Tag Archives: Noam Chomsky

Linguistics as a Window to Understanding the Brain

The ability to communicate through spoken language may be the trait that best sets humans apart from other animals. Last year researchers identified the first gene implicated in the ability to speak. This week, a team shows that the human version of this gene appears to date back no more than 200,000 years–about the time that anatomically modern humans emerged. The authors argue that their findings are consistent with previous speculations that the worldwide expansion of modern humans was driven by the emergence of full-blown language abilities.

The researchers who identified the gene, called FOXP2, showed that FOXP2 mutations cause a wide range of speech and language disabilities (ScienceNOW, 3 October 2002). In collaboration with part of this team, geneticist Svante Pääbo’s group at the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany, set about tracing the gene’s evolutionary history.

As a uniquely human trait, language has long baffled evolutionary biologists. Not until FOXP2was linked to a genetic disorder that caused problems in forming words could they even begin to study language’s roots in our genes. Soon after that discovery, a team at the Max Planck Institute for Evolutionary Anthropology in Leipzig, Germany, discovered that just two bases, the letters that make up DNA, distinguished the human and chimp versions ofFOXP2. To try to determine how those changes influenced the gene’s function, that group put the human version of the gene in mice. In 2009, they observed that these “humanized” mice produced more frequent and complex alarm calls, suggesting the human mutations may have been involved in the evolution of more complex speech.

When humanized mice and wild mice were put in mazes that engaged both types of learning,the humanized mice mastered the route to the reward faster than their wild counterparts, report Schreiweis, Graybiel, and their colleagues

The results suggest the human version of the FOXP2 gene may enable a quick switch to repetitive learning—an ability that could have helped infants 200,000 years ago better communicate with their parents. Better communication might have increased their odds of survival and enabled the new version of FOXP2 to spread throughout the entire human population, suggests Björn Brembs, a neurobiologist at the University of Regensburg in Germany, who was not involved with the work.

“The findings fit well with what we already knew about FOXP2 but, importantly, bridge the gap between behavioral, genetic, and evolutionary knowledge,” says Dianne Newbury, a geneticist at the Wellcome Trust Centre for Human Genetics in Oxford, U.K., who was not involved with the new research. “They help us to understand how the FOXP2 gene might have been important in the evolution of the human brain and direct us towards neural mechanisms that play a role in speech and language acquisition.”

Chomsky critiqued the field of AI for adopting an approach reminiscent of behaviorism, except in more modern, computationally sophisticated form. Chomsky argued that the field’s heavy use of statistical techniques to pick regularities in masses of data is unlikely to yield the explanatory insight that science ought to offer. For Chomsky, the “new AI” — focused on using statistical learning techniques to better mine and predict data — is unlikely to yield general principles about the nature of intelligent beings or about cognition.

 

Published on Oct 6, 2012

Steven Pinker – Psychologist, Cognitive Scientist, and Linguist at Harvard University

How did humans acquire language? In this lecture, best-selling author Steven Pinker introduces you to linguistics, the evolution of spoken language, and the debate over the existence of an innate universal grammar. He also explores why language is such a fundamental part of social relationships, human biology, and human evolution. Finally, Pinker touches on the wide variety of applications for linguistics, from improving how we teach reading and writing to how we interpret law, politics, and literature.

The Floating University

 

 

Chimpanzee Genome Project

The Chimpanzee Genome Project is an effort to determine the DNA sequence of the Chimpanzee genome. It is expected that by comparing the genomes of humans and other apes, it will be possible to better understand what makes humans distinct from other species from a genetic perspective.

