weighted units that
process vectors through the functions that characterise deep learning. It is widely
assumed that such hybrid systems will substantially enhance the performance of DNNs
on tasks requiring complex linguistic knowledge. In fact, this is not obviously the case.
Tree DNNs incorporate syntactic structure into a DNN, either directly through its
architecture, or indirectly through its training data and knowledge distillation. Socher
et al. (2011), Bowman et al. (2016), Yogatama et al. (2017), Choi et al. (2018), Williams
et al. (2018), Maillard et al. (2019), Ek et al. (2019) consider LSTM-based Tree DNNs.
These have been applied to NLP tasks like sentiment analysis, NLI, and the predic-
tion of human sentence acceptability judgments. In general they do not significantly

123
Assessing the Strengths and Weaknesses of Large Language Models 13

improve performance relative to the non-tree counterpart model, and, in at least one
case (Ek et al., 2019), performance was degraded in comparison with the non-enriched
models.2
More recent work has incorporated syntactic tree structure into transformers like
BERT, and applied these systems to a broader range of tasks (Sachan et al., 2021; Bai
et al., 2021). The experimental evidence on the extent to which this addition improves
the capacity of the transformer to handle these tasks remains unclear. Future work
may well show that hybrid systems of the sort that Tree DNNs represent do offer
genuine advantages over their non-enriched counterparts. The results achieved with
these systems to date have not yet motivated this claim.

2.3 Humans Don’t Learn Like That

Marcus (2022) and Chomsky et al. (2023) claim that DNNs do not capture the way
in which humans achieve knowledge of their language. In fact this remains an open
empirical question. We do not yet know enough about human learning to exclude
strong parallels between the ways in which DNNs and humans acquire and represent
linguistic knowledge, and other types of information.
But even if the claim is true, it need not constitute a flaw in DNN design. These
are engineering devices for performing NLP (and other) tasks. Their usefulness is
evaluated on the basis of their success in performing these tasks, rather than on the
way in which they achieve these results. From this perspective, criticising DNNs on the
grounds they do not operate like humans is analogous to objecting to aircraft because
they do not fly like birds.
Moreover, the success of transformers across a broad range of NLP tasks does have
cognitive significance. These devices demonstrate one way in which the knowledge
required to perform these tasks can be obtained, even if it does not correspond to the
procedures that humans apply. These results have important consequences for debates
over the types of biases that are, in principle, needed for language acquisition and
other kinds of learning.

3 Strong Arguments Against LLMs

3.1 LLMs as Models of Human Learning

Piantadosi (2023) claims that LLMs provide a viable model of human language acqui-
sition. He suggests that they provide evidence against Chomsky’s domain specific
innatist view, which posits a “language faculty” that encodes a Universal Grammar.3
This is due to their ablity to learn implicit representations of syntactic structure and
lexical semantics without strong linguistic learning biases.

2 See Lappin (2021) for discussion of LSTM Tree DNNs.

3 See Lappin and Shieber (2007), Clark and Lappin (2011) for earlier discussions of domain specific, strong
bias views of language learning, and arguments against them. These arguments are based on considerations
from computational learning theory.

123
14 S. Lappin

In Sect. 2.3 I argued that the fact that LLMs may learn and represent information
differently than humans does not entail a flaw in their design. Moreover they do
contribute insight into cognitive issues concerning language acquisition by indicating
what can, in principle, be learned from available data, with the the inductive procedures
that DNNs apply to this input. However, Piantadosi’s claims go well beyond the
evidence. His conclusions seem to imply that humans do, in fact, learn in the way
that DNNs do. This is not obviously the case.
Moreoever, although we do not know precisely how humans acquire and represent
their linguistic knowledge, we do know that they learn with far less data than DNNs
require. As Warstadt and Bowman (2023) observe, LLMs are trained on orders of
magnitude more data than human learners have access to. Humans also require inter-
action in order to achieve knowledge of their language (Clark, 2023), while LLMs are
trained non-interactively, with the possible exception of RL as a partial simulation of
human feedback.
Warstadt and Bowman argue convincingly that in order to assess the extent to which
deep learning corresponds to human learning it is necessary to restrict LLMs to the
sort of data that humans use in language acquisition. This involves reconfiguring both
the training data and the learning environment for DNNs, to simulate the language
learning situation that humans encounter. Only an experimental context of this kind
will illuminate the extent to which there is an analogy between deep and human
learning.

