Language by exposure?

[allhandsondex]

With a reasonably good grasp of German (Low C1 in speaking and writing, low C2 in comprehension), I moved to Belgium for work and one thing led to another and I found myself living in Leuven, in Flanders. Apart from my Ps and Qs and the first eight lessons of Assimil's Dutch with Ease, I've never really bothered trying to learn any Dutch, just kept my eyes and ears open a lot and spoken English when I need to do something more than say thanks to the cashiers in supermarkets.

I've been here now since September, so 9 months in total, and to my surprise find myself reading newspapers and letters and joining in conversations without much difficulty at all. I can't write well, but I'm sure that would come with some practice. Today just out of interest I had a go at interpreting some beginners speeches from Speech Repository from Dutch into English, and found that while I could cope with the Flemish speakers just fine, I really struggled to understand speakers of Netherlands Dutch, probably because I've never been very exposed to much of their variant.

Does anyone else have any experience of something similar to this - moving to a place that speaks a language closely related to one you already speak and learning it almost by accident?

[sillygoose1]

Never lived in another country, but that's basically how I learned Portuguese. After learning the grammar, I jumped right into native materials. I tried Assimil, but it was way too easy for me. Even before I started getting into Portuguese, just with my other Romance knowledge I could read basic newspaper articles and some light novels with relative ease.

[allhandsondex]

Never lived in another country, but that's basically how I learned Portuguese. After learning the grammar, I jumped right into native materials. I tried Assimil, but it was way too easy for me. Even before I started getting into Portuguese, just with my other Romance knowledge I could read basic newspaper articles and some light novels with relative ease.

Hm, maybe that should be my next project... once the Romanian is decent

[Lugubert]

Does anyone else have any experience of something similar to this - moving to a place that speaks a language closely related to one you already speak and learning it almost by accident?

In 1967, at the age of 24, I managed to get a trainee job at Shell in Amsterdam in their physics lab. (I was studying oil chemistry in Sweden.) When I arrived there, I knew little more Dutch than deciphering chocolate box labels. School English (8 years), German (6) and French (4). Fortunately, I was set up in a family where nobody knew any other languages than Dutch, except for a daughter of 15 who had some school English. So I just had to get by in Dutch to express my food likes and dislikes, when I might disappear for a round on the town and possibly return etc.

Not much Dutch training on job, of course, because with one notable exception, everyone wanted to practice their English. For a few weeks, I totally mixed up German and Dutch.

Then, another trainee, a Czech girl, arrived at the department. She knew no English or Dutch, but had very good German, so everyone approved when I volunteered to interpret between English and German. From that day, when I used both Dutch and German extensively every day, it quickly got easier to keep the languages apart. After two months, I managed most everyday situations, like shopping, getting a haircut, finding my way etc. in Dutch without resorting too much to English.

[reineke]

Has anyone learnt a language well without trying to learn it?

viewtopic.php?f=14&t=3061

Grammar through massive exposure

viewtopic.php?f=14&t=4571&hilit=Grammar+through+massive+exposure


Learning a closely related language

viewtopic.php?f=17&t=4814&hilit=Studying+related+languages

[reineke]

The Now-or-Never bottleneck: A fundamental constraint on language

Abstract: "Memory is fleeting. New material rapidly obliterates previous material. How, then, can the brain deal successfully with the
continual deluge of linguistic input? We argue that, to deal with this “Now-or-Never” bottleneck, the brain must compress and
recode linguistic input as rapidly as possible. This observation has strong implications for the nature of language processing:
(1) the language system must “eagerly” recode and compress linguistic input;
(2) as the bottleneck recurs at each new representational level, the language system must build a multilevel linguistic representation; and
(3) the language system must deploy all available
information predictively to ensure that local linguistic ambiguities are dealt with “Right-First-Time”; once the original input is lost,
there is no way for the language system to recover. This is “Chunk-and-Pass” processing. Similarly, language learning must also occur
in the here and now, which implies that language acquisition is learning to process, rather than inducing, a grammar... This approach promises to create a direct relationship between psycholinguistics and linguistic theory.

1. Introduction
Language is fleeting. As we hear a sentence unfold, we rapidly lose our memory for preceding material. Speakers, too, soon lose track of the details of what they have just said. Language processing is therefore “Now-or-Never”: If linguistic information is not processed rapidly, that information is lost for good.

The existence of a Now-or-Never bottleneck is relatively uncontroversial, although its precise character may be debated. However, in this article we argue that the consequences of this constraint for language are remarkably far-reaching, touching on the following issues:
1. The multilevel organization of language into sound-based units, lexical and phrasal units, and beyond;
2. The prevalence of local linguistic relations (e.g., in phonology and syntax);
3. The incrementality of language processing;
4. The use of prediction in language interpretation and production;
5. The nature of what is learned during language acquisition;
6. The degree to which language acquisition involves item-based generalization;
7. The degree to which language change proceeds itemby- item;
8. The connection between grammar and lexical knowledge;
9. The relationships between syntax, semantics, and pragmatics.