Human and chimpanzee chromosomes are very similar. The primary difference is that humans have one fewer pair of chromosomes than do other great apes. Humans have 23 pairs of chromosomes and other great apes have 24 pairs of chromosomes. In the human evolutionary lineage, two ancestral ape chromosomes fused at their telomeres, producing human chromosome 2.[3] There are nine other major chromosomal differences between chimpanzees and humans: chromosome segment inversions on human chromosomes 1, 4, 5, 9,12, 15, 16, 17, and 18. After the completion of the Human genome project, a common chimpanzee genome project was initiated. In December 2003, a preliminary analysis of 7600 genes shared between the two genomes confirmed that certain genes such as theforkhead-box P2 transcription factor, which is involved in speech development, are different in the human lineage. Several genes involved in hearing were also found to have changed during human evolution, suggesting selection involving human language-related behavior. Differences between individual humans and common chimpanzees are estimated to be about 10 times the typical difference between pairs of humans.[4]

About 600 genes have been identified that may have been undergoing strong positive selection in the human and chimp lineages; many of these genes are involved in immune system defense against microbial disease (example: granulysin is protective against Mycobacterium tuberculosis [8]) or are targeted receptors of pathogenic microorganisms (example: Glycophorin C and Plasmodium falciparum). By comparing human and chimp genes to the genes of other mammals, it has been found that genes coding fortranscription factors, such as forkhead-box P2 (FOXP2), have often evolved faster in the human relative to chimp; relatively small changes in these genes may account for the morphological differences between humans and chimps. A set of 348 transcription factor genes code for proteins with an average of about 50 percent more amino acid changes in the human lineage than in the chimp lineage.

Six human chromosomal regions were found that may have been under particularly strong and coordinated selection during the past 250,000 years. These regions contain at least one marker allele that seems unique to the human lineage while the entire chromosomal region shows lower than normal genetic variation. This pattern suggests that one or a few strongly selected genes in the chromosome region may have been preventing the random accumulation of neutral changes in other nearby genes. One such region on chromosome 7 contains the FOXP2 gene (mentioned above) and this region also includes the Cystic fibrosis transmembrane conductance regulator (CFTR) gene, which is important for ion transport in tissues such as the salt-secreting epithelium of sweat glands. Human mutations in the CFTR gene might be selected for as a way to survivecholera.[9]

Another such region on chromosome 4 may contain elements regulating the expression of a nearby protocadherin gene that may be important for brain development and function. Although changes in expression of genes that are expressed in the brain tend to be less than for other organs (such as liver) on average, gene expression changes in the brain have been more dramatic in the human lineage than in the chimp lineage.[10] This is consistent with the dramatic divergence of the unique pattern of human brain development seen in the human lineage compared to the ancestral great ape pattern. The protocadherin-beta gene cluster on chromosome 5 also shows evidence of possible positive selection.[11]

Results from the human and chimp genome analyses should help in understanding some human diseases. Humans appear to have lost a functional caspase-12 gene, which in other primates codes for an enzyme that may protect against Alzheimer’s disease.

The results of the chimpanzee genome project suggest that when ancestral chromosomes 2A and 2B fused to produce human chromosome 2, no genes were lost from the fused ends of 2A and 2B. At the site of fusion, there are approximately 150,000 base pairs of sequence not found in chimpanzee chromosomes 2A and 2B. Additional linked copies of the PGML/FOXD/CBWD genes exist elsewhere in the human genome, particularly near the p end of chromosome 9. This suggests that a copy of these genes may have been added to the end of the ancestral 2A or 2B prior to the fusion event. It remains to be determined if these inserted genes confer a selective advantage.

  • PGML. The phosphoglucomutase-like gene of human chromosome 2. This gene is incomplete and may not produce a functional transcript.[12]
  • FOXD. The forkhead box D4-like gene is an example of an intronless gene. The function of this gene is not known, but it may code for a transcription control protein.
  • CBWD. Cobalamin synthetase is a bacterial enzyme that makes vitamin B12. In the distant past, a common ancestor to mice and apes incorporated a copy of a cobalamin synthetase gene (see: Horizontal gene transfer). Humans are unusual in that they have several copies of cobalamin synthetase-like genes, including the one on chromosome 2. It remains to be determined what the function of these human cobalamin synthetase-like genes is. If these genes are involved in vitamin B12 metabolism, this could be relevant to human evolution. A major change in human development is greater post-natal brain growth than is observed in other apes. Vitamin B12is important for brain development, and vitamin B12 deficiency during brain development results in severe neurological defects in human children.
  • CXYorf1-like protein. Several transcripts of unknown function corresponding to this region have been isolated. This region is also present in the closely related chromosome 9p terminal region that contains copies of the PGML/FOXD/CBWD genes.
  • Many ribosomal protein L23a pseudogenes are scattered through the human genome.