3.2 Problems with Robust NLI

Transformers have scored well on natural language inference benchmarks. However,

they are easily derailed, and their performance can be reduced to close to chance by
adversarial testing involving lexical substitutions. Talman and Chatzikyriakidis (2019)
show that BERT does not generalise well to new data sets for NLI. Conversely Talman
et al. (2021) report that BERT continues to achieve high scores when fine tuned and
tested on corrupted data sets containing nonsense sentence pairs. These results suggest
that BERT is not learning inference through semantic relations between premisses and
conclusions. Instead it appears to be identifying certain lexical and structural patterns
in the inference pairs.
Dasgupta et al. (2022) points out that humans also frequently make mistakes in
inference. However, their reasoning abilities are more robust, even under adversarial
testing. Human performance declines more gracefully than transformers under gener-
alisation challenges, and it is more sharply degraded by nonsense pairs. Dasgupta et al.
(2022) discuss work that pre-trains models on abstract reasoning templates to improve
their performance in NLI. It is not clear to what extent the natural language inference
abilities of transformers constitute more than superficial inductive generalisation over
labelled patterns in their training corpora. Mahowald et al. (2023) review substantial
amounts of evidence for the claim that LLMs do not perform well on extra-linguistic
reasoning and real world knowledge tasks.

123
Assessing the Strengths and Weaknesses of Large Language Models 15

3.3 LLMs Hallucinate

LLMs are notorious for hallucinating plausible sounding narratives that have no factual
basis. In a particularly dramatic example of this phenomenon ChatGPT recently went
to court as a legal expert. A lawyer representing a passenger on Avianca Airline brought
a lawsuit against the airline citing 6 legal precedents as part of the evidence for his
case (Weiser, 2023). The judge was unable to verify any of the precedents. When the
judge demanded an explanation, the lawyer admitted to using ChatGPT-3 to identify
the legal precedents that he required. He went on to explain that he “even asked the
program to verify that the cases were real. It had said yes.”
The fact that LLMs do not reliably distinguish fact from fiction makes them danger-
ous sources of (mis)information. Notice that semantic grounding through multi-modal
training does not, in itself, solve this problem. The images, sounds, etc. to which
multi-modal transformers key text do not insure that the text is factually accurate. The
non-linguistic representations may also be artificially generated. A description of the
image of a unicorn may accurately describe that image. It does not characterise an
animal in the world.

3.4 Linguistic Universals

Natural languages display universal patterns, or tendencies, involving word order, mor-
phology, phonology, and lexical semantic classes. Many of these can be expressed as
conditional probabilities specifying that a property will hold with a certain likelihood,
given the presence of other features in the language.
Gibson et al. (2019) and Kågebäck et al. (2020) suggest that information theoretic
notions of communicative efficiency can explain many of these universals. This type
of efficiency involves optimising the balance between brevity of expression and com-
plexity of content. As there are alternative strategies for achieving such optimisations,
languages will exhibit different clusters of properties.
Chaabouni et al. (2019) report experiments showing that LSTM communication
networks display a preference for anti-efficient encoding of information in which the
most frequent expressions are the longest, rather than the shortest. They experiment
with additional learning biases to promote DNN preference for more efficient commu-
nication systems. Similarly, Lian et al. (2021) describe LSTM simulations in which
the network tends to preserve the distribution patterns observed in the training data,
rather than to maximise efficiency of communication.
If this tendency carries over to transformers, then they will be unable to distinguish
between input from plausible and implausible linguistic communication systems. They
will not recognise the class of likely natural languages. Therefore, they will not provide
insight into the information theoretic biases that shape natural languages.

3.5 Large Data and the Control of Deep Learning Architecture

Transformers require vast amounts of training data for their word and multi-modal
embeddings. Each significant improvement in performance on a range of tasks is
generally driven by a substantial increase in training data, and an expansion in the

