Another Linguist's Take on Blanc's Proposition of Gestalt Language Processing and Natural Language Acquisition: Part 2

---

What We're Here For

Image Description: a purple dart sticking out of the center of a bulls-eye.

Before we begin, I want to clarify for us all -- especially myself -- my intended purpose for these blog posts. While the topic of Gestalt Language Processing (GLP) is broad and complicated, my aim with these posts is to respond to very specific claims from specific sources, which I will try to make clear in each post.

This will mean leaving out a lot of important related discussion, and I believe it's important that the discussions that are left out here aren't neglected forever, but I think there's a place for discussing details one at a time as well. I'll try my best to do exactly that and not get sidetracked, though I don't anticipate it being easy -- I love talking about language, and there are way too many fascinating and meaningful tangents to pursue here.

---

A Quick Reflection On Parts 1 and 1.5

After my initial post on this blog and a related Instagram post, and the discussions they've led to in their respective comments sections, I've been able to draw this conclusion:

Different people can interpret the same evidence differently.

Image Description: a blue and orange checkerboard background with two magnifying glasses on it, one magnifying glass shows a yellow and purple pattern inside of it, the other magnifying glass shows a red and green pattern inside of it

This seems obvious in hindsight. But it looks like I have an answer to my initial question about why current GLP critics (that I'm aware of) haven't acknowledged the evidence that I'm currently discussing. If they're aware of it, they don't think it actually supports GLP/NLA. And plenty of other people think it does support it.

So my goal here isn't to change anyone's mind about how to interpret the evidence. It's simply to increase awareness of evidence that I see as relevant to the discussion.

---

Gestalts As A Possible Building Block For Language Acquisition

We're going to continue responding to the same claim made by Katharine Beals in the same journal article as in the first post:

[T]he purported linguistic units in GLP -- gestalts, defined as language that is echoed as analyzed [sic] stretches of speech sounds -- are not plausible either as building blocks in language acquisition, or as full-fledged linguistic phenomenon [...].

I'm responding here with what I believe is another piece of evidence that disproves at least part of the claim made above.

The following quote is, like the previous post's quotes, from Tomasello's 2003 text Constructing a Language: A usage-based theory of language acquisition.

[I]n languages that are less isolating than English (for example, polysynthetic languages such as many Eskimo [sic] languages) the whole-to-parts pattern of acquisition, requiring the segmenting of communicative intentions to extract meaningful elements, is the normal case. For example, in Inuktitut many early utterances are word-sentences such as Taartaulirtunga, meaning “Something is in my way,” or Tuqutaulangasivungaa! Meaning “I’m going to get killed!” (Allen, 1996) – which obviously require some segmenting to extract the productive elements involved. In any event, the general principle is that young children come equipped to move in either direction – part to whole or whole to parts – in learning to linguistically partition experiential scenes and indicate their constituents with different linguistic elements in multi-unit expressions and constructions. All children probably use both processes to some extent in different aspects of language acquisition.

There's a lot of terminology there that non-linguists might not have heard of, so let's break it down a bit:

Isolating languages vs. polysynthetic languages

Image Description: a multi-colored lego tower built with a variety of sizes of lego bricks

The Tomasello quote mentions two types of languages -- isolating languages and polysynthetic languages -- and a slight difference in how children tend to acquire each one. One of the primary differences between these types of languages has to do with morphemes, so let's explain that term first, starting with some examples from English:

Sentence:	Words (9):	Morphemes (10):
The quick brown fox jumps over the lazy dog.	the, quick, brown, fox, jumps, over, the, lazy, dog	the, quick, brown, fox, jump/, -s, over, the, lazy, dog

A morpheme is the smallest part of a word that carries some kind of meaning.

In "jumps", the base word "jump" is a morpheme that carries the meaning of the type of action being described, and the suffix "-s" is a morpheme that tells us about who's doing the action and when it's happening. Putting them together creates one word containing two morphemes.

In the above sentence, most of the words contain only one morpheme, with the exception of the verb "jumps". The word-to-morpheme ratio is almost 1:1.

Let's look at another English example:

Sentence:	Words (14):	Morphemes (21):
The mischievous, stealthy fox jumped into the underground tunnel and evaded the lethargic dog.	The, mischievous, stealthy, fox, jumped, into, the, underground, tunnel, and, evaded, the, lethargic, dog	The, mischiev/, -ous, stealth/, -y, fox, jump/, -ed, in/, -to, the, under/, -ground, tunnel, and, evade/, -ed, the, letharg/, -ic, dog

The word-to-morpheme ratio in this sentence is much farther from 1:1. There are quite a few words in this sentence that contain multiple morphemes, including base words, compound word parts, and suffixes.

In isolating languages, the word-to-morpheme ratio tends to be pretty close to 1:1, like the first example sentence. English is a relatively isolating language compared to many of the world's other languages. For the most part, English words contain a relatively small number of morphemes, and plenty of words only contain one single morpheme.

Polysynthetic languages, on the other hand, are the opposite. Most words contain multiple morphemes, and often entire sentences -- or complex messages that would require a multi-word sentence in a more isolating language like English -- can be conveyed with only one word that's made up of multiple morphemes:

[I]n Inuktitut many early utterances are word-sentences such as Taartaulirtunga, meaning “Something is in my way,” or Tuqutaulangasivungaa! Meaning “I’m going to get killed!” (Allen, 1996) – which obviously require some segmenting to extract the productive elements involved [...]

This is all, by necessity, a simplified explanation of an admittedly complex linguistic concept.

But it seems clear from this quote from Tomasello's text that young children acquiring this polysynthetic language produce these long strings of speech sounds as some of their first "words," and they serve as a functional building block for their further language acquisition. They start by using (at least some of) these multi-morphemic strings of speech sounds and only later break them down to use the smaller internal parts ("the productive elements involved") -- a whole-to-parts acquisition pattern.

A "whole-to-parts pattern of acquisition" sounds very similar to the way Gestalt Language Processors are described as beginning their language acquisition journey, in my opinion. The whole-to-parts pattern could be involved in the proposed NLA Stages 1, 2, and part of 3; with part-to-whole acquisition implicated in Stage 3 and continuing into the higher stages. Further analysis would be needed to pin down the exact nature of the relationships between the Stages and directional acquisition patterns, but the potential parallels seem clear to me with even this cursory look.

Tomasello states that this whole-to-parts acquisition pattern is seen more commonly in children acquiring some languages (polysynthetic) as opposed to others (isolating), based on specific characteristics of said languages (i.e. word-to-morpheme ratios). This is an example of a learner-external factor influencing the preponderance of whole-to-part vs. part-to-whole acquisition patterns.

This question then follows: Is it possible for acquisition pattern use to vary based on learner-internal factors, like the neurological differences associated with autism? To the best of my understanding, Blanc's GLP/NLA model proposes that it is indeed possible.

This all seems to me like a pretty clear rebuttal to Beals' claim that gestalts aren't feasible building block of language acquisition. But, as I've learned, it's likely that others will interpret this evidence differently. But at least it's out there now, so those that weren't previously aware can draw their own conclusions.

---

Part 1