The origin of language in the human species has been the topic of scholarly discussions for several centuries. In spite of this, there is no consensus on the ultimate origin or age of human language. One problem makes the topic difficult to study: the lack of direct evidence. Consequently, scholars wishing to study the origins of language must draw inferences from other kinds of evidence such as the fossil record, archaeological evidence, contemporary language diversity, studies of language acquisition, and comparisons between human language and systems of communication existing among other animals (particularly other primates). Many argue that the origins of language probably relate closely to the origins of modern human behavior, but there is little agreement about the implications and directionality of this connection.

This shortage of empirical evidence has led many scholars to regard the entire topic as unsuitable for serious study. In 1866, the Linguistic Society of Paris banned any existing or future debates on the subject, a prohibition which remained influential across much of the western world until late in the twentieth century.[1] Today, there are numerous hypotheses about how, why, when, and where language might have emerged.[2] Despite this, there is scarcely more agreement today than a hundred years ago, when Charles Darwin‘s theory of evolution by natural selection provoked a rash of armchair speculations on the topic.[3] Since the early 1990s, however, a number of linguists, archaeologists,psychologists, anthropologists, and others have attempted to address with new methods what some consider “the hardest problem in science.”[4]

Noam Chomsky, a prominent proponent of discontinuity theory, argues that a single chance mutation occurred in one individual in the order of 100,000 years ago, instantaneously installing the language faculty (a component of the mind-brain) in “perfect” or “near-perfect” form.[6] According to this view, emergence of language resembled the formation of a crystal; with digital infinity as the seed crystal in a super-saturated primate brain, on the verge of blossoming into the human mind, by physical law, once evolution added a single small but crucial keystone.[7][8] It follows from this theory that language appeared rather suddenly within the history of human evolution.

A majority of linguistic scholars as of 2015 hold continuity-based theories, but they vary in how they envision language development. Among those who see language as mostly innate, some — notably Steven Pinker[9] — avoid speculating about specific precursors in nonhuman primates, stressing simply that the language faculty must have evolved in the usual gradual way.[10] Others in this intellectual camp — notably Ib Ulbæk[5] — hold that language evolved not from primate communication but from primate cognition, which is significantly more complex.

Those who see language as a socially learned tool of communication, such as Michael Tomasello, see it developing from the cognitively controlled aspects of primate communication, these being mostly gestural as opposed to vocal.[11][12] Where vocal precursors are concerned, many continuity theorists envisage language evolving from early human capacities for song.[13][14][15][16]

Transcending the continuity-versus-discontinuity divide, some scholars view the emergence of language as the consequence of some kind of social transformation[17] that, by generating unprecedented levels of public trust, liberated a genetic potential for linguistic creativity that had previously lain dormant.[18][19][20] “Ritual/speech coevolution theory” exemplifies this approach.[21][22] Scholars in this intellectual camp point to the fact that even chimpanzees and bonobos have latent symbolic capacities that they rarely – if ever – use in the wild.[23] Objecting to the sudden mutation idea, these authors argue that even if a chance mutation were to install a language organ in an evolving bipedal primate, it would be adaptively useless under all known primate social conditions. A very specific social structure — one capable of upholding unusually high levels of public accountability and trust — must have evolved before or concurrently with language to make reliance on “cheap signals” (words) an evolutionarily stable strategy.