123
16 S. Lappin

size of the LLM. While GPT-2 has 1.5 billion parameters, GPT-3 has 175 billion, and
GPT-4 is thought to be 6 times larger than GPT-3.
Only large tech companies have the computing capacity, infrastructure, and funds to
develop and train transformers of this size. This concentrates the design and develop-
ment of LLMs in a very limited number of centres, to the exclusion of most universities
and smaller research agencies. As a result, there are limited resources for research on
alternative architectures for deep learning which focuses on issues that are not central
to the economic concerns of tech companies. Many (most?) researchers working out-
side of these companies effectively become clients using their products, which they
apply to AI tasks through fine tuning and minor modifications. This is an unhealthy
state of affairs. While enormous progress has been made on deep learning models, over
a relatively short period of time, it has been largely restricted to a narrow dimension
of tasks in NLP. In particular, work on the relation of DNNs to human learning and
representation is increasingly limited. Also, examination of learning from small data
with more transparent and agile systems is not a major issue in current research on
deep learning.
It is important to note that Reinforcement Learning does not alleviate the need for
large data, and massive LLMs. RL can significantly improve the performance of trans-
formers across a large variety of tasks (Dasgupta et al., 2023). It can also facilitate zero,
one, and few shot learning, where a DNN performs well on a new task with limited or
no prior training. However, it does not eliminate the need for large amounts of training
data. These are still required for pre-trained word and multi-modal embeddings.

3.6 LLMs are Opaque

Transformers, and DNNs in general, are largely opaque systems, for which it is difficult
to identify the procedures through which they arrive at the patterns that they recognise.
This is, in large part, because the functions, like ReLU, that they apply to activate their
units are non-linear. Autoregressive generative language models also use softmax to
map their output vectors into probability distributions.
These functions cause the vectors that a DNN produces to be, in the general case,
non-compositional. This is due to the fact that the representations of the input and
the output vectors cannot be represented by a homomorphic mapping operation. A
mapping f : A → B from group A to group B is a homomorphism iff for every
vi , v j ∈ A, and the group operation ·, f (A · B) = f (A) · f (B). As a result, it is
not always possible to predict the composite vectors that the units of a transformer
generate from their inputs, or to reconstruct these inputs from the output vectors in a
uniform and regular way.
Probes (Hewitt & Manning, 2019), and selective de-activation of units and attention
heads (Lasri, 2023) can provide insight into the structures that transformers infer.
These methods remain indirect, and they do not fully illuminate the way in which
transformers learn and represent patterns from data.
Bernardy and Lappin (2023) propose Unitary Recurrent Networks (URNs) to solve
the problem of model opacity. URNs apply multiplication to orthogonal matrices.
The matrices that they generate are strictly compositional. These models are fully
transparent, and all input is recoverable from the output at each phase in a URN’s

123
Assessing the Strengths and Weaknesses of Large Language Models 17

processing sequence. They achieve good results for deeply nested bracket matching
tasks in Dyck languages, a class of artificial context-free languages. They do not
perform well on context-sensitive cross serial dependencies in artificial languages, or
on agreement in natural languages. One of their limitations is the use of truncation
of matrices to reduce the size of their rows. This is necessary to facilitate efficient
computation of matrix multiplication, but it degrades the performance of a URN on
the tasks to which it is applied.

4 Conclusions and Future Research

LLMs are not simply “stochastic parrots” synthesising fluent sounding responses to
prompts from previously observed training data. They achieve a sophisticated level
of inductive learning and inference, with transferable skills, across a wide range of
tasks. They are able to identify hierarchical syntactic structure and complex semantic
relations. Through reinforcement learning on multi-modal input they can be trained
to respond appropriately to new questions and commands out of the domain of their
training data. This involves significant generalisation and few shot learning. Work on
LLMs has yielded dramatic progress across a broad set of AI problems, in a compar-
atively short period of time. It far exceeds the achievements of symbolic rule-based
AI over many decades.
It is not clear to what extent LLMs illuminate human cognitive abilities in the
areas of language learning and linguistic representation. While they surpass human
performance on many cognitively interesting NLP tasks, they require far more data
for language learning than humans do. It is not obvious that they learn and encode
linguistic knowledge in the way that humans perform these operations.
LLMs are far from human abilities in natural language inference, analogical rea-
soning, and interpretation, particularly for figurative language. Their performance in
domain general dialogue, while appearing to be fluent, remains informationally lim-
ited, and frequently unreliable. They do not distinguish fact from fiction, but freely
generate inaccurate claims. They also do not optimise informational efficiency in
communication.
The large scale of training data and model size that LLMs require has created a
situation in which large tech companies control the design and development of these
systems. This has skewed research on deep learning in a particular direction, and
disadvantaged scientific work on machine learning with a different orientation.
LLMs are opaque in the way that they generalise from data, which poses serious
problems of explainability. At present we can only understand their inference proce-
dures and knowledge representations indirectly, through probes, and through selective
oblation of heads and other units in the network.
These conclusions suggest the following lines of future research on deep learning.
To compare LLMs to human learners it is necessary to modify the data to which they
have access, and to alter their training regimen. This will permit us to examine the
extent to which there are correspondences and disanalogies between the two learn-
ing processes. It will also be necessary to study the internal procedures applied by
each type of learner, computational for LLMs and neurological for humans, more