2. The Now-or-Never bottleneck
Language input is highly transient. Speech sounds, like other auditory signals, are short-lived. Classic speech perception studies have shown that very little of the auditory trace remains after 100 ms (Elliott 1962), with more recent studies indicating that much acoustic information already is lost after just 50 ms (Remez et al. 2010). Similarly, and of relevance for the perception of sign language, studies of visual change detection suggest that the ability to maintain visual information beyond 60–70 ms is very limited (Pashler 1988). Thus, sensory memory for language input is quickly overwritten, or interfered with, by new incoming information, unless the perceiver in some way processes what is heard or seen.

The problem of the rapid loss of the speech or sign signal is further exacerbated by the sheer speed of the incoming linguistic input. At a normal speech rate, speakers produce about 10–15 phonemes per second, corresponding to roughly 5–6 syllables every second or 150 words per minute (Studdert-Kennedy 1986). However, the resolution of the human auditory system for discrete auditory events is only about 10 sounds per second, beyond which the sounds fuse into a continuous buzz (Miller & Taylor 1948). Consequently, even at normal rates of speech, the language system needs to work beyond the limits of auditory temporal resolution for nonspeech stimuli. Remarkably, listeners can learn to process speech in their native language at up to twice the normal rate without much decrement in comprehension (Orr et al. 1965).

Making matters even worse, our memory for sequences of auditory input is also very limited. For example, it has been known for more than four decades that naïve listeners are unable to correctly recall the temporal order of just four distinct sounds – for example, hisses, buzzes, and tones – even when they are perfectly able to recognize and label each individual sound in isolation (Warren et al. 1969). Our ability to recall well-known auditory stimuli is not substantially better, ranging from 7 ± 2 (Miller 1956) to 4±1 (Cowan 2000). A similar limitation applies to visual memory for sign language (Wilson & Emmorey 2006). The poor memory for auditory and visual information, combined with the fast and fleeting nature of linguistic input, imposes a fundamental constraint on the language system: the Now-or-Never bottleneck. If the input is not processed immediately, new information will quickly overwrite it.

Chunk-and-Pass language processing

How, then, is language comprehension possible? Why doesn’t interference between successive sounds (or signs) obliterate linguistic input before it can be understood? The answer, we suggest, is that our language system rapidly recodes this input into chunks, which are immediately passed to a higher level of linguistic representation. The chunks at this higher level are then themselves subject to the same Chunk-and-Pass procedure, resulting in progressively larger chunks of increasing linguistic abstraction.

Crucially, given that the chunks recode increasingly larger stretches of input from lower levels of representation, the chunking process enables input to be maintained over ever-larger temporal windows. It is this repeated chunking of lower level information that makes it possible for the language system to deal with the continuous deluge of input that, if not recoded, is rapidly lost. This chunking process is also what allows us to perceive speech at a much faster rate than nonspeech sounds (Warren et al. 1969): We have learned to chunk the speech stream. Indeed, we can easily understand (and sometimes even repeat back) sentences consisting of many tens of phonemes, despite our severe memory limitations for sequences of nonspeech sounds.

What we are proposing is that during comprehension, the language system – similar to SF – must keep on chunking the incoming information into increasingly abstract levels of representation to avoid being overwhelmed by the input. That is, the language system engages in eager processing when creating chunks. Chunks must be built right away, or memory for the input will be obliterated by interference from subsequent material. If a phoneme or syllable is recognized, then it is recoded as a chunk and passed to a higher level of linguistic abstraction. And once recoded, the information is no longer subject to interference from further auditory input.

In production, the process is reversed: Discourse-level chunks are recursively broken down into subchunks of decreasing linguistic abstraction until the system arrives at chunks with sufficient information to drive the articulators (either the vocal apparatus or the hands). As in comprehension, memory is limited within a given level of representation, resulting in potential interference between the items to be produced (e.g., Dell et al. 1997).

Thus, higher-level chunks tend to be passed down immediately to the level below as soon as they are “ready,” leading to a bias toward producing easy-to-retrieve utterance components before harder-to-retrieve ones (e.g., Bock 1982; MacDonald 2013). For example, if there is a competition between two possible words to describe an object, the word that is retrieved more fluently will immediately be passed on to lower-level articulatory processes. To further facilitate production, speakers often reuse chunks from the ongoing conversation, and those will be particularly rapidly available from memory.