Because the emergence of language lies so far back in human prehistory, the relevant developments have left no direct historical traces; neither can comparable processes be observed today. Despite this, the emergence of new sign languages in modern times — Nicaraguan Sign Language, for example — may potentially offer insights into the developmental stages and creative processes necessarily involved.[24] Another approach inspects early human fossils, looking for traces of physical adaptation to language use.[25][26] In some cases, when the DNA of extinct humans can be recovered, the presence or absence of supposedly language-relevant genes — FOXP2, for example — may prove informative.[27] Another approach, this time archaeological, involves invoking symbolic behavior (such as repeated ritual activity) that may leave an archaeological trace — such as mining and modifying ochre pigments for body-painting — while developing theoretical arguments to justify inferences from symbolism in general to language in particular.[28][29][30]

The time range for the evolution of language and/or its anatomical prerequisites extends, at least in principle, from the phylogenetic divergence of Homo (2.3 to 2.4 million years ago) from Pan (5 to 6 million years ago) to the emergence of full behavioral modernity some 150,000 – 50,000 years ago. Few dispute that Australopithecus probably lacked vocal communication significantly more sophisticated than that of great apes in general,[31] but scholarly opinions vary as to the developments since the appearance of Homosome 2.5 million years ago. Some scholars assume the development of primitive language-like systems (proto-language) as early as Homo habilis, while others place the development of symbolic communication only with Homo erectus (1.8 million years ago) or with Homo heidelbergensis (0.6 million years ago) and the development of language proper with Homo sapiens, currently estimated at less than 200,000 years ago.

Using statistical methods to estimate the time required to achieve the current spread and diversity in modern languages, Johanna Nichols — a linguist at the University of California, Berkeley — argued in 1998 that vocal languages must have begun diversifying in our species at least 100,000 years ago.[32] A further study by Q. D. Atkinson[14]suggests that successive population bottlenecks occurred as our African ancestors migrated to other areas, leading to a decrease in genetic and phenotypic diversity. Atkinson argues that these bottlenecks also affected culture and language, suggesting that the further away a particular language is from Africa, the fewer phonemes it contains. By way of evidence, Atkinson claims that today’s African languages tend to have relatively large numbers of phonemes, whereas languages from areas in Oceania (the last place to which humans migrated), have relatively few. Relying heavily on Atkinson’s work, a subsequent study has explored the rate at which phonemes develop naturally, comparing this rate to some of Africa’s oldest languages. The results suggest that language first evolved around 350,000-150,000 years ago, which is around the time when modern Homo sapiensevolved.[33] Estimates of this kind are not universally accepted but genetic, archaeological, palaeontological and much other evidence has led to a near-consensus that language probably emerged somewhere in sub-Saharan Africa during the Middle Stone Age, roughly contemporaneous with the speciation of Homo sapiens.[34]

Evolutionary linguistics is the scientific study of the psychosocial development and cultural evolution of individual languages as well as the origins and development of human language itself.[1] The main challenge in this research is the lack of empirical data: spoken language leaves practically no traces. This led to an abandonment of the field for more than a century. Since the late 1980s, the field has been revived in the wake of progress made in the related fields of psycholinguistics, neurolinguistics, evolutionary anthropology, evolutionary psychology, universal grammar, and cognitive science.

Evolutionary linguistics as a field is rapidly emerging as a result of developments in neighboring disciplines. To what extent language’s features are determined by genes, a hotly debated dichotomy in linguistics, has had new light shed upon it by the discovery of the FOXP2 gene.[7][8] An English family with a severe, heritable language dysfunction was found to have a defective copy of this gene.[9][10] Mutations of the corresponding gene in mice (FOXP2 is fairly well conserved; modern humans share the same allele as Neanderthals)[11][12] cause reductions in size and vocalization rate. If both copies are damaged, the Purkinje layer (a part of the cerebellum that contains better-connected neurons than any other) develops abnormally, runting is more common, and pups die within weeks due to inadequate lung development.[13] Additionally, higher presence of FOXP2 in songbirds is correlated to song changes, with downregulation causing incomplete and inaccurate song imitation in zebra finches. In general, evidence suggests that the protein is vital to neuroplasticity. There is little support, however, for the idea that FOXP2 is ‘the grammar gene’ or that it had much to do with the relatively recent emergence of syntactical speech.[14]

Another controversial dichotomy is the question of whether human language is solely human or on a continuum with (admittedly far removed) animal communication systems. Studies in ethology have forced researchers to reassess many claims of uniquely human abilities for language and speech. For instance, Tecumseh Fitch has argued that the descended larynx is not unique to humans. Similarly, once held uniquely human traits such as formant perception, combinatorial phonology and compositional semantics are now thought to be shared with at least some nonhuman animal species. Conversely, Derek Bickerton and others argue that the advent of abstract words provided a mental basis for analyzing higher-order relations, and that any communication system that remotely resembles human language utterly relies on cognitive architecture that co-evolved alongside language.