123
18 S. Lappin

closely to identify the precise mechanisms that drive inference, generalisation, and
representation, for each kind of acquisition.
It would be useful to experiment with additional learning biases for LLMs to see
if these biases will improve their capacity for robust NLI, and for communicative
efficiency. This work may provide deeper understanding of what is involved in both
abilities. If it is successful, it will produce more intelligent and effective DNNs, which
are better able to handle complex NLP tasks.
It is imperative that we develop procedures for testing the factual accuracy of the text
produced by LLMs. Without them we are exposed to the threat of disinformation on an
even larger scale than we are currently encountering in bot saturated social media. The
need to combat disinformation patterns together with the urgency of filtering racial,
ethnic, religious, and gender bias in AI systems powered by LLMs, that are used
in decision making applications (hiring, lending, university admission, etc). These
research concerns should be a focus of public policy discussion.
Developing smaller, more lightweight models that can be trained on less data
would encourage more work on alternative architectures, among a larger number
of researchers, distributed more widely across industrial and academic centres. This
would facilitate the pursuit of more varied scientific objectives in the area of machine
learning.
Finally, designing and implementing fully transparent DNNs will improve our
understanding of both artificial and human learning. Scientific insight of this kind
should be no less a priority than the engineering success driving current LLMs. Ulti-
mately, good engineering depends on a solid scientific understanding of the principles
embodied in the systems that the engineering creates.
Acknowledgements Many of the ideas proposed here were developed in discussion with Richard Sproat.
I am grateful to him for his contribution to my thinking on these issues. I am also indebted to Katrin Erk,
Steven Piantadosi, and two anonymous reviewers for valuable remarks and suggestions on previous drafts.
An earlier version of this paper was presented in the Cognitive Science Seminar of the School of Electronic
Engineering and Computer Science at Queen Mary University of London, on June 21, 2023. I thank the
audience of this talk for helpful questions and comments. I bear sole responsibility for any errors that remain.
My research reported in this paper was supported by grant 2014-39 from the Swedish Research Council,
which funds the Centre for Linguistic Theory and Studies in Probability (CLASP) in the Department of
Philosophy, Linguistics, and Theory of Science at the University of Gothenburg.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence,
and indicate if changes were made. The images or other third party material in this article are included
in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If
material is not included in the article’s Creative Commons licence and your intended use is not permitted
by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the
copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

References
Bai, J., Wang, Y., Chen, Y., Yang, Y., Bai, J., Yu, J., & Tong, Y. (2021). Syntax-BERT: Improving pre-trained
transformers with syntax trees. In: Proceedings of the 16th Conference of the European Chapter
of the Association for Computational Linguistics, (pp. 3011–3020). Association for Computational
Linguistics