Viewing language acquisition as continuous with other types of skill learning is very different from the standard formulation of the problem of language acquisition in linguistics. There, the child is viewed as a linguistic theorist who has the goal of inferring an abstract grammar from a corpus of example sentences (e.g., Chomsky 1957; 1965) and only secondarily learning the skill of generating and understanding language. But perhaps the child is not a minilinguist. Instead, we suggest that language acquisition is nothing more than learning to process: to turn meanings into streams of sound or sign (when generating language), and to turn streams of sound or sign back into meanings (when understanding language).

If linguistic input is available only fleetingly, then any learning must occur while that information is present; that is, learning must occur in real time, as the Chunk-and- Pass process takes place. Accordingly, any modifications to the learner’s cognitive system in light of processing must, according to the Now-or-Never bottleneck, occur at the time of processing. The learner must learn to chunk the input appropriately – to learn to recode the input at successively more abstract linguistic levels; and to do this requires, of course, learning the structure of the language being spoken.

But how is this structure learned? We suggest that, in language acquisition, as in other areas of perceptual-motor learning, people learn by processing, and that past processing leaves traces that can facilitate future processing. What, then, is retained, so that language processing gradually improves? We can consider various possibilities...

If aspects of language survive only when they are easy to produce and understand, then moment-by-moment processing will shape not only the structure of language but also the learning problem...

...grammatical structure is fundamentally the history of language processing operations within the individual speaker/hearer.

http://www.honeylab.org/wp-content/uploads/christiansen_chater_BBS_2016.pdf

[Tomás]

EuroRomCom was an interesting project to teach speakers of a Romance language to read in the others. I think they had another program for the Germanic languages. And then there was another similar project at a different university whose name I can't recall. Anyway, the EuroRomCom book is interesting. There is another, older book entitled "The Loom of Languages" that has a similar perspective.

[reineke]

There is no need to limit our interest to closely related languages. While the title is "language by exposure" I realize that the OP mentioned related languages in his post. Obviously, exposure to meaningful input and daily processing of a stream of language that is not too far removed from one's own linguistic and cultural background increases the likelihood of success. What I don't like about EUROM5 and similar projects is this huge overreliance on the written language right from the start and the concurrent processing of too many new languages. Of course, they're able to even attempt to do this in the first place because they are relying on written resources.

Effects of statistical learning on the acquisition of grammatical categories through Qur’anic memorization: A natural experiment

"Empirical evidence for statistical learning comes from articial language tasks, but it is unclear how these effects scale up outside of the lab. The current study turns to a real-world test case of statistical learning where native English speakers encounter the syntactic regularities of Arabic through memorization of the Qur’an. This unique input provides extended exposure to the complexity of a natural language, with minimal semantic cues. Memorizers were asked to distinguish unfamiliar nouns and verbs based on their cooccurrence with familiar pronouns in an Arabic language sample. Their performance was compared to that of classroom learners who had explicit knowledge of pronoun meanings and grammatical functions. Grammatical judgments were more accurate in memorizers compared to non-memorizers. No effects of classroom experience were found.

These results demonstrate that real-world exposure to the statistical properties of a natural language facilitates the acquisition of grammatical categories."

"One surprising feature of our findings was the limited benefit of classroom experience in the current task. While classroom learners often provided accurate translations of pronouns and identication of their grammatical categories, this knowledge did not increase sensitivity to co-occurrence patterns in the language sample. In fact, we found that more weekly exposure to Arabic led to less accurate judgments in Replacement trials, suggesting that explicit knowledge of the grammatical functions of closed-class items may have interfered with statistical learning. These results have implications for educational curricula in second language acquisition. Teaching methods traditionally rely on metalinguistic knowledge of the rst language to acquire grammatical features in the second (e.g., introducing ‘‘tum” as the plural form of the second person pronoun). Yet, our findings suggest that this top-down approach may negatively impact learners’ sensitivity to the bottom-up statistics of a language."

https://www.google.com/url?sa=t&source= ... IRSzQ_vaRg

[Finny]

Never lived in another country, but that's basically how I learned Portuguese. After learning the grammar, I jumped right into native materials. I tried Assimil, but it was way too easy for me. Even before I started getting into Portuguese, just with my other Romance knowledge I could read basic newspaper articles and some light novels with relative ease.

What the goose said. The only ever actual "studying" I've ever done of Portuguese was parallel reading a novel in it and English. In retrospect, I think I primarily read it in Portuguese and just used the English copy for reference. Since then, I've been able to read pretty much any normal text in the language without much trouble, as well as understand a decent amount of music in the language. I'm no language whiz; it's just very similar to Spanish. Ditto with Catalan.

[Brian]

I've never really bothered trying to learn any Dutch, just kept my eyes and ears open a lot

By doing so, you are actively engaging with the language and learning things in a real-life context. Many people would simply switch off if they knew they could get by in another language.