Forkhead box protein P2 (FOXP2) is a protein that, in humans, is encoded by the FOXP2 gene, also known as CAGH44,SPCH1 or TNRC10, and is required for proper development of speech and language.[1] Initially identified as the genetic factor of speech disorder in KE family, its gene is the first gene discovered associated with speech and language.[2] The gene is located on chromosome 7 (7q31, at the SPCH1 locus), and is expressed in fetal and adult brain, heart, lung and gut.[3][4] FOXP2 orthologs[5] have also been identified in other mammals for which complete genome data are available. The FOXP2 protein contains a forkhead-box DNA-binding domain, making it a member of the FOX group of transcription factors, involved in regulation of gene expression. In addition to this characteristic forkhead-box domain, the protein contains a polyglutamine tract, a zinc finger and a leucine zipper. The gene is more active in females than in males, to which could be attributed better language learning in females.[6]

In humans, mutations of FOXP2 cause a severe speech and language disorder.[1][7] Versions of FOXP2 exist in similar forms in distantly related vertebrates; functional studies of the gene in mice[8] and in songbirds[9] indicate that it is important for modulating plasticity of neural circuits.[10] Outside the brain FOXP2 has also been implicated in development of other tissues such as the lung and gut.[11]

FOXP2 is popularly dubbed the “language gene”, but this is only partly correct since there are other genes involved in language development.[12] It directly regulates a number of other genes, including CNTNAP2, CTBP1, and SRPX2.[13][14]

Two amino acid substitutions distinguish the human FOXP2 protein from that found in chimpanzees,[15] but only one of these two changes is unique to humans.[11] Evidence from genetically manipulated mice[16] and human neuronal cell models[17] suggests that these changes affect the neural functions of FOXP2.

FOXP2 is required for proper brain and lung development. Knockout mice with only one functional copy of the FOXP2 gene have significantly reduced vocalizations as pups.[28] Knockout mice with no functional copies of FOXP2 are runted, display abnormalities in brain regions such as the Purkinje layer, and die an average of 21 days after birth from inadequate lung development.[11]

FOXP2 is expressed in many areas of the brain[15] including the basal ganglia and inferior frontal cortex where it is and is essential for brain maturation and speech and language development.[13]

A knockout mouse model has been used to examine FOXP2’s role in brain development and how mutations in the two copies of FOXP2affect vocalization. Mutations in one copy result in reduced speech while abnormalities in both copies cause major brain and lung developmental issues.[11]

The expression of FOXP2 is subject to post-transcriptional regulation, particularly micro RNA, which binds to multiple miRNA binding-sites in the neocortex, causing the repression of FOXP2 3’UTR.[29]

Creatures Of Self Destruction?

Published on Oct 4, 2013

http://www.singularityweblog.com/noam…

Dr. Noam Chomsky is a famed linguist, political activist, prolific author and recognized public speaker, who has spent the last 60 years living a double life — one as a political activist and another as a linguist. His activism allegedly made him the US government’s public enemy number one. As a linguist he is often credited for dethroning behaviorism and becoming the “father of modern linguistics” (and/or cognitive science). Put together his accomplishments are the reasons why he is often listed as one of the most important intellectuals of the 20th century.

During our 28 minute conversation with Noam Chomsky we cover a variety of interesting topics such as: the balance between his academic and his political life; artificial intelligence and reverse engineering the human brain; why in his view both Deep Blue and Watson are little more than PR; the slow but substantial progress of our civilization; the technological singularity…

Minimalist Program

In linguistics, the Minimalist Program (MP) is a major line of inquiry that has been developing inside generative grammar since the early 1990s, starting with a 1993 paper by Noam Chomsky.[1]

Chomsky presents MP as a program, not as a theory, following Imre Lakatos‘s distinction.[2] The MP seeks to be a mode of inquiry characterized by the flexibility of the multiple directions that its minimalism enables. Ultimately, the MP provides a conceptual framework used to guide the development of grammatical theory. For Chomsky, there are minimalist questions, but the answers can be framed in any theory. Of all these questions, the one that plays the most crucial role is this: why language has the properties it has.[3] The MP lays out a very specific view of the basis of syntactic grammar that, when compared to other formalisms, is often taken to look very much like a theory.