123
Assessing the Strengths and Weaknesses of Large Language Models 19

Baroni, M. (2023). On the proper role of linguistically oriented deep net analysis in linguistic theorising.
In S. Lappin & J.-P. Bernardy (Eds.), Algebraic Structures in Natural Language (pp. 1–16). D: CRC
Press.
Bender, E. M., & Koller, A. (2020). Climbing towards NLU: On meaning, form, and understanding in
the age of data. In: Proceedings of the 58th Annual Meeting of the Association for Computational
Linguistics, (pp. 5185–5198). Association for Computational Linguistics,
Bender, E. M., Gebru, T., McMillan-Major, A., & Shmitchell, S. (2021). On the dangers of stochastic
parrots: Can language models be too big? FAccT ’21, (pp. 610–623). Association for Computing
Machinery
Bernardy, J.-P. & S. Lappin (2023).: Unitary recurrent networks. In S. Lappin & J.-P. Bernardy (Eds.),
Algebraic Structures in Natural Language (pp. 243–277). CRC Press.
Bowman, S. R., Gauthier, J., Rastogi, A., Gupta, R., Manning, C. D., & Potts, C. (2016). A fast unified
model for parsing and sentence understanding. In Proceedings of the 54th Annual Meeting of the
Association for Computational Linguistics (Volume 1: Long Papers), (pp. 1466–1477). Association
for Computational Linguistics, Berlin, Germany.
Chaabouni, R., Kharitonov, E., Dupoux, E., & Baroni, M. (2019). Anti-Efficient Encoding in Emergent
Communication. Curran Associates Inc.
Chandra, A., Tünnermann, L., Löfstedt, T., & Gratz, R. (2023). Transformer-based deep learning for pre-
dicting protein properties in the life sciences. eLife, 12.
Choi, J., Yoo, K. M., & Lee, S.-g. (2018). Learning to compose task-specific tree structures. In AAAI
Conference on Artificial Intelligence.
Chomsky, N., Roberts, I., & Watumull, J. (2023). The false promise of chatgpt. 2023.
Clark, E. (2023). Language is acquired in interaction. In S. Lappin & J.-P. Bernardy (Eds.), Algebraic
Structures in Natural Language (pp. 77–93). CRC Press.
Clark, A., & Lappin, S. (2011). Linguistic Nativism and the Poverty of the Stimulus. Wiley-Blackwell.
Dasgupta, I., Kaeser-Chen, C., Marino, K., Ahuja, A., Babayan, S., Hill, F., & Fergus, R. (2023). Collabo-
rating with language models for embodied reasoning. arXiv:2302.00763.
Dasgupta, I., Lampinen, A. K., Chan, S. C. Y., Creswell, A., Kumaran, D., McClelland, J. L., & Hill, F.
(2022). Language models show human-like content effects on reasoning. arXiv:2207.0705.
Davidson, D. (1967). Truth and meaning. Synthese, 17(1), 304–323.
Devlin, J., Chang, M.-W., Lee, K., & Toutanova, K. (2019). BERT: Pre-training of deep bidirectional trans-
formers for language understanding. In: Proceedings of the 2019 Conference of the North American
Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1
(Long and Short Papers), (pp. 4171–4186). Association for Computational Linguistics
Ek, A., Bernardy, J.-P., & Lappin, S. (2019). Language modeling with syntactic and semantic representation
for sentence acceptability predictions. In Proceedings of the 22nd Nordic Conference on Computational
Linguistics, , (pp. 76–85).
Gibson, E., Futrell, R., Piantadosi, S. P., Dautriche, I., Mahowald, K., Bergen, L., & Levy, R. (2019). How
efficiency shapes human language. Trends in Cognitive Sciences, 23(5), 38–407.
Goldberg, Y. (2019). Assessing bert’s syntactic abilities. arXiv:abs/1901.05287.
Hewitt, J., & Manning, C. D. (2019). A structural probe for finding syntax in word representations. In Pro-
ceedings of the 2019 Conference of the North American Chapter of the Association for Computational
Linguistics: Human Language Technologies, Volume 1 (Long and Short Papers), (pp. 4129–4138).
Association for Computational Linguistics, Minneapolis, Minnesota.
Kågebäck, M., Carlsson, E., Dubhashi, D., & Sayeed, A. (2020). A reinforcement-learning approach to
efficient communication. PLoS ONE 2020.
Lappin, S. (2021). Deep Learning and Linguistic Representation. CRC Press.
Lappin, S., & Shieber, S. (2007). Machine learning theory and practice as a source of insight into universal
grammar. Journal of Linguistics, 43, 393–427.
Lasri, K. (2023). Linguistic Generalization in Transformer-Based Neural Language Models. unpublished
PhD thesis
Lian, Y., Bisazza, A., & Verhoef, T. (2021). The effect of efficient messaging and input variability on neural-
agent iterated language learning. In Proceedings of the 2021 Conference on Empirical Methods in
Natural Language Processing, (pp. 10121–10129). Association for Computational Linguistics
Lu, J., Batra, D., Parikh, D., & Lee, S. (2019). ViLBERT: pretraining task-agnostic visiolinguistic represen-
tations for vision-and-language tasks. In Proceedings of the 33rd International Conference on Neural
Information Processing Systems, (pp. 13–23).

