Periodic Reporting for period 2 - LeibnizDream (Realizing Leibniz’s Dream: Child Languages as a Mirror of the Mind)
Reporting period: 2022-07-01 to 2023-12-31
The LeibnizDream project seeks to develop a novel approach to linguistic theory that places meaning and child data first. The primary way to accomplish this goal is to test predictions of this approach, called the Meaning First approach, for child languages across the world. Specifically, one prediction is that children should make specific “errors” of commission, meaning that they use a more verbose and redundant linguistic form or employ different linguistic structures than adults to express a thought linguistically.
The planned outcome of the project is a detailed model of the three components of the Meaning First approach: The Generator model captures the human capacity to form complex thoughts structured out of primitive concepts. The Compressor captures the general human capacity to relate a thought structure to a communicative act, such as the articulation of a sentence, as well as how this capacity is deployed in a specific language. The Acquirer component captures how the Generator and Compressor develop during childhood and especially how they interact with the linguistic environment of a child and their cognitive system as they learn a language or multiple languages. The Meaning First Approach will require a new vision of language acquisition which sees the process as one of creation of language to express an intended thought. The model we develop will be tested and adjusted using data from fifty different languages we collect using a global network of collaborators.
Human thoughts are extremely rich and complex. To communicate them, humans need to use language. Language, by its nature, has to compress thoughts by expressing them in a linear order (language unfolds in time), with specific words, in a concise and non-ambiguous way, so that what is communicated is enough for the hearer to recover the intended meaning conveyed by the speaker. The key idea of the project is that children have to yet acquire the specific way of compressing thoughts in their language and may thus express their thoughts in a more transparent and verbose way, through what we call “errors” of commission, that is, linguistic meaningful expressions which contains more linguistic materials than an adult sentence would. Along these lines, we have found that children may sometimes express a redundant verb expressing causation in French in sentences like “on va le faire couper” (we are going to make it cut) meaning “on va le couper” (we are going to cut it). This type of redundancy was observed also in other languages, like Turkish. Children speaking English or German may use two negative elements in the same sentence to express a single negation meaning. For example, they may say “I don’t see nobody” meaning “I saw nobody”. Or, they may produce relative clauses like “The one/child that mommy is combing the child is called John” (rather than “The one/the child that mommy is combing is called John”) using the same noun (or a pro-form) as the subject of the main sentence and as the object of the relative clause. These uses have been observed in a variety of languages, like Spanish, Italian, Mandarin, German, among others. Finally, children in a variety of languages say something equivalent to English “I want ice cream with without chocolate”, to intend “I want ice-cream without chocolate”. Our project intends to investigate through experiments designed for children using interactive activities how children express linguistically certain thoughts and which kind of “linguistic inventions” they resort to. We conjecture that these redundant expressions are closer to the structure of thoughts than the linguistic expressions used by adults, who have learned rules and convention about how one compresses in a given language. Our research extends world-wide, as our goal is not only to study languages like Italian or German, but languages around the world, like Cabacano (spoken in the Philippines), Baasa (spoken in Cameroon), Yoruba, Vietnamese, Thai, Mandarin, Guaraní (spoken in South America). Therefore, we have established collaborations with various researchers around the globe to carry out studies with children in their countries. In fact, since compression is executed differently in different languages and different languages pose different challenges to children, we expect that commission errors will be both similar and different across the various child languages. By comparing child languages across the globe, we will gain a better understanding of the structure of thought. Many children that will participate in our studies will be multilingual, as multilingualism is the norm in most non-Western countries, and is growing in Western countries as well. Since there is cross-linguistic influence in the languages of a multilingual child, we expect additional and unique contribution from multilingual children. For instance, if one language is more transparent in expressing a certain thought than the other language, we may expect children to transfer the more transparent expression to the other language or benefit from transparency in the acquisition of language, with certain expressions being acquired more rapidly in a language than in another. Although our focus are children acquiring one or more languages, adults acquiring second languages may also provide relevant information, as they also have to learn new rules and conventions to compress in the new language. In addition, we study adult language to establish a benchmark against which to interpret child data; therefore, we are also involving adult participants, through online experiments carried out world-wide with the goal to collect quantitative data on various structures of language and establish what is acceptable or preferred by speakers of that language.
In the end, we will have a new model of the relation of thought and language and its acquisition. In addition, we will be able to build a data base of child and adult productions and comprehension from typologically different languages.
Human thought and language are central to humanity's existence. The better understanding of both that we aim to achieve will have great impact for linguistics as well as cognitive science, philosophy, and education, and it also offers potential for novel language technologies.
The planned outcome of the project is a detailed model of the three components of the Meaning First approach: The Generator model captures the human capacity to form complex thoughts structured out of primitive concepts. The Compressor captures the general human capacity to relate a thought structure to a communicative act, such as the articulation of a sentence, as well as how this capacity is deployed in a specific language. The Acquirer component captures how the Generator and Compressor develop during childhood and especially how they interact with the linguistic environment of a child and their cognitive system as they learn a language or multiple languages. The Meaning First Approach will require a new vision of language acquisition which sees the process as one of creation of language to express an intended thought. The model we develop will be tested and adjusted using data from fifty different languages we collect using a global network of collaborators.