The MP appeals to the idea that the language ability in humans shows signs of being incorporated under an optimal design with exquisite organization, which seems to suggest that the inner workings conform to a very simple computational law or a particular mental organ. In other words, the MP works on the assumption that Universal Grammar constitutes a perfect design in the sense that it contains only what is necessary to meet our conceptual and physical (phonological) needs.[4]

From a theoretical standpoint, and in the context of generative grammar, the MP draws on the minimalist approach of the Principles and Parameters program, considered to be the ultimate standard theoretical model that generative linguistics has developed since the 1980s. What this approach suggests is the existence of a fixed set of principles valid for all languages, which, when combined with settings for a finite set of binary switches (parameters), may describe the specific properties that characterize the language system a child eventually comes to attain.[5]

The MP aims to get to know how much of the Principles and Parameters model can be taken as a result of this hypothetical optimal and computationally efficient design of the human language faculty. In turn, more developed versions of the Principles and Parameters approach provide technical principles from which the MP can be seen to follow.[6]

A major development of MP inquiry is Bare Phrase Structure (BPS), a theory of phrase structure (sentence building prior to movement) developed by Noam Chomsky.[9] Interestingly, the introduction of BPS has moved the Chomskyan tradition toward the dependency grammar tradition, which operates with significantly less structure than most phrase structure grammars.[10]

This theory contrasts with X-bar theory, which preceded it, in four important ways:

  1. BPS is explicitly derivational. That is, it is built from the bottom up, bit by bit. In contrast, X-Bar Theory is representational—a structure for a given construction is built in one fell swoop, and lexical items are inserted into the structure.
  2. BPS does not have a preconceived phrasal structure, while in X-Bar Theory, every phrase has a specifier, a head, and a complement.
  3. BPS permits only binary branching, while X-Bar Theory permits both binary and unary branching.
  4. BPS does not distinguish between a “head” and a “terminal”, while some versions of X-Bar Theory require such a distinction.

BPS incorporates two basic operations: Merge and Move. Although there is active debate on exactly how Move should be formulated, the differences between the current proposals are relatively minute. The following description follows Chomsky’s original proposal.

Merge is a function that takes two objects (say α and β) and merges them into an unordered set with a label (either α or β, in this case α). The label identifies the properties of the phrase.

Merge (α, β) → {α, {α, β} }

For example, Merge can operate on the lexical items ‘drink’ and ‘water’ to give ‘drink water’. Note that the phrase ‘drink water’ behaves more like the verb ‘drink’ than like the noun ‘water’. That is, wherever we can put the verb ‘drink’ we can usually put the phrase ‘drink water’:

I like to _____________ (drink)/(drink water).
(Drinking/Drinking water) __________ is fun.

Furthermore, we typically can’t put the phrase ‘drink water’ in places where we can put the noun ‘water’:

We can say “There’s some water on the table”, but not “There’s some drink water on the table”.

So, we identify the phrase with a label. In the case of ‘drink water’, the label is ‘drink’ since the phrase acts as a verb. For simplicity, we call this phrase a verb phrase or VP. Now if we were to Merge ‘cold’ and ‘water’ to get ‘cold water’, then we would have a noun phrase or NP with the label ‘water’. The reader can verify that the phrase ‘cold water’ can appear in the same environments as the noun ‘water’ in the three test sentences above. So, for ‘drink water’ we have the following:

Merge (drink, water) → {drink, {drink, water} }

We can represent this in a typical syntax tree as follows:

  • Minimalist Tree Drink Water.png

or, with more technical terms, as:

  • Minimalist Tree VP.png

Merge can also operate on structures already built. If it couldn’t, then such a system would predict only two-word utterances to be grammatical. Say we Merge a new head with a previously formed object (a phrase).