123
20 S. Lappin

Maillard, J., Clark, S., & Yogatama, D. (2019). Jointly learning sentence embeddings and syntax with
unsupervised tree-lstms. Natural Language Engineering, 25(4), 433–449.
Marcus, G. (2022). Deep learning alone isn’t getting us to human-like AI. Noema, August 112022.
McClelland, J. L. (2016). Capturing gradience, continuous change, and quasi-regularity in sound, word,
phrase, and meaning. In B. MacWhinney & W. O’Grady (Eds.), The Handbook of Language Emergence
(pp. 54–80). John Wiley and Sons.
Montague, R. (1974). Formal Philosophy: Selected Papers of Richard Montague. Yale University Press.
Edited with an introduction by R. H. Thomason.
Mahowald, K., Ivanova, A. A., Blank, I. A., Kanwisher, N., Tenenbaum, J. B., Fedorenko, E. (2023). Dis-
sociating language and thought in large language models: a cognitive perspective. arXiv:2301.06627
[cs.CL].
OpenAI. (2023). GPT-4 technical report. arXiv:2303.08774.
Piantadosi, S. (2023). Modern language models refute chomsky’s approach to language. Lingbuzz Preprint,
lingbuzz 7180.
Piantadosi, S. T., & Hill, F. (2022). Meaning without reference in large language models. arXiv:2208.02957.
Sachan, D. S., Zhang, Y., Qi, P., & Hamilton, W. (2021). Do syntax trees help pre-trained transformers extract
information? In: Proceedings of the 16th Conference of the European Chapter of the Association for
Computational Linguistics, (pp. 2647–2661). Association for Computational Linguistics
Sahlgren, M., & Carlsson, F. (2021). The singleton fallacy: Why current critiques of language models miss
the point. Frontiers in Artificial Intelligence, 4.
Shamshad, F., Khan, S., Zamir, S. W., Khan, M. H., Hayat, M., Khan, F. S., & Fu, H. (2023). Transformers
in medical imaging: A survey. Medical Image Analysis, 88.
Smolensky, P. (1987). Connectionist AI, symbolic AI, and the brain. Artificial Intelligence Review, 1(2),
95–109.
Socher, R., Pennington, J., Huang, E. H., Ng, A. Y., & Manning, C. D. (2011). Semi-supervised recur-
sive autoencoders for predicting sentiment distributions. In Proceedings of the 2011 Conference on
Empirical Methods in Natural Language Processing, (pp. 151–161). Association for Computational
Linguistics, Edinburgh, Scotland, UK.
Sørgaard, A. (2023). Grounding the vector space of an octopus: Word meaning from raw text. Minds and
Machines, 33(1), 33–54.
Talman, A., & Chatzikyriakidis, S. (2019). Testing the generalization power of neural network models across
NLI benchmarks. In Proceedings of the 2019 ACL Workshop BlackboxNLP: Analyzing and Interpreting
Neural Networks for NLP, (pp. 85–94). Association for Computational Linguistics, Florence, Italy.
Talman, A., Apidianaki, M., Chatzikyriakidis, S., & Tiedemann, J. (2021). NLI data sanity check: Assessing
the effect of data corruption on model performance. In Proceedings of the 23rd Nordic Conference
on Computational Linguistics (NoDaLiDa), (pp. 276–287). Linköping University Electronic Press,
Sweden, Reykjavik, Iceland (Online).
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I.
(2017). Attention is all you need. In I. Guyon, U. V. Luxburg, S. Bengio, H. Wallach, R. Fergus, S.
Vishwanathan, & R. Garnett (Eds.), Advances in Neural Information Processing Systems. (Vol. 30).
Red Hook, NY: Curran Associates Inc.
Warstadt, A., & Bowman, S. R. (2023). What artificial neural networks can tell us about human language
acquisition. In S. Lappin & J.-P. Bernardy (Eds.), Algebraic Structures in Natural Language (pp.
17–59). CRC Press.
Weiser, B. (2023). Here’s what happens when your lawyer uses ChatGPT. 2023.
Wilcox, E. G., Gauthier, J., Hu, J., Qian, P., & Levy, R. (2023). Learning syntactic structures from string
input. In S. Lappin & J. .-P. Bernardy (Eds.), Algebraic Structures in Natural Language (pp. 113–137).
CRC Press.
Williams, A., Drozdov, A., & Bowman, S. R. (2018). Do latent tree learning models identify meaningful
structure in sentences? Transactions of the Association for Computational Linguistics, 6, 253–267.
Yogatama, D., Blunsom, P., Dyer, C., Grefenstette, E., & Ling, W. (2017). Learning to compose words into
sentences with reinforcement learning. In 5th International Conference on Learning Representations,
ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Proceedings.

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps
and institutional affiliations.

123