Human thoughts are extremely rich and complex. To communicate them, humans need to use language. Language, by its nature, has to compress thoughts by expressing them in a linear order (language unfolds in time), with specific words, in a concise and non-ambiguous way, so that what is communicated is enough for the hearer to recover the intended meaning conveyed by the speaker. The key idea of the project is that children have to yet acquire the specific way of compressing thoughts in their language and may thus express their thoughts in a more transparent and verbose way, through what we call “errors” of commission, that is, linguistic meaningful expressions which contains more linguistic materials than an adult sentence would. Along these lines, we have found that children may sometimes express a redundant verb expressing causation in French in sentences like “on va le faire couper” (we are going to make it cut) meaning “on va le couper” (we are going to cut it). This type of redundancy was observed also in other languages, like Turkish. Children speaking English or German may use two negative elements in the same sentence to express a single negation meaning. For example, they may say “I don’t see nobody” meaning “I saw nobody”. Or, they may produce relative clauses like “The one/child that mommy is combing the child is called John” (rather than “The one/the child that mommy is combing is called John”) using the same noun (or a pro-form) as the subject of the main sentence and as the object of the relative clause. These uses have been observed in a variety of languages, like Spanish, Italian, Mandarin, German, among others. Finally, children in a variety of languages say something equivalent to English “I want ice cream with without chocolate”, to intend “I want ice-cream without chocolate”. Our project intends to investigate through experiments designed for children using interactive activities how children express linguistically certain thoughts and which kind of “linguistic inventions” they resort to. We conjecture that these redundant expressions are closer to the structure of thoughts than the linguistic expressions used by adults, who have learned rules and convention about how one compresses in a given language. Our research extends world-wide, as our goal is not only to study languages like Italian or German, but languages around the world, like Cabacano (spoken in the Philippines), Baasa (spoken in Cameroon), Yoruba, Vietnamese, Thai, Mandarin, Guaraní (spoken in South America). Therefore, we have established collaborations with various researchers around the globe to carry out studies with children in their countries. In fact, since compression is executed differently in different languages and different languages pose different challenges to children, we expect that commission errors will be both similar and different across the various child languages. By comparing child languages across the globe, we will gain a better understanding of the structure of thought. Many children that will participate in our studies will be multilingual, as multilingualism is the norm in most non-Western countries, and is growing in Western countries as well. Since there is cross-linguistic influence in the languages of a multilingual child, we expect additional and unique contribution from multilingual children. For instance, if one language is more transparent in expressing a certain thought than the other language, we may expect children to transfer the more transparent expression to the other language or benefit from transparency in the acquisition of language, with certain expressions being acquired more rapidly in a language than in another. Although our focus are children acquiring one or more languages, adults acquiring second languages may also provide relevant information, as they also have to learn new rules and conventions to compress in the new language. In addition, we study adult language to establish a benchmark against which to interpret child data; therefore, we are also involving adult participants, through online experiments carried out world-wide with the goal to collect quantitative data on various structures of language and establish what is acceptable or preferred by speakers of that language.
In the end, we will have a new model of the relation of thought and language and its acquisition. In addition, we will be able to build a data base of child and adult productions and comprehension from typologically different languages.
Human thought and language are central to humanity's existence. The better understanding of both that we aim to achieve will have great impact for linguistics as well as cognitive science, philosophy, and education, and it also offers potential for novel language technologies.
In its first two years, the project worked primarily with existing child data from databases and on the theoretical basis. Linguistic tests and questionnaires were prepared to be used across languages to ensure comparability of data, and studies were designed. The COVID-19 pandemic made collecting new child language data in person impossible for most of this reporting period, but the project instead developed remote data collection methods using internet-connected tablet computers. It also necessitated some familiarization with the software for handling remote data collection.
The researchers of the three project groups – acquirer group at UniMiB, compressor group at HUBER, and generator group at ZAS – formed around 35 small teams with mixed teams to formulate hypothesis and design experiments in the six areas of the project’s investigation. The teams worked together using video conferencing and a common cloud storage using shared data management standards. The research manager led a review of all teams’ progress.
At the same time, each of the three groups also worked on a common theoretical model. Work on the theoretical basis in morphology in the compressor group focused on the concept of multiple pronunciation in both child and adult language. Work on the theoretical basis in semantics in the generator group focused on developing an initial model of the algebra of thought. Work in the acquirer group focused on development of background tests and questionnaires for many languages to provide measures of general linguistic development for cross-linguistic comparison as well as to investigate heuristic strategies in language acquisition.
The researchers of the three project groups – acquirer group at UniMiB, compressor group at HUBER, and generator group at ZAS – formed around 35 small teams with mixed teams to formulate hypothesis and design experiments in the six areas of the project’s investigation. The teams worked together using video conferencing and a common cloud storage using shared data management standards. The research manager led a review of all teams’ progress.
At the same time, each of the three groups also worked on a common theoretical model. Work on the theoretical basis in morphology in the compressor group focused on the concept of multiple pronunciation in both child and adult language. Work on the theoretical basis in semantics in the generator group focused on developing an initial model of the algebra of thought. Work in the acquirer group focused on development of background tests and questionnaires for many languages to provide measures of general linguistic development for cross-linguistic comparison as well as to investigate heuristic strategies in language acquisition.
We could establish that a class of child language errors of interest to the project could be frequently found in corpora and also elicited in targeted experiments well. We worked on methodology to be able to compare productions of children across languages. We also worked on methodology to compare linguistic data from adults to be used as the target reference for children.