Merge (γ, {α, {α, β}}) → {γ, {γ, {α, {α, β}}}}

Here, γ is the label, so we say that γ ‘projects’ from the label of the head. This corresponds to the following tree structure:

  • Minimalist Syntax Tree 1.png

Note crucially that Merge operates blindly, projecting labels in all possible combinations. The subcategorization features of the head then license certain label projections and eliminate all derivations with alternate projections.

Merge (usually capitalized) is one of the basic operations in the Minimalist Program, a leading approach to generative syntax, when two syntactic objects are combined to form a new syntactic unit (a set). Merge also has the property of recursion in that it may apply to its own output: the objects combined by Merge are either lexical items or sets that were themselves formed by Merge. This recursive property of Merge has been claimed to be a fundamental characteristic that distinguishes language from other cognitive faculties. As Noam Chomsky (1999) puts it, Merge is “an indispensable operation of a recursive system … which takes two syntactic objects A and B and forms the new object G={A,B}” (p. 2).[1]

In some variants of the Minimalist Program Merge is triggered by feature checking, e.g. the verb eat selects the noun cheesecake because the verb has an uninterpretable N-feature [uN] (“u” stands for “uninterpretable”), which must be checked (or deleted) due to full interpretation.[2] By saying that this verb has a nominal uninterpretable feature, we rule out such ungrammatical constructions as *eat beautiful (the verb selects an adjective). Schematically it can be illustrated as:

  
          V
  ________|_________
 |                 |
eat [V, uN]   cheesecake [N]

Chomsky (2001) distinguishes between external and internal Merge: if A and B are separate objects then we deal with external Merge; if either of them is part of the other it is internal Merge.[3]

In other approaches to generative syntax, such as Head-driven phrase structure grammar, Lexical functional grammar and other types of unification grammar, the analogue to Merge is the unification operation of graph theory. In these theories, operations over attribute-value matrices (feature structures) are used to account for many of the same facts. Though Merge is usually assumed to be unique to language, the linguists Jonah Katz and David Pesetsky have argued that the harmonic structure of tonal music is also a result of the operation Merge.[4]

This notion of ‘merge’ may in fact be related to Fauconnier’s ‘blending’ notion in cognitive linguistics.

Standard Merge (i.e. as it is commonly understood) encourages one to adopt three key assumptions about the nature of syntactic structure and the faculty of language: 1) sentence structure is generated bottom up in the mind of speakers (as opposed to top down or left to right), 2) all syntactic structure is binary branching (as opposed to n-ary branching) and 3) syntactic structure is constituency-based (as opposed to dependency-based). While these three assumptions are taken for granted for the most part by those working within the broad scope of the Minimalist Program, other theories of syntax reject one or more of them.

Merge is commonly seen as merging smaller constituents to greater constituents until the greatest constituent, the sentence, is reached. This bottom-up view of structure generation is rejected by representational (=non-derivational) theories (e.g. Generalized Phrase Structure Grammar, Head-Driven Phrase Structure Grammar, Lexical Functional Grammar, most dependency grammars, etc.), and it is contrary to early work in Transformational Grammar. The phrase structure rules of context free grammar, for instance, were generating sentence structure top down.

Merge is usually assumed to merge just two constituents at a time, a limitation that results in tree structures in which all branching is binary. While the strictly binary branching structures have been argued for in detail,[5] one can also point to a number of empirical considerations that cast doubt on these strictly binary branching structures, e.g. the results of standard constituency tests.[6] For this reason, most grammar theories outside of Government and Binding Theory and the Minimalist Program allow for n-ary branching.

Merge merges two constituents in such a manner that these constituents become sister constituents and are daughters of the newly created mother constituent. This understanding of how structure is generated is constituency-based (as opposed to dependency-based). Dependency grammars (e.g. Meaning-Text Theory, Functional Generative Description, Word grammar) disagree with this aspect of Merge, since they take syntactic structure to be dependency-based.[7]